Single Best Factor?

[Molock]

What's the single best factor that seems to produce the highest (albeit a negative one) ROI? If someone were looking to just go to the races without the ability to thoroughly analyze the DRF, what would be a down and dirty way of doing it; a *gasp* just for fun approach. I've heard things like "bet the 2nd favorite to place," etc, which seems to have a modicum of success, but perhaps no better than picking colors, uniforms or names. Any thoughts?

[GR1@HTR]

BEL
ALW or Stakes top ranked MLO horse

40%
-2.00 ROI

[Lefty]

If I could do only one thing i'd look for the fastest horse at the 2nd call.

[Handle]

The thing is that it changes, from time to time, track to track, race type to race type. In some races you can cash in on some factors while they kill you in others.

It's also dependent on how you define each "factor" -- what I call an "ability". Some things, such as "2nd favorite" are more static, but even this is not invariable -- 2nd favorite _when_? 5 minutes to post? 2 minutes, when the gates open (or halfway through the race when the money is in)?

Here are some stats for most races in August (5692 of them total), Dirt/Turf -- everything:


(win% is based on starts, not races)

ROI Wise:

Dom E2+Late. There were 854 such races, 854 starts, 35.01% winners, +4.24% ROI.

Dom Speed was not far behind: 2025 races, 2025 starts, 32.69% wins, -4.55% ROI

First After Claim: 1016 races, 1329 starts, 16.4 % wins, -6.7% ROI.

For factors that appear in every race:
Best E1: 5691 races, 6335 starts, 19.98% Wins, -6.88% ROI

Best Speed: 5678 races, 6358 starts, 26.01 % wins, -9.93% ROI

Some High Win%:

Dom Prime Power: 1561 races 1561 starts, 40.10% wins, -10.17% ROI

Best Prime Power: 5673 races, 5676 starts, 30.66% wins, -13.04% ROI


The favorites:
ML: 5692 races, 30.27% wins, -19.43% ROI
PT: 5692 races, 34.8% wins, -15.75% ROI

Interesting...
First Time Starters: 975 races, 2571 starts, 7.66% wins, -39.25% ROI

First Lasix -- non first time starter: 904 races, 1173 starts, 11.43% wins, -31.79% ROI

Other "factors of interest":
Dominant Prime Power and Owner/Trainer:
299 races/starts, 41.4% wins, -1.67% ROI

Dom Prime Power and PT Favorite:
1164 races/starts, 47% wins, -4.14% ROI

Best E2 and Down in Class:
718 races/starts, 22.49% wins, +4.47% ROI

Best E2 and Recent Workout:
2203 races/starts, 23.23 % wins, -1.89% ROI

Take yer pick.

-Nathan

[JustMissed]

FIND OUT WHOSE THE BEST CAPPER/PLAYER AT YOUR TRACK AND STAND BEHIND HIM WHEN HE USES THE AMTOTE. LMAO.

JUSTMISSED

[MarylandPaul@HSH]

Originally posted by shanemooney
What's the single best factor that seems to produce the highest (albeit a negative one) ROI?


I do a lot of runs through my HSH database, and the factors that consistantly show the highest stand-alone $nets regardless of track or distance are Early Speed/Early Pace/First Fraction; best, 2nd best, and best 2 of the last 10.


MP

[smf]

the fastest final time of today's race.

NEXT!

[highnote]

The best factor I've found that will consistently produce the greatest number of winners is the final (or near final) odds. The horse with the lowest odds will win about 33% of the time.

Now, if you don't want to play every race, you might find a subset of races (say, $4,000 claiming) where there is a more powerful predictor than the final odds. However, I doubt it.

That said, someone once wrote that on the Aqueduct Inner Dirt track, at least I think it was the inner dirt, the 1-2 exacta made a flat bet profit in either sprints or routes.

Good luck and good betting,
John

[JimG]

Originally posted by smf
the fastest final time of today's race.

NEXT!

Even they lose occasionally. The stewards see to that.

[TenZin]

Ive got a computer with XP but I have a DOS program.
Can I run a DOS on XP?

[GR1@HTR]

Yes, you might have to download a couple .dll files to make it run.

[Jaguar]

A solid racing angle, beloved of the railbirds at Pimlico, box the inside 4 post positions at the home of the Preakness- in any route race- wins 62% of all exactas.

The tight turns mean the insiders get to the rail quick, and stay there. They run a shorter oval, while the outsiders have to expend more energy to get around them. Moreover, the outsiders have to fight physics and any tendency to go wide on those brutal turns.

