A cautionary note

[traynor]

I know that everyone on this board is familiar with how "automatic updates" work, but there may be a few lurkers (possibly from the Lake Minnetonka camp) who do not.

In the 9th race at Louisiana Downs yesterday, there was a $107.20 winner. In just about every software app that has a facility for "regression analysis" linked to mutuel payoffs or "overall profit," that winner is going to crop up. It is not really an issue if you understand what is happening, but it could be seriously misleading if you see something like, "140% ROI" for Factor Whatever. Even if you do an "all-races" regression for the past month or so, that one winner will skew the results.

Of course, this is not an issue for those of you who may have 100,000 races or whatever in your databases. For those using software apps with only a month or two of races, or those from the Sartin school of "what is winning now," the result of that one race can create the impression of profitability where it doesn't really exist.
Good Luck

[Tom]

I always throw out abberent prices anyway.

[traynor]

I always throw out abberent prices anyway.

I assumed you did, as does everyone who has any good sense. The problem is the newbies, or uncertain, or less confident, or whatever that makes people trust numbers on a screen more than they trust themselves. Without disparaging anyone who uses the apps, there are several that automatically update according to raw ROI. In that case, a mutuel that high is going to skew the algorithms as soon as they hit the "Update" button. If they only play a couple of tracks (and only have limited results for those tracks), the effect is even stronger.
Good Luck

[Murph]

What of the 8th race winner @ 12-1? Is this one considered an outlier? I would like to have a better understanding of outlier criteria you use to decide when a winner is "out" of acceptable range. For instance do you consider short prices for outliers as well?

Thanks traynor, anything you could add would be helpful to me.

Murph

[Blackgold]

Murph, in the 8th at LAD the 7 figured. I had the exacta and tri there.

In the 7's last race, he was reversing against the 1 (that had numers) that finished 2nd. And the 5 also came from that previous race and finished 4th to give someone a $10K super.

What was easy was throwing out the 1st and 2nd choices, both bet down because of perceived class drops.

The fav, the 10, opened with $900 in the win pool, usually a cause for concern at Larceny Downs. But the 10 could not meet the fulcrum pace of 46 for the half, yet the public was betting him like he was Barbaro.

And they crushed the 9 to 7/2 because of a perceived class drop. Hell, the owners and trainers had simply been running their stock over their heads. Maybe a drop to Penn National will do it.

Anyway, gool luck all.

[traynor]

It depends whether you are cashing the ticket or entering a mutuel price in a data model. If the latter, a fairly standard practice (in statistics) is to run what most stat apps call a "box and whisker plot."

Sort and rank. Toss the low quater and the top quarter. Average the "middle half" remaining. That is the "mean of the interquartile range" (or, more simply, the average of the middle half). Multiply that figure by 1.5. Anything over that is an outlier. It should be cut and dried, programmed in, and automatic. At least for data modeling that you are going to use for betting purposes.

It is far better to be pleasantly surprised when your 110% ROI model consistently cranks out 120-130% ROI than to discover it only generates 85-90% ROI in the real world because of outliers. Again, this is for data modeling--for constructing the algorithms used to predict future events. Both the process and the 1.5 multiplier are pretty standard statistics, rather than my personal opinion. The only opinion I have about it is that it works really, really well. If you correct for outliers and still show a decent profit, you have a real winner.
Good Luck

[GameTheory]

It depends whether you are cashing the ticket or entering a mutuel price in a data model. If the latter, a fairly standard practice (in statistics) is to run what most stat apps call a "box and whisker plot."

Sort and rank. Toss the low quater and the top quarter. Average the "middle half" remaining. That is the "mean of the interquartile range" (or, more simply, the average of the middle half). Multiply that figure by 1.5. Anything over that is an outlier. It should be cut and dried, programmed in, and automatic. At least for data modeling that you are going to use for betting purposes.The way I've always seen it done is to multiply 1.5 times the RANGE, not the mean value contained in the range. In other words, if the Q1 value was 10, and the Q3 value was 15, then your interquartile range is Q3-Q1 = 5, giving us a "step" of 1.5 * 5 = 7.5. And therefore anything below 2.5 (Q1 - 7.5) or over 22.5 (15 + 7.5) would be an outlier. If you used the mean value in the range, that value in this example would likely be around 12.5, giving us a much larger "step" size of 12.5 * 1.5 = 18.75. Of course in other cases it could lead to a much smaller step size.

But using the interquartile range rather than the interquartile mean is the standard way, no? The interquartile mean is also useful, though, and is usually used as a better estimate of the true center (or central tendency) of a range of data (much like the median).

Here is a more detailed description of the process of indentifying outliers:

http://www.purplemath.com/modules/boxwhisk.htm

[GameTheory]

Actually, I see you aren't using "steps" added to or subtracted from Q1 & Q3, but merely anything higher than 1.5 times the interquartile mean would be an outlier, which in my example would actually give you a tighter range. (Assuming you used .5 times the mean on the lower end to give you an symmetrical range.) Nevertheless, I believe that method to be non-standard. Were you taught it that way?

[traynor]

The mean is simply the mid-point of the range. In every business stats class I have taken (a considerable number, both undergraduate and graduate), the mean is used, rather than the range. That may be because business classes are more focused on the real world, rather than theory, and using the mean is a more concise way of expressing the same concept.

[traynor]

While the purplemath site is interesting, it is not particularly useful for our purposes. It suffers the same logical flaws as many "research" studies.

By establishing a median, rather than a mean, you are calculating a probable range. That is not particularly useful, either in handicapping or business. What we want is a realistic "average mutuel," not a "possible mutuel." Simply stated, 1.5 times the center half of mutuel prices will give you a much more realistic (as in the real world, as opposed to theoretical) value for truncating outliers. That is why we use it.

[GameTheory]

While the purplemath site is interesting, it is not particularly useful for our purposes. It suffers the same logical flaws as many "research" studies..It outlines the same exact method shown on hundreds of other sites and in all statistics textbooks I've ever seen. So your business professors with their "real-world" knowledge would apparently defy the textbooks and instruct use of the IQM*1.5 as the upper threshold (and IQM*.5 as the lower, presumably?) instead of the usual Q3+IQR*1.5. Interesting. But you said it was "pretty standard statistics" -- maybe you can name some statistics texts that advocate your method as being preferable in the real-world.

Of course, the interquartile method with a rigid value of 1.5 isn't the only way to go. You can also just exclude those values above or below any arbitrary z-score (plus or minus standard deviation units) of your choosing, or use Grubb's iterative method where you remove one outlier at a time. (Just Google "outliers Grubb" for more details.)

It should be pointed out that all of these methods assume you've got a normal distribution of payoffs, something not neccessarily true.

[traynor]

First, I have no interest whatsoever in long-winded, academic discourse about how many angels can dance on the head of a pin. That is waste of your energy and my time. I am a bit skeptical of your "expertise" if you cannot distinguish between descriptive statistics and inferential statistics. End of discussion.

[GameTheory]

First, I have no interest whatsoever in long-winded, academic discourse about how many angels can dance on the head of a pin. That is waste of your energy and my time. I am a bit skeptical of your "expertise" if you cannot distinguish between descriptive statistics and inferential statistics. End of discussion.Did I claim to be an expert? Did I even imply that your method wasn't useful?

I was simply trying to find out where your "standard" method came from, since it appears nowhere that I can find in any statistics text or anywhere on the vast internet. You are the expert, not me, to know about such an exclusively held secret that is only dispensed at advanced business stats classes. (I was a film major, for God's sake. I only know about Eisenstein and f-stops.) I was just hoping to reconcile the fact that it is "standard statistics" with the reality that it is so hard to find a description of it outside of your post. (Whereas a simple Google query turns up thousands of pages -- many from actual respected universities -- showing the "theoretical" IQR*1.5 method.)

I know it must be out there somewhere, because you would never feed us any BS, and would be the first to admit it if you had made any error, of that I am sure...

[PaceAdvantage]

Hey, didn't you read traynor's last post? He said "end of discussion", damn it!! :lol:

But seriously folks, he is a very serious player....and a stubborn one as well, apparently.

Traynor, I know this may be hard for you to do, but you owe GameTheory a little more respect than you showed in that last reply.

[traynor]

Hey, didn't you read traynor's last post? He said "end of discussion", damn it!! :lol:

But seriously folks, he is a very serious player....and a stubborn one as well, apparently.

Traynor, I know this may be hard for you to do, but you owe GameTheory a little more respect than you showed in that last reply.

Not hard at all. I apologize to GameTheory. Sorry, GameTheory.

It would be interesting to hear from the REAL experts--the software developers who are on this board--about how they correct for aberrant data, how they prevent unusual mutuels from corrupting their ROI figures, and what process they use to correct for outliers.

[DJofSD]

In a previous life, I learned that apparently corrupted data is left as is unless you can determine the aberrant data points are due to equipment failure and the such.

I had this point demonstrated one day in one of those impromptu discussions in one of those once in a lifetime meetings with a leader in my field of study. As an undergraduate doing research with a couple of graduate students, we collared a presenter during a conference. We had some questions about some data points and their possible interpretation. On a graph, he showed us that by just covering one point with his thumb how it might influence our thoughts.

Later, on a different project, the same point was made again. It appeared there was something wrong with the data. But, we left things alone then as more data accumulated, we realized our a priori assumptions were wrong. We actually had observational data that showed we had made some new discoveries worthy of publication.

[traynor]

In a previous life, I learned that apparently corrupted data is left as is unless you can determine the aberrant data points are due to equipment failure and the such.

I guess it depends on whether or not you are betting on calculations based on that corrupted data. I had hoped for a number of software developers to explain how they correct for outliers, and whether they use the "mean of the interquartile range" or the "median of the interquartile range" to be sure their applications are not presenting inflated and misleading ROI figures.

[traynor]

And not a word from the developers as to whether they prefer means or medians in correcting their datasets for anomalous outliers.

[traynor]

The question does not seem to be, "Is the mean or the median more statistically correct, or more in line with contemporary statistical theory?" but rather, "Why would I want to put a calculation in my software application that would eliminate deceptive, misleading anomalies in one fell swoop?" That is a rhetorical question--I already know the answer. So should every other user of a software application with "regression analysis" or "automatic algorithm update" features.

[Red Knave]

...--I already know the answer. So should every other user of a software application with "regression analysis" or "automatic algorithm update" features.
__________________
http://www.tekwrytrs.com/Home/FreeSample.aspx
Free Samples of Positive ROI Graded Ratings for Belmont, Arlington, Hollywood Park, and Woodbine posted daily.
Why does this sound so much like an ad?

And go steal someone elses avatar! :D :lol: :faint:

[traynor]

Why does this sound so much like what Ruth Ellen Boetcher-Joeres so elegantly referred to as "obfuscatory prose" and others less literate call "evading the issue by shoveling rapidly from the readily available pile of" whatever?

[Jeff P]

And not a word from the developers as to whether they prefer means or medians in correcting their datasets for anomalous outliers....

...It would be interesting to hear from the REAL experts--the software developers who are on this board--about how they correct for aberrant data, how they prevent unusual mutuels from corrupting their ROI figures, and what process they use to correct for outliers.
The simple truth is this: I don't worry about outliers at all. Not one single bit.

I simply take what I see in data samples and build algorithms and models around that. Of course I'm looking at some really big chunks of data - every race day for the same set of tracks for several years on end.

Funny thing though. After a while the only thing that really seems to change from one year to the next are the names of the horses, riders, trainers, and owners themselves. The rest of the factor mix that I use (and I DO use it) produces almost identical long term results in terms of win percent and roi whenever the big picture is looked at.

-jp

.

[keenang]

Wow!
This guy Traynor seems to have a very short fuse. Like Rodney said "Why can't we all get along.

Geno

[traynor]

Wow!
This guy Traynor seems to have a very short fuse. Like Rodney said "Why can't we all get along. Geno

I thought we were!

[traynor]

Jeff P wrote: <Of course I'm looking at some really big chunks of data - every race day for the same set of tracks for several years on end.>

That pretty much sums up my point. Small models do not accurately represent reality, and "regression analysis" of models containing less than (some big number) races should be viewed with skepticism. Especially if an application is programmed to "automatically update" algorithms based on mutuel payoffs in that short model, or the user is wagering based on the "ROI" of some factor (or combination of factors) in that short model.

[Murph]

The statistical anaylists are helping me along a fairly steep learning curve in this area. I realize that when I am using same meet data my wagering profiles are not consistent. It is only when I use 600 or more events that I can prove my profiles more reliable. Alas, this seems to lead to meager profits that often do not overcome the take.

I need to learn how to adjust my short term data sets for greater consistency. I believe (from my research) the answer to the dilema may lie in regression anaylisis. So as I struggle to reshape my data models please continue the debate. These concise postings have pointed me to information and ideas I have not considered in the past.

Mean or median gentlemen? Thanks for sharing your specialized knowledge.

Murph

[Red Knave]

Why does this sound so much like what Ruth Ellen Boetcher-Joeres so elegantly referred to as "obfuscatory prose" and others less literate call "evading the issue by shoveling rapidly from the readily available pile of" whatever?
So, you're saying it's NOT and ad? Okay :rolleyes:
And, not being interested in sociology or women's issues, I have no idea whether or not Ms. Boetcher-Joeres ever said anything like that.
My apparently obtuse point was that, since no one responded to your question, why would you answer yourself twice except to keep the thread bumped up ?
Admittedly, I didn't add much (as PA will attest, I rarely do ;) ) but my gut and my experience tell me that anyone that uses the term 'obfuscatory' knows what it means by practice.

[cj]

The simple truth is this: I don't worry about outliers at all. Not one single bit.

Bravo! The simple truth for me is the "outliers" provide a large chunk of the profits. Why would I want to classify them as flukes?

[traynor]

So, you're saying it's NOT and ad? Okay :rolleyes:
And, not being interested in sociology or women's issues, I have no idea whether or not Ms. Boetcher-Joeres ever said anything like that.
My apparently obtuse point was that, since no one responded to your question, why would you answer yourself twice except to keep the thread bumped up ?
Admittedly, I didn't add much (as PA will attest, I rarely do ;) ) but my gut and my experience tell me that anyone that uses the term 'obfuscatory' knows what it means by practice.

I assume everyone knows what obfuscatory means, but how often do you get to quote Boetcher-Joeres? Her reference was to something a little more complex than women's issues, although it involved them. Specifically, it was a reference to the tendency of discourse groups to use in-group jargon in an exclusionary fashion to denigrate (possibly) dissenting or conflicting opinion. As in the phrase, "Don't confuse me with the facts, my mind is already made up."

Consider the rhetorical strategy of "not being interested in sociology or women's issues." It implies that such a view is somehow superior, or more acceptable, than other alternatives. I can fully sympathize with that, because, as I stated in an earlier posting on this thread, I have no interest in pointless, long-winded debates about unrelated topics. Or would you like to continue a discussion of Ruth Ellen's undeniable contribution to the fields of psychology and communication?

[Overlay]

Funny thing though. After a while the only thing that really seems to change from one year to the next are the names of the horses, riders, trainers, and owners themselves. The rest of the factor mix that I use (and I DO use it) produces almost identical long term results in terms of win percent and roi whenever the big picture is looked at.

I agree with you, and the operative word in your post is "mix", which is what allows the consistent results over time (particularly with respect to roi) that a more narrowly-focused or less diverse approach can't maintain.

[traynor]

Bravo! The simple truth for me is the "outliers" provide a large chunk of the profits. Why would I want to classify them as flukes?

Simple. To make smaller data samples more in line with reality. Not everyone uses a 100,000 (or whatever number) race database. When you are using samples of a few hundred races (a LOT by Sartinista standards) or a few thousand races (still way too few if you layer by distance, surface, conditions, and so on), the inclusion of outliers misrepresents the potential replication of short-term results over a longer time span. In short, the return for betting on such models is likely to be substantially less than the ROI indicated by regression analysis of raw data would seem to indicate. (Gee, that sounds obfuscatory.)

[Jeff P]

I think it's important to understand what an outlier is - or isn't. For me the definition of an outlier extends well beyond an unusually high win mutuel. A true outlier in my eyes is a horse that wins or runs big despite the fact that none of the factors I use come anywhere close to suggesting that horse.

For example, take the three of us: Traynor, CJ, and myself. We each use different methodologies to arrive at our plays. A highly rated horse by Traynor or CJ might be viewed by me as garbage taking up space on the track. And sometimes it will be the other way around. Despite the fact that each of us uses very different methods to arrive at selections, I'd make an educated guess that the bulk our collective wagering profits is generated whenever the public chooses to ignore a horse that one of us has played in a race where that horse outruns its odds.

So what might be an outlier in Traynor's eyes could very well be a frequently repeatable event for CJ or myself... and of course sometimes it will be the other way around.


-jp

.

[Red Knave]

Consider the rhetorical strategy of "not being interested in sociology or women's issues." It implies that such a view is somehow superior, or more acceptable, than other alternatives. No it doesn't. It implies nothing. It's a statement not an implication. And neither is it rhetoric. You saying it is doesn't make it so.I can fully sympathize with that, because, as I stated in an earlier posting on this thread, I have no interest in pointless, long-winded debates about unrelated topics.And yet here you are doing just that.Or would you like to continue a discussion of Ruth Ellen's undeniable contribution to the fields of psychology and communication?Undeniable? Hardly. She's a harridan, a harpy and a hack, IMHO.

[traynor]

Undeniable? Hardly. She's a harridan, a harpy and a hack, IMHO.

I used to think the same thing. It took some serious exploration in textual deconstruction to get beyond the surface content to understand what she is actually saying.

Think of it in Korzybski's frame of reference; as soon as people label something, they tend to respond only to the label, not the thing itself. Your perception of Ruth Ellen may be based on your categorization, rather than the content of her writing.

[traynor]

No it doesn't. It implies nothing. It's a statement not an implication. And neither is it rhetoric. You saying it is doesn't make it so.

On the contrary, your denial does not diminish the rhetorical implications in the least. BTW, a definition may be in order; "rhetoric" is used in contemporary media to imply "lack of valuable content," as in the phrase "nothing but empty rhetoric." That would be insulting, and was not my intent. My use of the word "rhetoric" is that used by academics and scholars, rather than by Anne Coulter. Specifically, rhetoric was defined by Aristotle as "the use of all available means of persuasion." Defining your statement as a rhetorical strategy was in no way intended to be demeaning or insulting; it was to define it as a persuasive statement.

[traynor]

I think it's important to understand what an outlier is - or isn't. For me the definition of an outlier extends well beyond an unusually high win mutuel. A true outlier in my eyes is a horse that wins or runs big despite the fact that none of the factors I use come anywhere close to suggesting that horse.

An interesting explanation, but it doesn't address the key issue of why anyone should care about outliers in the first place. Unusual data points in a small sample may diminish the predictive ability of that model; they may not represent the distribution of data points in a different sample. In short, aberrant data in a sample is not "normally distributed" through other samples, or through the general population from which the samples are drawn.

Outliers are easy to locate. Run a bootstrap algorithm on your dataset of mutuels (random sampling with replacement). If a graph of the subsets shows spikes, it is from outliers. If the samples are fairly symmetrical, the mutuel prices are fairly evenly distributed. SPSS has bootstrap built in, and there are free add-ins for Excel that work every bit as well.

Instead of debating whether the "median" or the "mean" of the interquartile range is more appropriate, or Boetcher-Joeres' contribution to group dynamic theory, just cut to the chase, run a bootstrap, and look at your data. If most for the peaks are at (some arbitrary height) and a few are spiked substantially above that height, you have outliers. Crop them, re-run the bootstrap until your dataset is fairly symmetrical, and you may have something worth betting on.

The value of a handicapping software application is in predicting the future, not simply reiterating the past. For any model less than 5000-10,000 races, cropping the outliers makes a more predictive (and more representative of reality) data model for wagering.

[GameTheory]

Instead of debating whether the "median" or the "mean" of the interquartile range is more appropriateYou are the only one debating that since you pulled it out of the blue earlier in the thread. (Just as you seem to have made up the interquartile mean method despite pretending that it is actually taught that way somewhere.) The standard method is to use the range*1.5 + Q3 for the upper threshold, and Q1 - range*1.5 for the lower. (The range value being Q3-Q1.) The interquartile median never comes into play at all.

Of course, I'm with Jeff and CJ. I never eliminate any data points, although I do smooth them in some circumstances with bootstrap methods.

Just which software apps do this dastardly "automatic updating" that are you so concerned about? (Concern which does come across as an ad for your own service -- do you see Jeff or CJ starting threads warning people to stay away from other people's software?) You seem to have a grudge against something from 20 years ago...

[Jeff P]

Instead of debating whether the "median" or the "mean" of the interquartile range is more appropriate, or Boetcher-Joeres' contribution to group dynamic theory, just cut to the chase, run a bootstrap, and look at your data. If most for the peaks are at (some arbitrary height) and a few are spiked substantially above that height, you have outliers. Crop them, re-run the bootstrap until your dataset is fairly symmetrical, and you may have something worth betting on. I think we'll just have to agree to disagree.

I have a computer model of a spot play. I developed this particular model in 2003 and have been using it ever since. The research... the working out of the rules... the coding and testing of the code to make sure I am getting the right plays... all of that was completed years ago... I make no handicapping decisions whatsoever on a daily basis when it comes to this model's plays. I simply load race card files, get scratches, and run a report that has the plays on it.

This is a summary of all the plays produced by this model at the tracks that I followed during calendar year 2005:

Data Summary Win Place Show
Mutuel Totals 3706.20 2829.70 2531.40
Bet -2500.00 -2500.00 -2500.00
Gain 1206.20 329.70 31.40

Wins 244 423 579
Plays 1250 1250 1250
PCT .1952 .3384 .4632

ROI 1.4825 1.1319 1.0126
Avg Mut 15.19 6.69 4.37


Taking a closer look, here are the model's plays broken out by post time odds:

By: Odds Range

>=Min <Max Gain Bet Roi Wins Plays Pct Impact
-999.00 0.00 0.00 0.00 0.0000 0 0 .0000 0.0000
0.00 5.00 180.70 916.00 1.1973 131 458 .2860 1.4653
5.00 10.00 251.60 892.00 1.2821 70 446 .1570 0.8041
10.00 15.00 398.30 366.00 2.0883 29 183 .1585 0.8118
15.00 20.00 96.50 156.00 1.6186 7 78 .0897 0.4598
20.00 25.00 -29.00 76.00 0.6184 1 38 .0263 0.1348
25.00 30.00 122.00 40.00 4.0500 3 20 .1500 0.7684
30.00 35.00 -24.00 24.00 0.0000 0 12 .0000 0.0000
35.00 40.00 139.60 10.00 14.9600 2 5 .4000 2.0492
40.00 45.00 80.50 10.00 9.0500 1 5 .2000 1.0246
45.00 50.00 -2.00 2.00 0.0000 0 1 .0000 0.0000
50.00 55.00 -2.00 2.00 0.0000 0 1 .0000 0.0000
55.00 60.00 -2.00 2.00 0.0000 0 1 .0000 0.0000
60.00 65.00 -4.00 4.00 0.0000 0 2 .0000 0.0000
65.00 70.00 0.00 0.00 0.0000 0 0 .0000 0.0000
70.00 75.00 0.00 0.00 0.0000 0 0 .0000 0.0000
75.00 80.00 0.00 0.00 0.0000 0 0 .0000 0.0000
80.00 85.00 0.00 0.00 0.0000 0 0 .0000 0.0000
85.00 90.00 0.00 0.00 0.0000 0 0 .0000 0.0000
90.00 999999.00 0.00 0.00 0.0000 0 0 .0000 0.0000


You could make the argument that the plays going to post at odds of 20-1 and up are really just outliers.