All the best,

Jaguar

[Molock]

My father-in-law, after spending years amateur handicapping, finally gave up and strictly boxes the 1-2-3 posts whenever he goes to the races (which is regularly but fairly infrequent). Somedays he does really well, others he doesn't. Anyone know if this angle or the 1-2-3-4 wins on a consistent enough basis across all tracks to use, or is it too sporadic to even consider?

[Jaguar]

Shanemooney,

Doesn't work because the layout of most tracks allows the outside horses to get around the broad, sweeping turns.

All the best,

Jaguar

[Foolish Pleasure]

MarylandPaul is the only guy in the entire thread that is a winner.

Front speed is 4X as likely to be associated with a positive ROI then any single other quantified variable to date.

[Foolish Pleasure]

How misleading are small population statistics?

The favorites:
ML: 5692 races, 30.27% wins, -19.43% ROI
PT: 5692 races, 34.8% wins, -15.75% ROI


This stat over a sample size 15 times that has the ML fav winning the ROI battle by 1%.

[Handle]

Statistics are not misleading. Only the application of statistics can be misleading.

Your point about the ROI derived from ML and PT favorites is relevant to my initial point -- the efficacy of every handicapping factor is dynamic, it changes. But, they are not completely random. Some reasons for the change of a particular factor can be known. There are other reasons that we simply don't know. Why would a factor perform so well for, say, 5900 races and then bomb in the next 5000?

But patterns exist in the way certain factors perform. You don't have a factor that wins 30% of the time "at random". That's why I'm actually a very big fan of small population statistics taken from larger populations. Part of the key becomes finding out when a particular factor is performing well, and when its not. Your massive approach doesn't debunc my sample set in the least -- rather it shows the fallacy of relying on a massive sample set to prove something. This is as bad as saying a small sample set represents everything. In horse racing, the only thing an all encompassing, massive population is good for is to disprove everything. Congratulations. You've shown that horse racing is a dynamic sport.

-Nathan

[highnote]

Has anyone found a factor that can be applied to every race on every card at every track in the world that can predict the winner more than 33% of the time?

So far, I've only found one factor that comes close to a 33% strike rate -- the post time favorite.

[keilan]

Handle -- you are one sharp guy!! Continue to post.

[Handle]

swetyejohn ,

Are you asking about what factor, overall, will find 33% winners?
That is, if you look at a race -- ANY race -- then, given that factor, you have a 33% chance of winning.

If I understand what you're asking/stating, the nature of your question/statement implies that you're going to play every single race in order to get this statistic (or, not exactly play since you don't want to be on the favorite anyway, but maybe you'll use him with that 33% win rate in exotics....).

Back to August. At APX, in Sprints, the darned favorite won 42% of the time in 69 races with a _positive_ 2.32% ROI. Not bad for chalk. In MSW Sprints, the PT fave won 58.3% of the time (!) in 12 races for a +53% ROI.

Sure, that's a _tiny_ sample from the world of horse racing. But the point should be obvious given the huge difference between 33% and the percentages above -- you can't pick horses using a 5000 pound statistic the same way you wouldn't use a baseball bat to swat a fly on your forehead -- in both instances you end up knocking yourself out.

You don't need to go far to find tremendously wide variance in how well the PT favorite does, or how well any "factor" does for that matter. But, I have good reason to believe that you can make better use of this information if its packaged and broken down. To go back to the example above -- from 8/1 to 8/15 APX Sprint Faves won 38.24% of the time. From 8/16 to 8/31 they won 45% of the time. In other words, there was (is?) a definite trend in the favorites winning very often in APX sprints, clearly shown by the fact that the favorites didn't win every race on 2 or 3 days and lose every other race.

As of yesterday, PT faves were winning APX sprints at a 36.36% clip during September.

-Nathan

[Aussieplayer]

Nathan,

Out of curiosity (I am enjoying this), at what ROI? (For the PT favs. at APX in sprints) in September?

Also, how has the MSW sprints subcategory gone?

Cheers
AP

[Handle]

It's down (not that I'm saying it shouldn't be, or that I've been heatedly tracking the PT fave at AP, it was just the first example I dug out).

In september.
22 races, 36.36% wins, -15.91% ROI.

PT fave for MSW was 25% winners out of 4 races.

Now, what interests me is what's going to happen next. Will we see another flare up of post time favorites winning in the second half of the month? 36.36% is actually a high (too high!) rate for the faves to win at. Or is the trend going to duck down to 33, or will we have a "market correction" (sorry, couldn't resist) so that, overall, it gets closer to that 33% mark? Always tough to say, but those are the things I'm working on.

If super short term indicators are worth much, today saw the PT fav win % dip to 36% even.