But here is why I refuse to crop them:

I have a decision to make. Do I want to play this model's plays at 20-1 and up? Or do I want to pass them?

I know from looking at every data sample I have going back to the first year that I started downloading files from Bris that this model's plays are profitable - despite the very low win rate - at 20-1 and up. I also know from first hand experience - and data samples from the first two quarters of 2006 - the model's plays continue to be profitable at 20-1 and up.

For me it boils down to a simple play or pass decision. And my decision is to play them - whether somebody else thinks of them as outliers or not.

-jp

.

[mainardi]

Bravo! The simple truth for me is the "outliers" provide a large chunk of the profits. Why would I want to classify them as flukes?
I agree.... but maybe not for the common reasons that most folks cling to.

I'm one of those people that traynor is calling for, as I write handicapping software that I use AND sell (I have been for 18 years). Near the start of my long sojourn, I learned some very good lessons -- from a guy that made a living off of mathematics and "number crunching" (sadly, though, not at the horses) -- that stick with me to this day. His forte was neural networks, and his novice level interest in horse racing drove him to stuff scads of numbers into his 'net.

After much analysis and consideration, we concluded the following:
1. Patterns focused on tracks -- aka biases -- can't be computed quickly enough to provide a long-term advantage. My favorite example was when we converted a "rail is dead" bias, to the real long-term bias of "throw out jockeys too stupid to avoid getting caught on the rail" :bang: . That's why I am not a bias handicapper.

2. Collecting results data isn't nearly as valuable when you don't bet on the race. Bear with me on this one... I hope to make a point. When your money is on the line, it's not nearly as easy to stick with something that you've only seen work through some algorithm. Betting is a game of confidence, and if you can't see yourself betting on certain patterns that you've never actually had a vested interest in, then all the computations in the world aren't going to help. More to cj's point, if you know that a pattern works that YOU HAVE ACTUALLY BET ON, you're much less likely to dismiss it as an anomaly or outlier. And conversely, if you mostly run your "wagers" on paper, when it comes time to lay out the cash, your gambling instincts will lead you to back off.

I track only the wagers I ACTUALLY make, because that's where I learn what works when I put up my own money. And I don't throw out any wagers... how else am I going to learn from them? I don't do nearly as much wagering as many others out there -- a wife, child, day job and my horse racing business don't leave much free time -- but I'll take my 6% ROI (330 races in 5+ years time) and all the lessons that I've learned in order to get better over any "mass data calculations" that toss out "anomalies" and "outliers" and don't consider the actual dollars wagered.

Two cases that prove my point (at least to me). First, I haven't had a winner at more than 12-1 in two years... but I stick to my guns because even though I'm 1-for-14 -- yes, I'm VERY selective -- I'm still ahead in that odds category. Also, after 3+ years and 150+ races, I had enough data to do an initial analysis of my wagers (and NOT computations of what I would have played)... and what I found was downright frightening. My ROI was a putrid -13.5%. So I analyzed what DID work for me during that time, eliminated my "bad bets", and my ROI jumped to +6% (that's an ROI of +17% since that day).

No brag, just fact... and if I threw out the "anomalies" and "outliers", I no doubt would still be milling about below the water line... or maybe even out of the game entirely!!!

Enough said... whew, I'm worn out... I'm going to bed... :sleeping: :sleeping:

[traynor]

A brief explanation. I have no quarrel with handicapping software developers per se. I have no interest in converting the world to the use of the mean of the interquartile range as a faster, easier, simpler way to crop outliers. I don't even really care about outliers (except in my own data models), other than in one, specific case.

That case is a relatively inexperienced user who uses the regression analysis feature that is standard on many applications to create a betting model. My entire point is that anomalies like the $107 win at LaD will badly skew the results if accepted at face value as a "normal" result. It was not normal.

Because I build a LOT of betting models, I learned early on that small samples are more predictive if outliers are truncated. Rather than building a 5000 race sample, I can use 500-1000 races provided the sample is controlled for outliers, and get pretty much the same distribution. In plainer language, it gives me most of the advantages of a 5000 race model, with a lot less time involved.

The bottom line is that if you control for outliers, you can use a more concise dataset (weeks of races rather than months and years) and make useful approximations and predictions. That is convenient at places like Saratoga, Del Mar, and others. Oddly enough, those small samples that Sartin and Brohamer recommended worked really well--as long as the users had the good sense to realize that mixing $107 mutuels into a small sample of races created seriously defective predictions.

I apologize if my comments were perceived as insulting by any developers on this board. That was not my intent.

[traynor]

mainardi wrote: <More to cj's point, if you know that a pattern works that YOU HAVE ACTUALLY BET ON, you're much less likely to dismiss it as an anomaly or outlier. And conversely, if you mostly run your "wagers" on paper, when it comes time to lay out the cash, your gambling instincts will lead you to back off.>

Betting or not betting on a particular race has no relationship to the definition of an outlier. An outlier is simply a non-standard result, or a result that is not evenly distributed through a larger sample from the same population. It is not a subjective evaluation. Cashing big tickets is good. Betting large amounts of money on models that are based on anomalies and wishful thinking is bad.

[traynor]

You are the only one debating that since you pulled it out of the blue earlier in the thread. (Just as you seem to have made up the interquartile mean method despite pretending that it is actually taught that way somewhere.)

You may find the link below illuminating. Oddly, it wasn't that difficult to find.

http://www.answers.com/topic/interquartile-mean

[GameTheory]

You may find the link below illuminating. Oddly, it wasn't that difficult to find.

http://www.answers.com/topic/interquartile-meanThe calculation of the interquartile mean? Yeah, so what? Did you think I was arguing that the interquartile mean doesn't exist? Doesn't anyone around here actually bother to read what's written?

Now please find me the page where it says the standard upper threshold for determining outliers is the interquartile mean * 1.5 instead of the interquartile range * 1.5 + Q3...

[traynor]

The calculation of the interquartile mean? Yeah, so what? Did you think I was arguing that the interquartile mean doesn't exist? Doesn't anyone around here actually bother to read what's written?

Now please find me the page where it says the standard upper threshold for determining outliers is the interquartile mean * 1.5 instead of the interquartile range * 1.5 + Q3...

You should be able to do that yourself.

[traynor]

The calculation of the interquartile mean? Yeah, so what? Did you think I was arguing that the interquartile mean doesn't exist? Doesn't anyone around here actually bother to read what's written?

Now please find me the page where it says the standard upper threshold for determining outliers is the interquartile mean * 1.5 instead of the interquartile range * 1.5 + Q3...

I would be more interested (as I am sure many others on this board would be) in you finding the page where it says outliers can be ignored in small samples, when those samples are specifically intended for use in "predicting" future events.

[sjk]

I'm with Jeff. All races are pretty much the same. Why would there be an reason to use a small sample?

[traynor]

I'm with Jeff. All races are pretty much the same. Why would there be an reason to use a small sample?

If that is true, shouldn't software developers warn their clients that anything less than a very large database is unsuitable for predictive purposes? Or is it simply caveat emptor and too bad for the newbies?

[sjk]

I have never used anyone else's software so I don't know what is available or what caveats are presented. I would have my doubts about anything that made significant revisions based on small slices of data.

What is latin for "bettor"? I think "bettor beware" is good advice in general.

[Jeff P]

posted by Traynor - If that is true, shouldn't software developers warn their clients that anything less than a very large database is unsuitable for predictive purposes? Or is it simply caveat emptor and too bad for the newbies? Yikes - here we go again... :bang:

Warn them about what? The algorithms I've created are based on extensive research... research that I conducted myself first hand... using databases spanning years and several hundred thousand races. The compound numbers generated by my software's algorithms - and I suspect the algorithms of a great many other handicapping software programs on the market today - are hard coded into the program. Individual horses like the one you pointed out won't change the output of those algorithms one single bit - not unless the author/developer decides to go back in and re-code the application and release a new version.

-jp

.

[GameTheory]

You should be able to do that yourself.No -- YOU should because you're the one claiming it is standard procedure. But you can't because it doesn't exist. You really go will to any length to avoid admitting you've made an error, won't you? I'll leave it to your therapist...

[traynor]

No -- YOU should because you're the one claiming it is standard procedure. But you can't because it doesn't exist. You really go will to any length to avoid admitting you've made an error, won't you? I'll leave it to your therapist...

If you are really curious, you can discover it for yourself.

[traynor]

posted by Traynor - Yikes - here we go again... :bang:

Warn them about what? The algorithms I've created are based on extensive research... research that I conducted myself first hand... using databases spanning years and several hundred thousand races. The compound numbers generated by my software's algorithms - and I suspect the algorithms of a great many other handicapping software programs on the market today - are hard coded into the program. Individual horses like the one you pointed out won't change the output of those algorithms one single bit - not unless the author/developer decides to go back in and re-code the application and release a new version. -jp .

Are you saying that 10 people using your software will all rate the horses in a given race in the same way? And all will select the same horse as "best," the same horse as "second best," and so on?

[traynor]

I have never used anyone else's software so I don't know what is available or what caveats are presented. I would have my doubts about anything that made significant revisions based on small slices of data.

What is latin for "bettor"? I think "bettor beware" is good advice in general.

Try AllWays automatic updates.

[PaceAdvantage]

Try AllWays automatic updates.

I knew AllWays would come up sooner or later....

[traynor]

I knew AllWays would come up sooner or later....

Sorry. Is that topic off-limits? I did not mention it in a disparaging way, but simply as an example of "automatic update" software.

[traynor]

Jeff,
Before this goes off the deep end, operationalizing terms may be in order. Specifically, "algorithm." I understand that you hard-wire code that does stuff (a somewhat inelegant way to express it) and that is called an "algorithm." If I "discover" a process that consists of using your application to find the top jockey, on a horse with the highest average purse value, in a turf race that has been switched to the dirt--that is also an algorithm. "Algorithm" simply refers to a step-by-step procedure for solving a problem, or accomplishing something; it includes hard-coded formulas, but also includes a number of other things.

[PaceAdvantage]

Sorry. Is that topic off-limits? I did not mention it in a disparaging way, but simply as an example of "automatic update" software.

No, why would it be off limits? It's just that since the beginning of this thread, I've always had it in my head that you were referring to software like AllWays....

[Jeff P]

posted by Traynor - Are you saying that 10 people using your software will all rate the horses in a given race in the same way? And all will select the same horse as "best," the same horse as "second best," and so on?In terms of ability in the each of the following areas: Early Speed, Late Speed, Total Pace, Class, Form, and then all of those separate areas compounded into JRating, Win Probability, and Odds Line:

YES.

In terms of the built in models that ship with the software selecting a single horse when all of the rules of the model converge on a single horse that qualifies as a play under the rules of the model:

There is NO gray area... NO user interpretation... just the name of the qualifying horse listed on a separate report...

YES.


-jp

.

[Jeff P]

posted by Traynor: Jeff,
Before this goes off the deep end, operationalizing terms may be in order. Specifically, "algorithm." I understand that you hard-wire code that does stuff (a somewhat inelegant way to express it) and that is called an "algorithm." If I "discover" a process that consists of using your application to find the top jockey, on a horse with the highest average purse value, in a turf race that has been switched to the dirt--that is also an algorithm. "Algorithm" simply refers to a step-by-step procedure for solving a problem, or accomplishing something; it includes hard-coded formulas, but also includes a number of other things. I understand that. The Newbie User of my software is given profitable models out of the box to start with that were created by the author/developer: Me. After they gain enough experience, they of course are no longer Newbies. Once their own experience and data research allow them to discover profitable areas outside the scope of the beginning models that I provide, then they do have the ability of course to create store and use profitable algorithms of their own. I do need to stress that with any active user defined model in my software there is NO gray area or user interpretation involved whatsoever. The rules for the models are absolute. If a horse qualifies as a play under the rules for the model it is a play and will be listed as a play on a separate report. If the horse does not qualify as a play under the rules of an active model then it is not a play and will not be listed on the report that lists the plays. The information is black and white - True or False - it's a pure yes or no decision with no gray area whatsoever.

-jp

.

[traynor]

posted by Traynor - In terms of ability in the each of the following areas: Early Speed, Late Speed, Total Pace, Class, Form, and then all of those separate areas compounded into JRating, Win Probability, and Odds Line:

YES.

In terms of the built in models that ship with the software selecting a single horse when all of the rules of the model converge on a single horse that qualifies as a play under the rules of the model:

There is NO gray area... NO user interpretation... just the name of the qualifying horse listed on a separate report...

YES.


-jp

.

Then I apologize, because my comments were in no way directed to an application such as yours. It sounds like something really worthwhile.

[traynor]

posted by Traynor: I understand that. The Newbie User of my software is given profitable models out of the box to start with that were created by the author/developer: Me. After they gain enough experience, they of course are no longer Newbies. Once their own experience and data research allow them to discover profitable areas outside the scope of the beginning models that I provide, then they do have the ability of course to create store and use profitable algorithms of their own. I do need to stress that with any active user defined model in my software there is NO gray area or user interpretation involved whatsoever. The rules for the models are absolute. If a horse qualifies as a play under the rules for the model it is a play and will be listed as a play on a separate report. If the horse does not qualify as a play under the rules of an active model then it is not a play and will not be listed on the report that lists the plays. The information is black and white - True or False - it's a pure yes or no decision with no gray area whatsoever.

-jp

.

That is exactly what we have been doing, and exactly our argument; if you have to "interpret" ratings, or if 10 people using the software all come up with different "predictions," the software is not particularly useful, other than for recordkeeping and number crunching.

Again, I apologize for anything that you may have interpreted as hostility or as demeaning your application. It obviously does not fit into the category of software to which I was referring. It sounds like an impressive bit of work, and I appreciate you taking the time to explain it.

[Tom]

That is exactly what we have been doing, and exactly our argument; if you have to "interpret" ratings, or if 10 people using the software all come up with different "predictions," the software is not particularly useful, other than for recordkeeping and number crunching.



In your eyes, maybe, but many other people are using non-black box software, that allows us to land on different horses, and are doing just fine. Not for record keeping. Frankly, I have no use for a program that gives me a number to bet.

[Little Tommy]

I've lurked here for a while and found PA to be quite interesting and resourceful for the most part.

I do find Traynor to be quite an enigma I must say. Why?

1. He's repeatedly pointed out how reluctant most anyone with anything really profitable, whether a system or selections, would be in sharing this information, even for a price, as the mutuals would generally be so impacted that it would be of little value to provide these tools to others considering the cost to the promoter's own betting profits.

2. He's ridden Anderon unceasingly while putting up selections himself that he has not presented cumulative ROI's for as of late. He did it briefly I remember, but until a day or so ago, did not even post his daily ROI and is not currently posting his cumulative ROI. I wonder why? :p

3. In spite of his argument in point #1 (which I've paraphrased from a number of his posts), he sells, or I imagine attempts to sell, his own supposedly highly valuable, mutual-reducing selections. The ROI figures on his site seem quite out of whack with the probable ROI of the selections he's posted on PA, an interesting note for sure.

4. Now he demeans the value of any program which leaves users to decide on their own selections rather than simply being a black box. That's hard to figure his take on that.

5. Then when presented with a seeming lack of documentation about his claim of standard outlier calculation by Game Theory, he suggests that Game Theory should search for the very documentation that he seems unable to provide to back-up his claim, instead of merely admitting that he spoke an untruth. Look, there's no shame in admitting to being a debate team dropout, come out of the closet already.

Since he's so willing to be such a harsh critic with Anderon, let's see his cumulative numbers since he's been posting here. I would assure you that the ROI is negative and probably below that of Anderon, so assuming I'm correct on that point, exactly where is his ground to speak on others picks. At least Anderon posts his results daily, which Traynor only began doing very recently after a period of no ROI postings.

You seem to like to throw rocks, but I see a whole lot of glasswork on your house.

[traynor]

<Since he's so willing to be such a harsh critic with Anderon, let's see his cumulative numbers since he's been posting here. I would assure you that the ROI is negative and probably below that of Anderon, so assuming I'm correct on that point, exactly where is his ground to speak on others picks. At least Anderon posts his results daily, which Traynor only began doing very recently after a period of no ROI postings.>

I thought Anderon was the new fair-haired boy, not because of the accuracy of his predictions, but because of the number of page hits his postings create. I have made no negative comments about Anderon, his software (linked by various "others"), or his strategies, or his "teaching seminars in handicapping" in some time. I don't even bother looking at his threads, because there seems to be nothing there of value. Is that what bothers you? Or are you simply yet another of the many faces of Anderon and his "marketing crew"?

[traynor]

<1. He's repeatedly pointed out how reluctant most anyone with anything really profitable, whether a system or selections, would be in sharing this information, even for a price, as the mutuals would generally be so impacted that it would be of little value to provide these tools to others considering the cost to the promoter's own betting profits.>

Off the wall.

[traynor]

<2. He's ridden Anderon unceasingly while putting up selections himself that he has not presented cumulative ROI's for as of late. He did it briefly I remember, but until a day or so ago, did not even post his daily ROI and is not currently posting his cumulative ROI. I wonder why? :p>

When you use simplistic, do-exactly-as-I-say-dummy methods, with a specific strategy clearly defined, stating an ROI (for that specific strategy) is feasible. The reason I stopped posting ROI is much simpler; I changed the format to graded ratings, which can be used in numerous ways, and those various ways have been posted along with the ratings. There is more than one way to bet a horse race, and to cover every possible eventuality would require either a two pages of instructions with every posting, or spurious claims of "success." I don't care to do either, mainly because of the time involved.

[traynor]

<3. In spite of his argument in point #1 (which I've paraphrased from a number of his posts), he sells, or I imagine attempts to sell, his own supposedly highly valuable, mutual-reducing selections. The ROI figures on his site seem quite out of whack with the probable ROI of the selections he's posted on PA, an interesting note for sure.>

And this surprises you because???? It is VERY clearly, and repeatedly stated that free samples are for informational purposes only, and are not intended to be as accurate as the other ratings. Duh. You get what you pay for, dude. To imagine you would get the same ratings for free that others pay (a substantial amount, soon to be increased) for is wishful thinking and a reality disconnect.

[traynor]

<4. Now he demeans the value of any program which leaves users to decide on their own selections rather than simply being a black box. That's hard to figure his take on that.>

Is English your native language? I demean the value of any software application that fails to truncate outliers when building "models," "track profiles," "automatic updates," or any other recommendation for wagering that uses small samples. Period. The outliers distort the "predictions" of those models.

[traynor]

<5. Then when presented with a seeming lack of documentation about his claim of standard outlier calculation by Game Theory, he suggests that Game Theory should search for the very documentation that he seems unable to provide to back-up his claim, instead of merely admitting that he spoke an untruth. Look, there's no shame in admitting to being a debate team dropout, come out of the closet already.>

Off the wall. Not worth a response.

[traynor]

<Since he's so willing to be such a harsh critic with Anderon, let's see his cumulative numbers since he's been posting here. I would assure you that the ROI is negative and probably below that of Anderon, so assuming I'm correct on that point, exactly where is his ground to speak on others picks. At least Anderon posts his results daily, which Traynor only began doing very recently after a period of no ROI postings.>

My question would be, why does "Anderon" (not the name used when he submitted the link to his "seminar software tote application whatever" software some months ago) feel so threatened by me? It looks to me like a hissy fit precipitated by a prolonged losing streak.

Please provide links to any "harsh criticism" I have made about Anderon since PA told me to stop posting such. This may come as a surprise, but I don't care about Anderon, his schemes, strategies, seminars, tote-board apps, or anything connected to them. They are not interesting.

[Little Tommy]

<Since he's so willing to be such a harsh critic with Anderon, let's see his cumulative numbers since he's been posting here. I would assure you that the ROI is negative and probably below that of Anderon, so assuming I'm correct on that point, exactly where is his ground to speak on others picks. At least Anderon posts his results daily, which Traynor only began doing very recently after a period of no ROI postings.>

I thought Anderon was the new fair-haired boy, not because of the accuracy of his predictions, but because of the number of page hits his postings create. I have made no negative comments about Anderon, his software (linked by various "others"), or his strategies, or his "teaching seminars in handicapping" in some time. I don't even bother looking at his threads, because there seems to be nothing there of value. Is that what bothers you? Or are you simply yet another of the many faces of Anderon and his "marketing crew"?

I have no relationship whatsoever with Anderon. The jealousy (not there's anything to be jealous of with regards to Anderon mind you) doesn't wear well on you, try another shade of green. You've made no comments about Anderon? You only further polluted his threads for weeks. What memory killing meds. are you on this week, oh dumb question, since if you're on them you naturally wouldn't remember. So how long have you been a master comedian?

[Little Tommy]

<1. He's repeatedly pointed out how reluctant most anyone with anything really profitable, whether a system or selections, would be in sharing this information, even for a price, as the mutuals would generally be so impacted that it would be of little value to provide these tools to others considering the cost to the promoter's own betting profits.>

Off the wall.

Again, come out of the closet, Debate Team Dropout.

[traynor]

<I do find Traynor to be quite an enigma I must say. Why?>

You need to get a life. If pondering why I am such an enigma is the best thing you have to do with your time, you have a serious, serious problem. I don't think anyone on this forum has even a moment's interest in the topic, including me. I think they would be far more interested in why Anderon's postings have become so sporadic since he started losing consistently. Perhaps time off to iron out the last few wrinkles in the software before release day?

[Little Tommy]

<2. He's ridden Anderon unceasingly while putting up selections himself that he has not presented cumulative ROI's for as of late. He did it briefly I remember, but until a day or so ago, did not even post his daily ROI and is not currently posting his cumulative ROI. I wonder why? :p>

When you use simplistic, do-exactly-as-I-say-dummy methods, with a specific strategy clearly defined, stating an ROI (for that specific strategy) is feasible. The reason I stopped posting ROI is much simpler; I changed the format to graded ratings, which can be used in numerous ways, and those various ways have been posted along with the ratings. There is more than one way to bet a horse race, and to cover every possible eventuality would require either a two pages of instructions with every posting, or spurious claims of "success." I don't care to do either, mainly because of the time involved.