On a slightly different, but related note, it may often be a good thing for systems players when the PT favorite wins a lot. It means (perhaps) that, overall, the predictable factors are predicting. ES's simulations are winning at a 32% clip in similar races (AP sprints) for a + 21.60% ROI so far in Sept. Oddly enough, the first part of August, which saw 38.2 % favorite winners, saw 32.35% sim winners for a 15% ROI whereas the latter half saw that huge 45% favorite winners and 31.43% sim winners with a -7.14% ROI. I'm still screwing around looking for "patterns" in this stuff, but I think they can be found and trends can be put to good use.

-Nathan

[Aussieplayer]

Thanks Nathan. Just interested in the same sort of thing is all.

I think it's a dilema to be honest, that I'm not sure has been at all solved. I'm a believer in big sample AND small samples - IN THE RIGHT CONTEXT.

However, back to the dilema, or racetrack tragedy: How often do you see such an example as you have provided happen? All too often! You get a thing that's winning in August - great! You put the money in, in September, and watch yourself bleed away, lol

Not trying to be melodramatic, it's one of those things!! I guess we just hope to hit a "wave" where the short term turns into the long term, and that's the key.

Cheers
AP

[Handle]

No doubt that its a dilemma, but I think you can pin it to the wall in several places.

You can never hope to stay on a trend forever. It still requires work figuring out what's going to happen next.

As an example, this AP race I've been looking at. Happens to have a Speed Index of "Slow". Speed Index is based on the Quirin model where you look at the top Quirin speed points of 3 or 4 horses in a race.

We are in Sept. where we've got favorites winning 36% of the time . The favorite is .6/1 and we're coming off of August where faves were winning 45% of the time toward the end of the month. Does this horse really warrant that 36% chance status, or even 33% status?

Remember that speed index. When looking at sprints that have a "slow" speed index, we have only 20% winning favorites from 15 races . The more refined we can make our analysis and still get a decent set of races to compare to, the more apt we are to find out how races really shape up on a case -by-case basis.

In my example race (happens to be the first race on 9/5 at AP that I brought up), the PT favorite is out gunned by the second favorite in a number of "abilities" or factors statistics wise.

Without the PT favorite angle winning 36% of the time, the choice is obvious. Hawk Royal has the Best E2+Late ability -- its been winning 38.89% of the time for a +80% ROI. He has the Best Speed, another 38.89% ability, and a trainer with dominant wins at the meet to boot (just a 2/5 ability with a positive ROI).

In contrast, the PT fave Fusto has the Best Prime Power -- a 26.67% ability with a -38% ROI, Best E1, an 18.75% ability with (-30%) roi and Best E2, a 20% ability with a (-26%) roi. This is an MSW race so I can understand the public jumping on the early speed (but with a trainer who's 0/20 at the meet....).

Fusto goes off as the favorite at .6/1. Hawk Royal rolls over him at 2.8/1.

-Nathan

[highnote]

Originally posted by Handle
Are you asking about what factor, overall, will find 33% winners? That is, if you look at a race -- ANY race -- then, given that factor, you have a 33% chance of winning.

If I understand what you're asking/stating, the nature of your question/statement implies that you're going to play every single race in order to get this statistic

Handle,
I just wondered if there could possibly be another single factor, other than PT odds, that could pick 33% winners, day in and day out; every race on every card; everywhere.
John

[Handle]

Right -- I don't think there is even one such factor. The PT fave, overall, masquerades as this type of predictor. But, I'm trying to say that at the level you refer to ("every race on every card"), this mark of 33% is indeed misleading because you're applying a statistic culled from the universe of races to "every [individual] race on every [individual] card".

If you play every race on every card with no regard to every other race, perhaps this statistic is more meaningful - you might have a general expectation of 33% in a context so broad it becomes less meaningful. If you don't, then it pays to know how this factor measures up given the particular race you are playing.

As to other factors that, overall, hit at 33% of the time, there are a few numbers that do well but with a negative expectation. Prime Power is near 30% I believe but I'd have to check.

-Nathan

[highnote]

Originally posted by shanemooney
What's the single best factor that seems to produce the highest (albeit a negative one) ROI? If someone were looking to just go to the races without the ability to thoroughly analyze the DRF, what would be a down and dirty way of doing it

Nathan,
I agree with you. However, my response was geared towards shanemooney's thread starting message. I think that PT odds are the "single best factor to produce the highest ... ROI". He didn't mention betting only certain races, so I figured maybe he wanted to bet every race on the card.

I'm certain if you segmented the races then there could be certain factors that could be applied to certain types of races. But then you'd need several factors and not just one.

I suppose I'm being somewhat semantical. I think both our views are correct.

John

[Handle]

Yes, I think I'm being too semantical in my response too.

[kenwoodall]

A good trainer placing the horse at the level it was last claimed at!

[Foolish Pleasure]

Statistics are not misleading. Only the application of statistics can be misleading.

Mumbo jumbo used to introduce subejctive analysis to an objective problem.

Your point about the ROI derived from ML and PT favorites is relevant to my initial point -- the efficacy of every handicapping factor is dynamic, it changes. But, they are not completely random. Some reasons for the change of a particular factor can be known. There are other reasons that we simply don't know. Why would a factor perform so well for, say, 5900 races and then bomb in the next 5000?

Bad news but the ROI in infinity plus one trials in a negative sum game will almost exclusively revert to the exact percentage that makes it a negative sum game to begin with. There is an enormous amount of unprovable supposition in this paragraph, you say it's dynamic, I say ultimately it's static.

But patterns exist in the way certain factors perform. You don't have a factor that wins 30% of the time "at random". That's why I'm actually a very big fan of small population statistics taken from larger populations. Part of the key becomes finding out when a particular factor is performing well, and when its not. Your massive approach doesn't debunc my sample set in the least -- rather it shows the fallacy of relying on a massive sample set to prove something. This is as bad as saying a small sample set represents everything. In horse racing, the only thing an all encompassing, massive population is good for is to disprove everything. Congratulations. You've shown that horse racing is a dynamic sport.

[b]More subjective dreamery, a big fan of small population statistics? Best of luck, they all revert to the take eventually, your delusion of picking out segmented samples and thus creating predictive information is laughable at best. Your part of the key is pointless unless you can anticpate what will or won't do well in the future.

Lastly, no your orginal ML v Off time chalk was misleading it suggests there is a notable, bettable difference when in reality the whole world knows they are in essence equivalent, the fact that you managed to dig up 5000+ races that didn't occur, has absolutely zero bearing on the next 5000.


-Nathan

Nice to meet you Nate, you guys crack me the F up.

[Handle]

Looky, looky, folks, another person to add to my ignore list!

[GameTheory]

Originally posted by Foolish Pleasure
More subjective dreamery, a big fan of small population statistics? Best of luck, they all revert to the take eventually, your delusion of picking out segmented samples and thus creating predictive information is laughable at best.

More insulting nonsense to avoid backing up your arguments.

[Derek2U]

Just a point about samples, sizes, etc etc.
At the financial comp I work for they have a big dept of Math/Stat guys; they invest millions in predicting just about everything to do
with money. One technique used is this: Say you have a database of 150,000 races. You draw RANDOM samples of say
sets of 300 races. Maybe you should draw 200 such RANDOM
samples of 300 races each. You draw conclusions on the success
of the DISTRIBUTION of those samples. This is highly preferable
to working with a distribution of the 150,000 races individually.
The KEY here is that you must take pains to make each race
drawn a RANDOM selection. This way you can talk about how
many samples of 300 races showed you went bust or made $$.
Your measure of Return & Variability improves with this technique.
In investigating factors themselves, things improve when you
stick to two (2) principles. State -- AHEAD of TIME -- what your
population of races will be. In my case, ONLY NY races at the
major 3 NY tracks. Therefore, according to these guys, I should
NOT use races from any other track. Then, use ANY method of
arriving at your factors. Then, after the factors are selected,
discard that SAMPLE of races FOREVER. Test the factors on a
FRESH sample taking repeated RANDOM samples of a particular
size (say 300 races each). Take many such sets of RANDOM 300's.

[Rick]

Derek,

You're right.


Everyone else,

What the heck are you talking about? I couldn't understand it so I had a couple of beers. That usually helps, but in this case I still don't get it. Explain please. Some people seem to be saying that nothing works. I can tell you positively, that's just not true. Some things work REALLY well and some things discovered from small samples don't work at all. You never know in advance how it's going to work out. After about 10-20 years you DO get a better idea of which ideas might work though, and your success rate improves tremendously.

One of the things people don't understand very well is that if you do enough studies, you'll find a lot of things that seem significant but really aren't. Usually, statistical significance is established at the 95% confidence level. That means that one out of twenty times the result would be achieved by chance alone. So, if you looked at twenty different things it wouldn't be impressive at all to find one that seemed significant. There are ways to deal with this problem but the simplest thing to do is to test it on a completely different sample.

Another thing is that you have to use is common sense when you're looking for significant factors. For example, if you compare morning line odds to actual odds, the differences will almost never be significant. If their IS a significant difference for some time period, then it will disappear soon because the information is available to everyone. Normally, about 80% of the variance in actual odds can be explained by the variance in morning line odds, so if you notice that the morning line oddsmaker is doing well, so will everyone else. It's just common sense.

[Tom]

Originally posted by Handle
Looky, looky, folks, another person to add to my ignore list!

Soundslike one of the "Franks" is out of rehab!