Quote "That is exactly what we've been doing, and exactly our argument. If you have to 'interpret' ratings, or if ten people using the software all come up with different 'predictions', the software is not particularly useful, other than for record-keeping and number crunching."

And just how are your "selections" so useful in spite of the fact that since you've changed your format, since there are so many ways people can use your selections.

The point you're avoiding is this. You're still posting your selections to win, yes or no? If you're selections were creating a great ROI using them as you suggest you'd certainly let us know about the 45 ways you could have used them profitably. They're not doing so, so instead you try and weasel around the fact and point out how meaningless ROI numbers are since people can bet your selections in so many different ways. Please, Mr. Expert in Obfuscation. I suppose we're presumed to be too stupid to see that that's all there is in your argument.

If you're going to make the selections, be willing to post a flat bet ROI, or other method that you state in advance and cut the weasel routine, will ya? You obviously want to have your cake and eat it too, and it's ALL TO EVIDENT.

[Little Tommy]

<3. In spite of his argument in point #1 (which I've paraphrased from a number of his posts), he sells, or I imagine attempts to sell, his own supposedly highly valuable, mutual-reducing selections. The ROI figures on his site seem quite out of whack with the probable ROI of the selections he's posted on PA, an interesting note for sure.>

And this surprises you because???? It is VERY clearly, and repeatedly stated that free samples are for informational purposes only, and are not intended to be as accurate as the other ratings. Duh. You get what you pay for, dude. To imagine you would get the same ratings for free that others pay (a substantial amount, soon to be increased) for is wishful thinking and a reality disconnect.

You seem so concerned with the welfare of other information sellers "new customers", yet do you disclose to your subscribers that "Hey these profit figures I post for my paid selections are real, even though they aren't documented by any third party, and it's only those publicly available freebie picks which PA members can easily document, that totally suck horse dookie. Wildly profitable paid selections (supposedly) and KNOWN horribly negative ROI freebies, interesting contrast.

[Little Tommy]

<4. Now he demeans the value of any program which leaves users to decide on their own selections rather than simply being a black box. That's hard to figure his take on that.>

Is English your native language? I demean the value of any software application that fails to truncate outliers when building "models," "track profiles," "automatic updates," or any other recommendation for wagering that uses small samples. Period. The outliers distort the "predictions" of those models.

You post to Jeff P. that I'm referring to above was not regarding outliers, if you'll simply take another look, it was regarding the uselessness of programs that aren't black boxes. Are you really this full of BS or are you utterly forgetful?

[traynor]

In your eyes, maybe, but many other people are using non-black box software, that allows us to land on different horses, and are doing just fine. Not for record keeping. Frankly, I have no use for a program that gives me a number to bet.

I would assume that anyone else who was lucky enough, persistent enough, and intelligent enough to achieve your level of handicapping proficiency would feel the same way. That is not the issue. You were the first to respond on this thread, indicating you routinely correct for aberrant results. The issue was a caution for less experienced users that the inclusion of one or two monster mutuels could seriously tweak the "regression analysis" performed by most software that users tend to depend on for wagering models. Nothing more.

The issue was not, and is not, the use of software by skilled handicappers. If you have not learned to avoid distortions caused by anomalies, you don't last long enough in the game to become a skilled handicapper. My "cautionary note" was intended for those less experienced, and less capable than you.

Lastly, even the best black box software is hostage to the available odds at post time; "a number to bet on" is not quite what is produced. The production is, rather, a matrix of probabilities, each of which has a range of constraints. Without a sufficient level of handicapping skill, such software is of no more use to a novice than (sorry, I almost wrote "that software from Minnetonka") many other types of software. To call it a "black box application" would be a misnomer.
Good Luck

[Hank]

<3. In spite of his argument in point #1 (which I've paraphrased from a number of his posts), he sells, or I imagine attempts to sell, his own supposedly highly valuable, mutual-reducing selections. The ROI figures on his site seem quite out of whack with the probable ROI of the selections he's posted on PA, an interesting note for sure.>

And this surprises you because???? It is VERY clearly, and repeatedly stated that free samples are for informational purposes only, and are not intended to be as accurate as the other ratings. Duh. You get what you pay for, dude. To imagine you would get the same ratings for free that others pay (a substantial amount, soon to be increased) for is wishful thinking and a reality disconnect.

Traynor, come on man your better than that.:ThmbDown:

[Little Tommy]

<I do find Traynor to be quite an enigma I must say. Why?>

. If pondering why I am such an enigma is the best thing you have to do with your time, you have a serious, serious problem. I don't think anyone on this forum has even a moment's interest in the topic, including me. I think they would be far more interested in why Anderon's postings have become so sporadic since he started losing consistently. Perhaps time off to iron out the last few wrinkles in the software before release day?

At least Anderon "just started" losing consistently, whereas you have quite a heaq start on him in that department with a much longer track record of it.

Iron out the wrinkles? Now there's something you might want to consider, considering your track record.

You don't think anyone at this forum has any interest in exposing the hypocrisy and obfuscation that you seem to spew regularly? I do.

[Little Tommy]

To imagine you would get the same ratings for free that others pay (a substantial amount, soon to be increased) for is wishful thinking and a reality disconnect.

The reality disconnect is your post. What a hoot!!! Well, at least for your subsribers sake, there won't be any of them paying the higher prices, since there are no doubt none paying the current prices, unless you've somehow managed to extract money from imbodied spirits. I hear Casper the friendly ghost is quite interested in subscribing, by the way. You're better than Anderon and Ult. Selector combined in the nonsesical hyperbole dept. Keep up the good work, remember "It's never reality, It's always imagination" I like that motto better than Anderon's anyway, it's so much more truthful.

[traynor]

The reality disconnect is your post. What a hoot!!! Well, at least for your subsribers sake, there won't be any of them paying the higher prices, since there are no doubt none paying the current prices, unless you've somehow managed to extract money from imbodied spirits. I hear Casper the friendly ghost is quite interested in subscribing, by the way. You're better than Anderon and Ult. Selector combined in the nonsesical hyperbole dept. Keep up the good work, remember "It's never reality, It's always imagination" I like that motto better than Anderon's anyway, it's so much more truthful.


Again, is English your native language? If not, one should make allowances for the obscure, disconnected, rambling tone. Is there some point in this?

[traynor]

I have no relationship whatsoever with Anderon. The jealousy (not there's anything to be jealous of with regards to Anderon mind you) doesn't wear well on you, try another shade of green. You've made no comments about Anderon? You only further polluted his threads for weeks. What memory killing meds. are you on this week, oh dumb question, since if you're on them you naturally wouldn't remember. So how long have you been a master comedian?

Again, off the wall. I have not posted anythng on Anderon's threads for weeks. You must have me confused with someone else.

[traynor]

You post to Jeff P. that I'm referring to above was not regarding outliers, if you'll simply take another look, it was regarding the uselessness of programs that aren't black boxes. Are you really this full of BS or are you utterly forgetful?

Again, is English your native language? You seem to have difficulty processing language.

[traynor]

Translation: I got caught in the bathroom with a little boy, move along, nothing to see here folks.

Off the wall.

[traynor]

Traynor, come on man your better than that.:ThmbDown:

What you see in free samples are a segment. Because you don't see the entire view, the results of that segment may or may not replicate the results of the larger sample. To use Anderon's threads as an example, extract every fourth posting (or third or seventh), and use that segment as a criteria for evaluating his performance. Absurd, no?

Similarly, no matter how good the full sample may be (or how bad), a segment of that sample provides an inaccurate view of the full sample. In the case of selections, that means that extracting a number of representative races (before the races, before the post time odds are known) can rarely be as accurate as the full sample. Better results, worse results, possibly; but not an accurate representation.

The free samples, like the postings, are only a small segment of the day's races. There are ups, and there are downs. Neither is an accurate representation of the full sample of the day's races. That is why we very emphatically state that the samples are for informational purposes only.

[traynor]

At least Anderon "just started" losing consistently, whereas you have quite a heaq start on him in that department with a much longer track record of it.

Iron out the wrinkles? Now there's something you might want to consider, considering your track record.

You don't think anyone at this forum has any interest in exposing the hypocrisy and obfuscation that you seem to spew regularly? I do.

Is there some point in this, beyond a hissy fit by the disenfranchised?

[traynor]

The reality disconnect is your post. What a hoot!!! Well, at least for your subsribers sake, there won't be any of them paying the higher prices, since there are no doubt none paying the current prices, unless you've somehow managed to extract money from imbodied spirits. I hear Casper the friendly ghost is quite interested in subscribing, by the way. You're better than Anderon and Ult. Selector combined in the nonsesical hyperbole dept. Keep up the good work, remember "It's never reality, It's always imagination" I like that motto better than Anderon's anyway, it's so much more truthful.

The quote is, "It's never reality, it's always perception."

[Little Tommy]

What you see in free samples are a segment. Because you don't see the entire view, the results of that segment may or may not replicate the results of the larger sample. That is why we very emphatically state that the samples are for informational purposes only.

Blah blah blah, or should I say "Off the wall".

The informational purpose is to determine if there really are people on PA who would actually pay for Traynor's "service" once they digest enough of his losing "freebie" picks. I can answer that one for you Traynor, but you won't want to know the truth, seeing, at this time, relating to topics in this thread, the truth would be rather painful for you.

English is my fourth language by the way, but I've been wanting to learn another, can you tutor me in Obfuscation, I've noticed that you speak it better than anyone around.

[Little Tommy]

The quote is, "It's never reality, it's always perception."

Oh, it that what "the" quote is? It should be "It's never reality, It's always delusion" You know what I'm talking about. Kind of like the way you delude yourself into believing that you have subscribers.

I have more faith that Anderon is "feeding" $20,000 into the win pool than I do that any subscribers are "feeding" you for your hot, occaisional $2.80 Winners and $8.40 Exactas. Traynor's "CHALK DUST SPECIALS" ready for you to digest, comes with a free antacid chaser.

Come off the heavy meds and smoke some ragweed or something, you're just too delusional.

[Little Tommy]

Is there some point in this, beyond a hissy fit by the disenfranchised?

I don't know, again that's something for you to answer. Just like it's for you to answer Game Theory regarding your pompous claim that you can't back up.

Since you're the disenfranchised (no subscriber base) and you're the one unable to coherently and maturely address the valid points I've raised, I think it's quite clear who's having the hissy fit. Maybe your Trainer Diapers got messed and you're feeling a bit uncomfortable wallowing in your own dung?

[traynor]

Blah blah blah, or should I say "Off the wall".

The informational purpose is to determine if there really are people on PA who would actually pay for Traynor's "service" once they digest enough of his losing "freebie" picks. I can answer that one for you Traynor, but you won't want to know the truth, seeing, at this time, relating to topics in this thread, the truth would be rather painful for you.

English is my fourth language by the way, but I've been wanting to learn another, can you tutor me in Obfuscation, I've noticed that you speak it better than anyone around.

Do I think there are people on PA who would buy a rating service? Not in the least. Any more than they would buy Anderon and company's "handicapping seminars" or "tote-board software." Posting selections on PA as a business venue is not really good strategy, as Anderon will discover soon. Page hits do not equal dollars spent.

I post selections more for people to use than as a "marketing strategy," which may be why I am not particularly obsessed with whether the selections on a given day are up or down. I don't seem to share your performance anxiety. I just post selections that I think have a good shot at winning and producing a profit, rather than as a "demonstration" of my "handicapping expertise."