:eek:

[Tom]

Excellent points. A new feature in HTR lets us easily sample just
one track at a time. I prefer small, recent samples of the track I am playing alone. It has much more real meaning to me that data from tracks I never will play. I want to know what is going in the last 40-50 races. I use the long term stats as a baseline to see if things are differenet now than they have been.

[Jeff P]

More subjective dreamery, a big fan of small population statistics? Best of luck, they all revert to the take eventually, your delusion of picking out segmented samples and thus creating predictive information is laughable at best. Your part of the key is pointless unless you can anticpate what will or won't do well in the future.


Foolish Pleasure seems to be implying that nothing works. I can tell you from my own experience that this just isn't the case. There are lots of things that do work. Anybody with a well designed database can discover what works and what doesn't work if they look long enough and hard enough. True, racing is a dynamic game. Some things (that the public never really believes in) seem to work very well over an extended period of time and other things (Beyers are one example that I can think of) only seem to work until enough of the public catches on. And, of course, there are those things that can and do matter (handicapping in the paddock for example- if you take the time to develop a keen eye) that never make it into a past performance file.

It's always very easy in this kind of forum to shoot someone down (without offering proof of anything) and make it sound good while doing it.

So I'll ask a simple question: Foolish Pleasure- What factor(s) do you think truly matter in this game?

The name of this thread is 'Single Best Factor.' I'd be very interested to know if you have anything relevant to contribute.

[Rick]

What "works" is putting several things together. For example, if best early speed loses 5% and claimed last race loses 5%, if you put them together you might find something that wins. I'd expect something like that to get about 13% profit (.95 * .95 / .8 = 1.128) or maybe a little less since the factors aren't totally independent.

People who believe that nothing works are usually only looking at obvious single factors or combining correlated factors.

[wes]

Thorobrain5: Training inaccuracy is the control used to test how well the selector handicapps. Every 6th fact is removed from the training set for each selector and put into a "test set" This means that in a typical 2,000 fact network, 333 facts are reserved for the test set, leaving 1667 facts to train on. Periodically during the training, the selector attempts to "solve"
the facts set aside in the test set. Training inaccuracy is the measure of how well the selector performs on these facts. Lower the number, the better the selector will be at handicapping.


I think Huey Mahl said, the best way to become an educated guesser is through a combination of knowledge, common sense, analyses, and patience, among other things. Luck will take care of itself if these factors are apportioned properly.

wes

[hurrikane]

All some pretty good ideas folks.
I can only add..In No America..early speed is king. The only thing that gets close to a pos ROI as a one factor filter.

The good thing about the PT fav is they only win 33% of the time. If they won more I don't know if we could make money at this. We need the other 66% to make money.

One last thing...before everyone runs to Pim to box the inside 4 on routes..

to date this meet.
26 bets
11 wins
42%
$312.00 bet ($1 ex bx = $12.00)
$92.50 won
$-219.50
ROI 0.29

Railbirds..oh yeah..the ones that are always asking me to buy them a beer.

[andicap]

That Pimlico angle hasn't worked in years. I think Andy Beyer wrote about that some months back, maybe around Preakness time.

[Dave Schwartz]

Andicap,

What Pimlico angle?

Dave

[Jaguar]

Re: impact values,

the "true facts" of racing- taking the racing universe as a whole- are brilliantly and definitively analyzed in Mike Nunamaker's magnum opus.

Nunamaker derived his results from studying an enormous database of races and I believe it will be awhile before there are any demonstrable variations from these algorithms which will shed new light on handicapping.

All the best,

Jaguar

P.S. I define "railbird" to mean a guy who 2-years ago asked to borrow $20., never paid it back, and today asks for another hit.

This is when I reply: "Sure, I'll loan you $20.- just as soon as you pay me back the $20. I loaned you 2-years ago."

Invariably, they frown and walk away.

I stopped "lending" money at OTB in '94 and have never regretted it.

[hurrikane]

Dave,
earlier in the thread someone wrote in to box the horses 1-4 in all routes. Show a pos return on the exacta. I think they were just trying to get some more money in the exacta pool. :D
I made the mistake of taking them seriously. Should have known better up here. :D

[Dave Schwartz]

Hurrikane,

Thanks.

Dave

[Foolish Pleasure]

Jeff P,

Interesting interpretation of my post, naturally you missed the my first one in this thread, so let me paste it here for you,

MarylandPaul is the only guy in the entire thread that is a winner.

Front speed is 4X as likely to be associated with a positive ROI then any single other quantified variable to date.

And you contributed what exactly? that endless data mining will optimize one's results at the racetrack? I stand by my post, Rick's 5% chance is understated in a negative sum game.

Thanks again for Game's brilliant post, what proof would you like me to offer? Intro. to numerical analysis 101? Seriously, do I need to prove that 99.99% of the angles you mine will regress to the take over infinity plus one trials?

[Foolish Pleasure]

And one other thing

I can tell you from my own experience that this just isn't the case. There are lots of things that do work. Anybody with a well designed database can discover what works and what doesn't work if they look long enough and hard enough.

Anybody with a well designed database can tell you what WORKED IN THE PAST, finding predictive information for the future is a whole different ballgame and those of us that have been at the track for the last 30 years and don't sell books to supplement their income are well aware of what works and what doesn't.


www.gosellanotherbook.com

[GameTheory]

Originally posted by Foolish Pleasure

Thanks again for Game's brilliant post, what proof would you like me to offer? Intro. to numerical analysis 101? Seriously, do I need to prove that 99.99% of the angles you mine will regress to the take over infinity plus one trials?

No, no, just attacking the character of the poster instead of offering reasoned analysis as a rebuttal is VERY persuasive.

Really, it is.

[Jake]

Originally posted by Foolish Pleasure
Jeff P,

Front speed is 4X as likely to be associated with a positive ROI then any single other quantified variable to date.[/q]


I would like to see this data.

[Tom]

Originally posted by Jake


I would like to see this data.

What data would that be? Since db's are useless, there would be no data. Maybe this guy is Kreskin?
Re-read that single statment and you will see it is pure nonsense.
Or the results of the last race.

[ranchwest]

Maybe he meant to say front speed is 4x overbet?

[Jeff P]

Anybody with a well designed database can tell you what WORKED IN THE PAST, finding predictive information for the future is a whole different ballgame and those of us that have been at the track for the last 30 years and don't sell books to supplement their income are well aware of what works and what doesn't.


The game is dynamic. There is no guarantee that what has worked in the past will continue to work in the future. And a database only lets you see what has worked in the past.

But...

What conclusions can you draw as a horseplayer when you consistently apply a combination of factors that you have observed to have worked in the past- and this combination continues to work for you, more often than not, week in and week out. The ONLY conclusions that I can draw as a horseplayer is that you should: A. Keep Playing and B. Continue to evaluate factors.

I will agree, and over the years it has always amazed me, that about 99.5% of the new queries I run against the database show absolutely no promise. Thank God for the aberration- that half percent that DOES show promise.

[Jeff P]

And you contributed what exactly? that endless data mining will optimize one's results at the racetrack? I stand by my post, Rick's 5% chance is understated in a negative sum game.


What have I contributed so far? I guess nothing concrete. It's only fair- the name of this forum is Single Best Factor. So here's what I think truly matters:

I believe that track bias or track profile is paramount to consistently turning a profit at the windows. I want to have a clear understanding ahead of time just what type of horse might be dominant on today's surface. I'm not talking about just knowing in general whether or not speed is holding or whether closers are winning. I have written my own software that employs specific tools for analyzing (in detail) the running surface(s) and distance(s) of the tracks that I am going to play.

The first thing I do is run an analysis using a home grown Energy Distribution. This will tell me just what type of horse (in fact identifying a minimum energy distribution as well as a maximum Energy Distribution) has the best chance for success for each given distance at each track. I do this for several tracks and then, based on what I find, decide WHICH TRACK(s) I am going to play today. I think that far too many players simply choose to play a given track (say Belmont for example) because they think it may offer quality racing- as opposed to playing a track because they believe that they have a solid understanding of just what type of horse has the best chance to win on the surface there.

Once that has been done- I try to evaluate, within the previously identified enregy distribution ranges, which horse or horses in each race is the BEST FIT for the profile of that particular track and distance.

Does this work? Yes- at least for me- and more often than not. Unless there has been a change in the running surface I find that horses within the same energy distribution ranges continue to do very well at certain tracks and distances. Quite often these same track and distance profiles are successful from one meet to the next. For example, the profile I used for Del Mar's 2001 meet was every bit as effective for the Del Mar 2002 meet. Likewise, my 2001 Gulfstream profiles held up nicely for Gulfstream 2002. But this is not always the case. My 2001 Calder profiles seem completely useless for Calder 2002. Remember- I did say more often than not.

Well that's it. That's what I think really matters when you want to leave the track with more money than you came with.

So- FP- What do YOU think truly matters?

[Foolish Pleasure]

No offense Gameguy,

But you always struck me as one of the brighter candles int his box, I view your posts in this thread as nothing more than patronization.

The last time I posted a proof here, what happened?

About 3 dozen posts of,

"can someone please decipher this," then rather then anyone proposing a better risk of ruin model for this application, it was let's prove FP's equation wrong.