If I were into demonstrations (which I am NOT, especially for free), I would do like Anderon, and post every selection, for every race, at every track, over a period of time. I have never, and will never, even consider such a thing. It is a rather pointless display, given the venue; most of the people on this forum have gone way beyond the point of EITHER buying ratings or buying software based on "tote-board analysis" (whatever that might mean).

[traynor]

I don't know, again that's something for you to answer. Just like it's for you to answer Game Theory regarding your pompous claim that you can't back up.

Since you're the disenfranchised (no subscriber base) and you're the one unable to coherently and maturely address the valid points I've raised, I think it's quite clear who's having the hissy fit. Maybe your Trainer Diapers got messed and you're feeling a bit uncomfortable wallowing in your own dung?

Such eloquence! Should we tell everyone that you are upset because my spam filter has been tossing your e-mails into oblivion, and you feed on responses like some spoiled child ignored by his mommy?

[Little Tommy]

So you post your selections here, not for "Marketing Strategy", not for "demonstrations" but for people to use. How nice, but useless, since previous posts of your implied that you believe no one in their right mind ever shares profitable selections or information that could depress their mutuals.

So, you're not doing it for "Marketing Strategy" but you make sure a link to your site is on every post of yours.

What use could people have for your cumulative negative ROI selections? Don't you think they'd have a lot more fun and fascination coming up with their own losing picks? When did it occur to you that people need help losing money at this game, do you know numbers of people that are having problems acheiving that goal? Must be a terrible problem to have, but luckily you're here to save them from that fate.

For everyone tuning in, I have nothing personal against Traynor. I have a problem with his lack of ability to candidly address the obviously quite valid points I raised. If he'd simply address things in a straight forward manner I'm willing to move on without haste. Until then, I'm not going to sit back and mutely listen to his propaganda which he's trying to pass off as valid rebuttals.

So far, you're O for 5 and your statements are starting to look worse than your "Freebie" Selections, which is quite a testimony.

[Little Tommy]

The real "Traynor" comes out when pinned to the wall. Instead of addressing valid critiques it's nothing but this crap:

"Is english your first language"

" Nobody cares"

"Off the wall"

"You look it up, I could care less"

"How eloquent"

I've concluded that you're obviously a masochist who likes to be exposed as the pompous fraud you are. Lucky for you, I'll gladly oblige since I'm a recent graduate of Dominitrix U. Go ahead, spew some more mindless banter and I'll whip you again and again.

[Little Tommy]

You've claimed to be a "big bettor". That brings to mind a strategy for ending up with a small fortune betting Traynor's selections. It's quite simple really, "start with a large fortune"

[traynor]

So you post your selections here, not for "Marketing Strategy", not for "demonstrations" but for people to use. How nice, but useless, since previous posts of your implied that you believe no one in their right mind ever shares profitable selections or information that could depress their mutuals.

So, you're not doing it for "Marketing Strategy" but you make sure a link to your site is on every post of yours.

What use could people have for your cumulative negative ROI selections? Don't you think they'd have a lot more fun and fascination coming up with their own losing picks? When did it occur to you that people need help losing money at this game, do you know numbers of people that are having problems acheiving that goal? Must be a terrible problem to have, but luckily you're here to save them from that fate.

For everyone tuning in, I have nothing personal against Traynor. I have a problem with his lack of ability to candidly address the obviously quite valid points I raised. If he'd simply address things in a straight forward manner I'm willing to move on without haste. Until then, I'm not going to sit back and mutely listen to his propaganda which he's trying to pass off as valid rebuttals.

So far, you're O for 5 and your statements are starting to look worse than your "Freebie" Selections, which is quite a testimony.


And when you move on, another new poster arrives who walks like you, talks like you, and writes like you. What is your point? Your behavior is beginning to seem more and more like uncontrollable OCD in response to Anderon's losses. A pity.

[Little Tommy]

Based on your lack of candidly addressing your hypocrisy, grossly inconsistent statements, hyperbole and useless propaganda, it does not appear I will be moving on anytime soon. The only disorder I see around is your obsessive compulsive lying disorder. Were you born a congenital liar or did you study it intently in graduate school?

Anybody here want to point out any invalid criticisms I have made of Traynor:

1. When you are proved wrong, you only confess occasionally, and then in a over the top, mocking tone.

2. You claim that any software that does not simply spit out selections is useless, then you go back and clarify that you were only talking about it being useless for novices, so I guess you presumed the first time around you were speaking to a group of novices here at PA. Then you claim that you stop posting your ROI on your freebie selections because there are so many ways to play them.

In other words you are claiming that you ARE posting selections, as you certainly note the mutuals of your top choices, but then when it comes to showing an ROI number, all of a sudden, you are not posting selections but merely the scores and order of the selections your amazing program spits out, so they are not to be considered selections, unless of course a miracle $35 winner comes up once a year, (a true outlier) in your book, in which case you will be sure to point that out to us as one of your winners.

3. Game Theory quite patiently asks you to back up your claim and all you can do is get smart ass and dismiss his point, kind of like you did with all mine.

4. You post seriously losing selections (HUNDREDS, IF NOT THOUSANDS OF THEM) and then whine about the small sample being the reason your "freebies" are doing so poorly in relation to your NON-DOCUMENTED, NON-PUBLICLY RELEASED, paid selections, which are amazingly profitable, particularly in relation to the horrid performance of your "freebies".

Do you think that people here can not clearly see through all your rhetoric, and I do not mean that in anything but a perjorative context? That is what is really the most insulting about your posts. The fact that you think that people here are so dense that they buy any of your excuses and deflections. Your credibility, if you ever had any here, is seriously depleted, I would suggest that over the weekend you see if you can find somewhere you can buy some more, because you are really lacking there ace.

[Little Tommy]

Your behavior is beginning to seem more and more like uncontrollable OCD in response to Anderon's losses.

I see. It is all about Anderon. Is not that the guy you have never heard of, never seen any of his posts, find uninteresting, could care less about, are not obsessed with, so you say?

Funny that a guy such as that you take to be the root of the issues people have with your inconsistencies. You really should stop lying so much. When you actually tell the truth, you do not need to keep such an accurate record of what you have said in the past. In your case, it is obviously a full-time job to keep track of what you have previously said, since it usually will contradict what you are saying today.

We understand that you will make mistakes when attempting to cover your many tracks, it is OK, we have seen your type before and will see your type again after YOU MOVE ON, which should not be long, now that your weak cover of lies has been removed. Such a pity. You really should not take us all for such dupes, have a little more respect for our intelligence.

[traynor]

I see. It is all about Anderon. Is not that the guy you have never heard of, never seen any of his posts, find uninteresting, could care less about, are not obsessed with, so you say?

Funny that a guy such as that you take to be the root of the issues people have with your inconsistencies. You really should stop lying so much. When you actually tell the truth, you do not need to keep such an accurate record of what you have said in the past. In your case, it is obviously a full-time job to keep track of what you have previously said, since it usually will contradict what you are saying today.

We understand that you will make mistakes when attempting to cover your many tracks, it is OK, we have seen your type before and will see your type again after YOU MOVE ON, which should not be long, now that your weak cover of lies has been removed. Such a pity. You really should not take us all for such dupes, have a little more respect for our intelligence.

Ah, the imperial "we." So much like Richie. So much llke the Anderon clones. And so boringly pathetic. Smells like kitty litter.

[Suff]

I You really should not take us all for such dupes, have a little more respect for our intelligence.

you too. you signed up today after lurking....and have a blow by blow account of the last two months of anderon threads?


and your purpose registering was a hatfield/mcooy thread with Traynor?

no safe ground for you to land on no matter who you are.

and btw .. Who are you?

[Little Tommy]

you too. you signed up today after lurking....and have a blow by blow account of the last two months of anderon threads?


and your purpose registering was a hatfield/mcooy thread with Traynor?

no safe ground for you to land on no matter who you are.

and btw .. Who are you?

Yea, I'm forcing Traynor to obfuscate (his favorite activity next to that other bate) bs and throw out ad hominem attacks instead of addressing my simple but valid critiques. That's it, I must be his troll to make him look good, although I'm not doing a very good job of that, admittedly.

Are you threatening me there Suff, or is that Tuff? No safe ground to land on, what does that mean? Traynor won't fess up to being full of crap, simple. For his sake, I'd hoped he'd maturely address my points and we'd put the isssue to bed, no he'd rather weasel on as seems to be his nature, as revealed by this thread, so fine, we'll uncover his real character a bit more, if that's what he'd like.

Who are you? Give me your phone number, name, street address, and where your kids live, then I'll reciprocate, until then, nunya, boy.

[Suff]

Yea, I'm forcing Traynor to obfuscate (his favorite activity next to that other bate) bs and throw out ad hominem attacks instead of addressing my simple but valid critiques. That's it, I must be his troll to make him look good, although I'm not doing a very good job of that, admittedly.

Are you threatening me there Suff, or is that Tuff? No safe ground to land on, what does that mean? Traynor won't fess up to being full of crap, simple. For his sake, I'd hoped he'd maturely address my points and we'd put the isssue to bed, no he'd rather weasel on as seems to be his nature, as revealed by this thread, so fine, we'll uncover his real character a bit more, if that's what he'd like.

Who are you? Give me your phone number, name, street address, and where your kids live, then I'll reciprocate, until then, nunya, boy.

My Name is Michael Dudley
I live at One Devonshire Place, Boston 02109
My phone number is 617-733-0496

I have no kids.


No safe ground to land on implys... no matter who you are, or why your doing it.... its bizarre. You look wierd. Not wierd like me. I'm soft in a Kitten kinda way... meow.


btw... bout 600 pictures of me plastered all over this place as well.

now...who are you?

[Little Tommy]

Smells like kitty litter.

The only thing that smells worse than kitty litter is the 40% Profits you supposedly show on your PAID selections while returning a close to 25% Loss on your Freebie selections, funny how the ability to document those picks turns one set from -25% to 40% just like that.

Mr. I'm so concerned about the noobie customers of other system sellers, and you have the gall to post those obviously phoney results on your website. What's wrong, nobody buying your $300 a month losing garbage so you need to fudge the numbers a whole lot to pay off your graduate school loans, since you obviously can't make it in the real world.

Did you really have to attend graduate school to learn how to be a lying, stinkin tout? A lot of people have learned to do that spending a lot less money on their education than you did, too bad for you, now get back on your Trainer Bike and pedal on out of here unless you'd like more of your nonsense uncovered.

[Little Tommy]

My Name is Michael Dudley
I live at One Devonshire Place, Boston 02109
My phone number is 617-733-0496

I have no kids.


No safe ground to land on implys... no matter who you are, or why your doing it.... its bizarre. You look wierd. Not wierd like me. I'm soft in a Kitten kinda way... meow.


btw... bout 600 pictures of me plastered all over this place as well.

now...who are you?

Sorry, if you have no kids and can't tell me where they go to school, no go.
No safe ground to land on implys "it's bizarre"? Wow, that's a weird expression.

Look, no offense, but why don't you start a thread about who I am and how weird I am OK? I'll welcome it and throw my two cents in, everyone will be happy that wants to talk about weird, OK?

This thread was simply pointing out a number of hypocritical and other inconsistent statements and interesting anomalies between Traynor's public record on PA and his "supposed" winning record on his website.

I calmly layed out my points, he's spent 3 pages totally dodging around them, don't you think that is a little weird? Really it's not weird, it simply shows that he's full of crap largely, that's what it quite clearly shows. He refuses to address the points period. For a guy who says he doesn't want to discuss how many angels can dance on the head of a pin, he sure spends a lot of time theorizing about that to avoid a real discussion.

Why do you think that is? I'll tell you why, because I've pinned his misrepresenting inconsistent ass to the wall on every point and rather than clearly address it, he's rather try and weasel out of it all, isn't that clear?

Are you going to tell me that my harsh tone justifies his dancing around?
Is he too stupid to understand my points?
On his site he claims to have tremendously positive ROI's yet all we see here after hundreds of selections is probably a negative 20% ROI, don't you find that a little strange?

[Suff]

Are you going to tell me that my harsh tone justifies his dancing around?
?

I'm telling you that your a F***ing idiot.

Ok. We clear.

[traynor]

Yup. Definitely kitty litter.

[Suff]

Yup. Definitely kitty litter.

and so are you.

[Little Tommy]

I'm telling you that your a F***ing idiot.

Ok. We clear.

Quite. Glad to know your opinion. Is there anything else you have an opinion on that might be more meaningful or relevant? I'd certainly love to hear it.

Does the existance of this thread somehow diminish your existance?
Does it cause you pain? Does it make you wonder what planet I came from?
Is it keeping you up at night? Did aliens anally probe you and then force you to visit this thread, causing this thread to bring back extremely painful abreactions? Is this too many questions for you? Are you worried that the Bush worshippers here will turn your information over to the NSA and report you?

[Little Tommy]

and so are you.

Uh, Traynor, consider yourself Bitch-Slapped once again. Oh, but I forgot, being the masochist you are, you like it.

[Little Tommy]

Suff,

Thanks for joining our tryst. I never would have known you liked threesomes.
Now that you're here, make yourself comfortable. I'm sorry it's not quite the park bathroom style accomodations you're used to, but it will do. Let me finish giving Traynor all the whippings he's pleading for and then I'll get to you.

[Little Tommy]

How about some of those longshot exactas paying $8.40, or maybe a longshot winner paying $2.40, do you think you can get us one of those tonight?

No one else out there can provide those like you can. Abuttry gets us those $40.00 horses, but that's only every other day, you can always seem to get us one of those big winners of yours every day, out of about 36 selections. Come on Traynor, I desperately need a tax write-off, see what you can do for us ole buddy.

As a term of endearment, because I appreciate all you've done for my bankroll, do you mind if I call you "My Little Chalk Monkey"?

[Hosshead]

The reason for starting this thread didn't make sense.

To tell everybody that there was a $107.20 winner that would skew the results/ROI etc. of a small database sample ?

What, we don't know that?
What's the point?

[PaceAdvantage]

Who are you? Give me your phone number, name, street address, and where your kids live, then I'll reciprocate, until then, nunya, boy.

So, I guess you are retracting this statement? Suff puts it up, and you welsh....not good.

Boy has this thread gotten totally useless. Who knew anyone cared this much about traynor?

Talk about a shitstorm.....better get your parting shots in quickly....I smell a good ol' fashioned CLOSURE coming down....

[Little Tommy]

Two questions no one's discussed.

Did I have valid points regarding the consistency of his statements?

Isn't his track record on his website rather suspect considering the dreadful performance of his selections on PA?

Did he at all address these points in all the Kitty Litter he spewed about in this thread?

Don't his responses say a whole lot about his credibility?

Anyone like to come defend Traynor since he's apparently totally unable to defend his own words and actions?

[GameTheory]

Two questions no one's discussed.

Did I have valid points regarding the consistency of his statements?

Isn't his track record on his website rather suspect considering the dreadful performance of his selections on PA?

Did he at all address these points in all the Kitty Litter he spewed about in this thread?

Don't his responses say a whole lot about his credibility?

Anyone like to come defend Traynor since he's apparently totally unable to defend his own words and actions?My final thoughts.

Tommy, you are a bit suspect because you showed up just to weigh in on this thread, and you do seem a tad overly ... enthusiastic ... about it. Probably more to the picture there. On the other hand, you are 100% correct about Traynor.

And Traynor, you really do need a therapist. Trying to save face by acting as if you are a semi-retarded mental patient. Interesting strategy.

[traynor]

:lol:

[Little Tommy]

My final thoughts.

Tommy, you are a bit suspect because you showed up just to weigh in on this thread,

Suspect shuspect Game. No, I didn't show up to weigh in on this thread. I've been observing this Traynor for a while and got sick of his holier than thou bullcrap while running a service, the records of which, I believe are not accurate. If Mr. Obfuscation himself would have addressed the valid points I made I'd have made ONE post in this thread. He wanted to run and play weasel so I chased him down.

I'm not fond of folks like Traynor, particularly when they're overly concerned with the clients of other software developers, selection sellers.....

Did I get a certain thrill out of pinning him to the wall, yea, so I missed my calling as a prosecuting attorney, forgive me. Am I the one who's words dictated that this thread go on a wild goose chase as Traynor ran away, NO. He did, I merely persued. Had he come clean up front, like I said, I would have been done in one post.

I addressed him quite respectfully, he's the one that refused to address anything and wanted to throw Kitty Litter. I merely asked him questions, he's the one that wanted to spew bullshiite and play the clown. I'd say we know a lot more about Traynor after this thread. I already had strong suspicions about him and that's why I asked him the questions repeatedly.

I would guess that there's a few others here that would have liked to hear Traynor maturely addresss the questions, at this point, I'd doubt anybody would want to hear anything from this slimey cat who's too cheap to buy an ad for his "premium" service.

If you'd like, as I suggested to Suff, let's start 18 threads on psychoanalysis of me if you'd like, fine, I'm all for it. Please commence the Little Tommy psychoanalysis thread.

[traynor]

Two questions no one's discussed.

Did I have valid points regarding the consistency of his statements?

Isn't his track record on his website rather suspect considering the dreadful performance of his selections on PA?

Did he at all address these points in all the Kitty Litter he spewed about in this thread?

Don't his responses say a whole lot about his credibility?

Anyone like to come defend Traynor since he's apparently totally unable to defend his own words and actions?

Why would I bother? Anderon starts serious losing, so disappears into the woodwork. You pop-up like some little toy clown, bobbing your head and spewing drivel. Bad clown. Get back in your box. Anderon will have his app fixed soon, and return to his former (badly faded) glory. :lol:

[traynor]

"Little Tommy," how come you write just like Anderon? :D

[Little Tommy]

Why would I bother?

Ask Anderon if he knows me. He doesn't but since you're running low on Red Herring's he's a handy one to pull out of the hat again, and again, anything to avoid addressing your nature.

It's not why would you bother, you've bothered to for four pages now, it's that you're incapable of defending yourself because I've exposed you for what you really are.

By the way, you do display one sign of mental health, and one sign only. That is the ability to laugh at yourself, you're certainly deserving of it.

Can we take up a Vcash offering to help Traynor pay his bills? Please everyone donate $1,000 Vcash to help him. Maybe he can dope out one of those $2.80 longshots he hits every now and then and make a score. Doesn't anyone care?

[Little Tommy]

Anything for attention. Poor, pathetic little loser. :D

Now at least you've finally identified for yourself what we've been seeing in you all along. Good self-realization is good, you should pick up Swami Muktananda's teachings and carry on, although if I were you, I'm not so sure I'd want to know much more about myself. Maybe in your case, it really is better to live in delusion.

[Little Tommy]

"Little Tommy," how come you write just like Anderon? :D

Because we both speak English without obfuscation, something you might want to try sometime when you get some cash and don't have to keep your front as a profitable handicapper to buy your wonder bread. You must be terribly busy with that subscriber base of hundreds of thousands. No doubt Dave, CJ, Jeff P. Barry Meadow, Selvin, Andy Beyer, Steve Davidowitz, and every one else of prominence in the game is a subscriber to your amazingly profitable service.

Of course, everyone's keeping it on the down low, as no one wants to talk about it when they're hooked up to the hottest handicapper on the globe.
That must be a problem, being so profitable that no one wants to write a testimonial letter for you since they don't want everyone on board to lower those juicy mutuals, speaking of which, I can't remember the last time I hit a juicy $8.00 Exacta with a $2 Ticket. No wonder they're gaga over you.

I understand you're opening an investment firm to handle all the funds people want to throw at you, isn't the name Blind'em & Rob'em, how appropriate.

[traynor]

:lol:

[mainardi]

... Cashing big tickets is good. Betting large amounts of money on models that are based on anomalies and wishful thinking is bad.
now that you've painted yourself into that corner. ;)

Think about your last line. You may not have meant to, but you actually confirmed what I was saying. Wait... wait... think some more... now you've got it!

That's right. An anomaly is defined as "one that is peculiar, irregular, abnormal, or difficult to classify", and by betting what you KNOW fits your own wagering style is NOT an anomaly. Of course, if you are saying that you can analyze a bunch of races before actually betting to get comfortable with pushing money through the windows, I'll give you that... to a point. What happens if you look at it and you don't have the "courage" (cojones) to actually bet? And what happens if you PREMATURELY eliminate an "outlier", when maybe you haven't actually used the right algorithm to trim your dataset?

Trust me, I have an advanced degree in Artificial Intelligence, and the hardest thing to do is to make machines "think"... and neural nets (or any other algorithm, for that matter) can't sweat bullets when pulling a C-note out to drop on a theoretical selection... you have to have done it a few times to know the feeling (good AND bad).

As far as "wishful thinking" goes, I'm not convinced that depending on a potentially flawed algorithm to eliminate "anomalies" and "outliers" isn't in itself hoping that you got your algorithm tuned correctly... talk about wishful thinking!!! My belief is that -- oh, the blasphemy, coming from a guy that sells handicapping software! -- without the human condition as the counterpoint, we'd never get the FEEL of a winner... and that's what builds our confidence to put down the cash when we believe that we're right.

Not that I could ever convince the "stat freaks" (and I'm not singling you out, traynor) to change their minds... after all, I'm a big stat freak myself... but I don't go overboard overanlayzing things when it comes to the horses.

'Nuff said again... I gotta "practice" for the upcoming Tournament in Vegas... focus... focus... oh, have another beer, Joe!!!!!!!!!!!!!!!!!!!!!!!!!!!! :lol:

[PaceAdvantage]

God, what an incredible waste of time this thread has become.

Little Tommy is now banned. Anyone who devotes this much time to traynor automatically deserves the boot. Not to mention I had to edit his ass to kingdom come because he was breaking just about every law on the board.

In addition, Little Tommy is a re-branding of a formerly kicked nick......so right there is grounds for dismissal.

Carry on....

[xtb]

Nice


(RE: Joe's post)

[Tom]

It started out so nice...
Al least I got a few good websites - thanks guys for the links.

btw Little TOMMY is no relation! :eek:

[traynor]

now that you've painted yourself into that corner. ;)

Think about your last line. You may not have meant to, but you actually confirmed what I was saying. Wait... wait... think some more... now you've got it!

That's right. An anomaly is defined as "one that is peculiar, irregular, abnormal, or difficult to classify", and by betting what you KNOW fits your own wagering style is NOT an anomaly. Of course, if you are saying that you can analyze a bunch of races before actually betting to get comfortable with pushing money through the windows, I'll give you that... to a point. What happens if you look at it and you don't have the "courage" (cojones) to actually bet? And what happens if you PREMATURELY eliminate an "outlier", when maybe you haven't actually used the right algorithm to trim your dataset?

Trust me, I have an advanced degree in Artificial Intelligence, and the hardest thing to do is to make machines "think"... and neural nets (or any other algorithm, for that matter) can't sweat bullets when pulling a C-note out to drop on a theoretical selection... you have to have done it a few times to know the feeling (good AND bad).