SOrry chum, I get nothing out of educating anyone, and you already know exactly what I was referring to. A basic seventh grade algebra class could derive the proof, so sorry if you can't.

TO the guy who wants to see the data, and TOm, the comedian.

Disagree with the post TOmmie? Think it's incorrect? Or just merely mocking? Why not just post, yes FP is correct, Fr1 is clearly the dominant factor???? Instead, like Game you what??? Don't want to believe in finite math, want to tell the folks, that XYZ is a +1.5 ROI, and will be forever??? C'mon you know better.

Lasty, to the Jeff guy, that's an interesting post, however only in the absolute extreme cases, will the track profile supercede FR1 as an indicator of future profitability, and FWIW you're dealing with a guy that's been keeping his own bias information since 1983. Before it's too late, realize new technology has made most advanatages extremely short lived and often there is value on the other side of your neat model, regardless of whatever data you'd like to spit out.

Lastly, to the guy that would like to see the data, yea I wish someone did all the work for me too over the years, just think how much more time I'd have had over the years with the family.

I stand by all my posts thus far. To the defensive crowd that refuses to admit 99.95% of all mined data is virtually worthless, YOU ALL FRICKIN' KNOW BETTER.

[Rick]

The 5% chance that I mentioned has to do with commonly used tests of statistical significance. It makes no difference whether it's a negative sum game or not. You can set the level anywhere you want to. Obviously it's going to be harder to find a method that has a 95% chance of being profitable than one that is just significantly better than the track take.

Having a database just allows you to test ideas more quickly. If your ideas aren't good then you won't be successful. Of course, experience helps here but testing a method by database is not any less reliable than testing it by paper and pencil.

[Foolish Pleasure]

Yes, I do enjoy reading your posts,

and yes am overtly familiar with hypothesis testing. I did not understand your original post, so my understatement post was not exactly topical to your post.

However, my point is the difficulty of finding angles that will return a pos roi is of course inversely proportional to the takeout.

[andicap]

Re: Jeff P's energy post:

Several years ago when I was using Thorovision, a great software program that is no more, it had two figures that I would consistently use to nail longshots that didn't figure that much on speed/pace, etc.

One compared a horses' energy distribution to the rest of the fields. I have a descrption at home on how it was compiled. Had to do with taking a horses' good races, etc. Anyway it was on a scale fro 0.1 to 8.0 with the lower the figure the earlier the energy. At most sprints, a figure over 5.5 was a complete throwout unless the pace figured to absolutely collapse. but often times - not always, but it wasn't uncommon -- tracks would have a fairly narrow energy distribution. That is, I'd model the winners figure (using a blocking feature to take out PPs that were aberrant because the energy figure used all of a horses' good races -- you'd look at a horse's record and his energy in 6 good races would be in the 800-900 range and the 7th would be a 600. So you'd chuck the 600 and just use the energy figues he commonly ran. )
Anyway, you'd sometimes have a pretty narrow model, say, from 1.0 to 3.5 so that a middling horse with a 2.5 would beat a faster horse with a 4.0 and pay $20.

The other figure measured where the horse liked to run in his good races -- that is what position he liked to be in and I'd model this too. Sometimes this model worked as well. Again on a scale of 1 to 8.

I sure wish I could recreate those figures. It worked amazingly well.

The only problem was I didn't believe the models sometimes and couldn't believe these horses had a chance.

[Lefty]

FP, whether you realize it or not, you're essentially agreeing with the DB guys. How do you come to the conclusion that f1 best factor? PAST RESULTS. There ya go.
Andicap, sounds like Thorovision kinda borrowed that energy distribution from Sartin.

[Show Me the Wire]

Owning a horse a must for every handicapper.

Not a bottom level claiming type and not the expensive future development type, and put some learning time in on the backside. This is a great way to understand what factors are and are not important.

Regards,
Show Me the Wire

Through sight colors may be seen, but too much color blinds us.

[keilan]

Jeff P -- In regards to your post on track profile and energy dispersion. This measurement has been around for a long time and yes it is still a huge advantage to those who implement in their handicapping process.

I'm sure you use many other handicapping factors before selecting the horse you wager on but the thread is titled "single best factor" and I agree this measurement has paid for itself many times over and will continue to do so for those that make their own numbers and are track selective.

There was a decent discussion on HTR website regarding this very thing. Titled "energy calculation" date March 2002.

[Tom]

Now I have just the opposite opinion of percent early-for me, it has been a worthless factor. Not to say the factor is worthless, but in my hands, it has no value. I have been tutored by Tom Brohamer himself, Howard Sartin, Glen Connely, heck, even Dick Schmidt slapped me and gave me an explantation once <G>.
For me, it is a fool proof method of throwing out winnners.
I applaud you if you make it work - I tried for years.