As far as "wishful thinking" goes, I'm not convinced that depending on a potentially flawed algorithm to eliminate "anomalies" and "outliers" isn't in itself hoping that you got your algorithm tuned correctly... talk about wishful thinking!!! My belief is that -- oh, the blasphemy, coming from a guy that sells handicapping software! -- without the human condition as the counterpoint, we'd never get the FEEL of a winner... and that's what builds our confidence to put down the cash when we believe that we're right.

Not that I could ever convince the "stat freaks" (and I'm not singling you out, traynor) to change their minds... after all, I'm a big stat freak myself... but I don't go overboard overanlayzing things when it comes to the horses.

'Nuff said again... I gotta "practice" for the upcoming Tournament in Vegas... focus... focus... oh, have another beer, Joe!!!!!!!!!!!!!!!!!!!!!!!!!!!! :lol:

I think you may not understand the concept of outlier, or how it is used. You seem to be saying that outliers are somehow eliminated from wagers, in favor of "safe" bets, for those who lack cojones. The only outliers eliminated are in the models used to determine bets, and then ONLY aberrant prices that are likely to corrupt the model. Anyone who would not wager on a selection because the odds are too high (or who thinks that is the purpose of defining outliers) is missing the point completely.

[mainardi]

I think you may not understand the concept of outlier, or how it is used. You seem to be saying that outliers are somehow eliminated from wagers, in favor of "safe" bets, for those who lack cojones. The only outliers eliminated are in the models used to determine bets, and then ONLY aberrant prices that are likely to corrupt the model. Anyone who would not wager on a selection because the odds are too high (or who thinks that is the purpose of defining outliers) is missing the point completely.
Nope... it's your turn to miss the point.

I know what an outlier is (a data point "far" from others), and I know that statisticians love to throw them out... but were talking about them in the context of gambling, where throwing something out means that you don't plan on learning from it. Why on earth would you throw out a "corrupt" value? Maybe because your models are for statistical purposes, and not for wagering. And, NO, I never even implied that I toss high-odds horses, nor do I consider them outliers... so I can only assume that you were speaking in general terms, and not about me when talking about missing the point.

So, here's your chance to make your point. Without giving away any of your secrets, give us an example of a REAL outlier -- and word it in gambling terms, please -- and how you've used the elimination of it to make a profit. Conversely, since you no doubt regularly re-analyze your full datasets to ensure that you haven't produced a false outlier -- tell me that you do that, please, or else your credibility amongst statisticians will drop like the Dow this week -- tell us about a false outlier that you found that made you re-evaluate one of your algoritms.

[traynor]

Nope... it's your turn to miss the point.

I know what an outlier is (a data point "far" from others), and I know that statisticians love to throw them out... but were talking about them in the context of gambling, where throwing something out means that you don't plan on learning from it. Why on earth would you throw out a "corrupt" value? Maybe because your models are for statistical purposes, and not for wagering. And, NO, I never even implied that I toss high-odds horses, nor do I consider them outliers... so I can only assume that you were speaking in general terms, and not about me when talking about missing the point.

So, here's your chance to make your point. Without giving away any of your secrets, give us an example of a REAL outlier -- and word it in gambling terms, please -- and how you've used the elimination of it to make a profit. Conversely, since you no doubt regularly re-analyze your full datasets to ensure that you haven't produced a false outlier -- tell me that you do that, please, or else your credibility amongst statisticians will drop like the Dow this week -- tell us about a false outlier that you found that made you re-evaluate one of your algoritms.


Piece of cake, Joe. I built a model of races at Birmingham Greyhound a few years back, and accumulated about 300 races for each grade and distance. That is about 2000 races. I ran a normal regression that indicated a certain combination of factors was generating a very healthy profit. I was busy doing other things for a week or so, and when I re-ran another regression, the same combination of factors turned up a very healthy profit. Not as much as the original one, but enough to be impressive.

Not being overly shy in the betting department, I layed into the sucker with a vengeance. What I saw was a win percentage around 36%, and an ROI of well over 130%. I bet, I won, I bet more, I won more, but at the end of about 10 days I was down an embarassing amount of money, despite the fact that I was winning more than a third of the bets.

The size of the bets may have had an influence, because Birmingham is a fairly small track. Even given that, I still couldn't figure out what was wrong. So I went back and looked at the mutuel prices for every bleeping race at Birmingham for the past several months. Tucked neatly away off to the side was a $98 winner. It fit the exact profile that I was betting, except, for whatever reason, it paid $98.

My computer, being a dumb ass, dumped it in with the rest and averaged it to find the ROI. The combination of a high ROI and a healthy win percentage was irresistible; I continued betting beyond the point that any reasonable person would have done, because I "believed" in my algorithms.

I know this would have been better if it had been a horse race, but it wasn't. It was a greyhound race. And the way that made money for me was that I completely redesigned the applications I use for greyhound racing to truncate any mutuel that is more than $12. I could have built a function to calculate the "median of the interquartile range" or the "mean of the interquartile range," but lopping mutuels off at $12 took one line of code.

I re-ran all my greyhound mutuels with that one line changed, and discovered that several tracks that showed 110-120% ROI without truncation dipped to 90-95% when corrected for outliers.

So what? So I realized that my wonderful app was not nearly as wonderful as I thought it was. So, instead of doing whatever it is that people do when they discover they are losing money, I fixed it. I rewrote the algorithms to increase the accuracy, without diminishing the mutuels. It only took about six months of full-time, up-to-my-ears-in-computer-printouts frustration, but I fixed it. If it says there is an ROI of whatever, and a win percentage of whatever, it is relatively dependable, and consistent over time. Have I made money with it? Yes. A lot.

The bottom-line is that I have very little patience with convoluted explanations about why it isn't necessary to correct for outliers, or whether the "median" is better than the "mean." I can pretty well guarantee that if you are building a betting model based on 500-1000 races (not the underlying algorithms hard-coded into the application, but rather the betting models generated by regression facilities in many apps), if you don't correct for outliers, your chances of losing are substantially increased, while believing that the next big score is right around the corner.

When I saw the $107 winner at LaD, it reminded me of the race in Birmingham, as well as a number of months of extensive losses and extreme frustration using AllWays "automatic update" feature and downloading "models" created by other users that all claimed positive ROIs and lost money faster than even I could believe. So I started a thread, and said, essentially, "if you got a big price in your sample, it may mislead you." End of explanation.

I know it would have made a better tale if it had involved Sunday Silence and Easy Goer in the Preakness, half a dozen math professors, or a loud-mouthed "wise guy" from Philly chomping on a cigar while pontificating about the chances of the favorite in the next race, but it didn't. It was a dull, boring $98 winner at an obscure little track that most people don't even realize exists.

[mainardi]

but how does this translate to the "unwashed" out there that can rub their butt for free all day, but if someone bets them $5 to do it, they couldn't find it wearing oven mitts while looking in a mirror? ;)

In other words, you and I are SIMILAR (NOT the SAME) in that we analyze and then re-analyze (through whatever methods work for us) to try to improve... but what are you trying to say -- in an executive summary, because that's our collective attention span these days -- to those that buy someone else's data/reports WITHOUT UNDERSTANDING THE CONCEPTS OF OUTLIERS AND ANOMALIES? :confused:

Do they go on "blind faith" (because -- unlike you -- they've never actually placed bets with it before they get their hands on it... hence no wagering confidence) and then have to continue to buy updates to the data/reports or else be right back where they started? What I'm saying is that to understand outliers and anomalies, one would have to know how to create the algorithms, or else they would be at the mercy of the author to keep them constantly updated.

BTW... I think we do agree-to-disagree about the value of removing outliers and anomalies... but that's the beauty of analysis... nobody is going about it in exactly the same way. :)

[traynor]

but how does this translate to the "unwashed" out there that can rub their butt for free all day, but if someone bets them $5 to do it, they couldn't find it wearing oven mitts while looking in a mirror? ;)

In other words, you and I are SIMILAR (NOT the SAME) in that we analyze and then re-analyze (through whatever methods work for us) to try to improve... but what are you trying to say -- in an executive summary, because that's our collective attention span these days -- to those that buy someone else's data/reports WITHOUT UNDERSTANDING THE CONCEPTS OF OUTLIERS AND ANOMALIES? :confused:

Do they go on "blind faith" (because -- unlike you -- they've never actually placed bets with it before they get their hands on it... hence no wagering confidence) and then have to continue to buy updates to the data/reports or else be right back where they started? What I'm saying is that to understand outliers and anomalies, one would have to know how to create the algorithms, or else they would be at the mercy of the author to keep them constantly updated.

BTW... I think we do agree-to-disagree about the value of removing outliers and anomalies... but that's the beauty of analysis... nobody is going about it in exactly the same way. :)

It is not even necessary to do the actual adjustment (unless it is your own application, or you are using Access or something similar and can program it fairly readily). The only thing necessary is to be aware of the fact that ROI may be unrealistically inflated by outliers. It should be fairly easy to track the "average" mutuel by hand for a few weeks or a month (using whatever application) and get a rough idea of what to expect in a longer term. Outliers (if any) that occur during that period should be fairly easy to recognize and correct. Use that average multiplied by win percentage to give an ROI that is more in line with reality.

Are today's modern, sophisticated, knowledgeable computer handicappers aware of all this? Possibly, but most are too strongly manipulated by their high hopes to do anything about it. That may be the reason there are so many "hobbyist" handicappers who spend all their time "analyzing" computer readouts and making $2 bets.
Good Luck

[Murph]

God, what an incredible waste of time this thread has become.

Carry on....

Not from where I was sitting PA. This thread had it all. Information plus excitement, drama and action.

Thru post #65, there is the GT - Traynor debate, then we get bombed by lil tommy who is strikingly relentless. At post #101 Suff has seen enough, bull charges the thread with a matter of fact tell it like it is, including admonishing the home team in post #108. A touch of class. Then come the action scenes where we get to see you BOOT the guy.

Extremely entertaining for me. My interest in outliers is twofold in that our selection routines DO NOT consider them when making selections. Maybe it should in some form and the debate gives me different ways to consider doing it.

When evaluating the selections and making ratings and adjustments for wagering profiles I have always considered outliers on a case by case basis instead of with a statistical measure. I may try a method such as this.

There are mostly fine comments and opinions from a wide section of folks whom I would like to thank for making their opinions on the subject so clearly. Everyone who posted really.

Tom summed up the entire issue nicely for everyone in post #2, the rest of the posts were really a bonus. Great fun.

Murph

[prank]

Piece of cake, Joe. I built a model of races at Birmingham Greyhound a few years back, and accumulated about 300 races for each grade and distance. That is about 2000 races. I ran a normal regression that indicated a certain combination of factors was generating a very healthy profit. I was busy doing other things for a week or so, and when I re-ran another regression, the same combination of factors turned up a very healthy profit. Not as much as the original one, but enough to be impressive.

.....
So I went back and looked at the mutuel prices for every bleeping race at Birmingham for the past several months. Tucked neatly away off to the side was a $98 winner. It fit the exact profile that I was betting, except, for whatever reason, it paid $98.

My computer, being a dumb ass, dumped it in with the rest and averaged it to find the ROI. The combination of a high ROI and a healthy win percentage was irresistible; I continued betting beyond the point that any reasonable person would have done, because I "believed" in my algorithms.


I appreciate this thread on outliers, etc. It seems much could be improved, though, by looking into the bootstrap and cross-validation methods. Not only would you be able to mitigate outliers, but you'd also have some improvements in your model selection methods.

(Sorry for the first post in a long time - it's been a very, very busy summer, and I'm just catching up on PA and all the rest of the literature I'd expected to read.)