[so.cal.fan]

Show Me The Wire:

"Owning a horse a must for every handicapper.

Not a bottom level claiming type and not the expensive future development type, and put some learning time in on the backside. This is a great way to understand what factors are and are not important.

Regards,
Show Me the Wire

Through sight colors may be seen, but too much color blinds us."

It is always a pleasure to read a post from our resident"Taoist"
You are correct, of course!

;)

[Rick]

Although I've never owned a horse I've found it more than worthwhile to read some of the books written from the perspective of the trainer or owner.

[Jake]

Originally posted by Foolish Pleasure
[B]No offense Gameguy,

Lastly, to the guy that would like to see the data, yea I wish someone did all the work for me too over the years, just think how much more time I'd have had over the years with the family.

Front speed is 4X as likely to be associated with a positive ROI then any single other quantified variable to date.



Like you, I think early speed is a strong predictor. However, you're stating that front speed is 4X as likely to be associated with a positive ROI than any single other quantified variable. Really? I have looked at tons of data over the past 5 years and wish it was that damn easy! I would love to see the sample size and parameters that makes that case.

[keilan]

Tom -- if you are isolating the energy dispersion factor it will probably never make sense, but if it is looked at in conjunction with pace, class and form it is a superior factor. It not always points towards the winner but often steers one away from horses that look good in most all other categories.
Any software program which does distance equalization (they all do) often skews the energy factor and consequently the measurement is unreliable. My suggestion would be to only look at distances (races) which are within a furlong of today's race or ask that all measurements be based only on the actual distance raced. But this is only useful if one has studied where one wagers(track profile)
I believe for these reasons and some others that percent energy is a measurement often ignored.

[smf]

After going to La Downs today and watching what all came to the winner's circle time after time, I'd say the damn vetrinarian is the MVP of Bossier City these days.

Look at Eagle Lake's fractions in that 2 turn race. If you'd saw him in the paddock you'd swear the horse couldn't win.

[Bruddah]

The Hottalkers site was promoting Eagle Lake to be the winner for the last two days. That site puts out a lot of winner tips almost daily. I have made several bets on horses that have been promoted on the site and I am way ahead on those tips.:cool:

[smf]

Bruddah,

Nice to know they give winning tips @ HT. They seem to believe they are the only ones that do so. I'll keep the money I won today (w/o having to look at their picks)..., that is, if it's ok with the Ned Beatty chasers over at HT.

Eagle Lake kicked up more dirt in the walking ring than Earl Weaver ever did at an umpire before getting tossed.

[MarylandPaul@HSH]

Originally posted by Tom
I have been tutored by Tom Brohamer himself, Howard Sartin, Glen Connely, heck, even Dick Schmidt slapped me and gave me an explantation once <G>.


Tom, are you still in contact with Glen Connely? Haven't seen him in years. We shared a table at the Laurel Sports Palace nearly every racing day for a long time, exchanging ideas, burning laptop batteries (running "Energy" mostly), and choking down ten dollar club sandwiches.

Hell, we live not too far from each other, just lost touch after awhile. Hope he's doing well.


Paul

[andicap]

Originally posted by Lefty
FP, whether you realize it or not, you're essentially agreeing with the DB guys. How do you come to the conclusion that f1 best factor? PAST RESULTS. There ya go.
Andicap, sounds like Thorovision kinda borrowed that energy distribution from Sartin.

Well, one of the co-founders of ThoroVision, Bill Burns, was an old Sartin guy (he's acknowledged in the Pace Makes the Race book);
I just used it in a different way than everyone else. You see we also had a median energy figure called MPPR on a scale to 1000, the higher the earlier. But modelling that figure didn't work as well as modeling the other figures I described above. You could eliminate horses at the extremes but that was about it.

I have a description of the really excellent energy figure if someone is interested in looking at it and maybe programming it.

[Tom]

Haven't heard from Glen since the last Sartin outing in Baltimore in the early 90's. He was a nice guy and a big help with T-mation.
If nothing else, he was systematic in his approach. That is what I miss most about the Sartin group - meeting the people at the seminars. I would go to HTR seminar, but they are always in LasVegas in July, which is my busy season. It was always nice to get together with fellow users and sort or re-charge the batteries.

[GameTheory]

Originally posted by andicap
I have a description of the really excellent energy figure if someone is interested in looking at it and maybe programming it.

I'm interested.

[Rpd]

I'm interested too, Andicap. Thanks!

[andicap]

Anyone interested, please send me an email -- its easier that way -- that includes Game Theory and Rpd.

There is one hitch. In the definition, the author said it uses the horses' "Good races," but he would not define -- probably could not recall -- his definition of good race. I've seen all kinds of definitions.
In the money and within 6 lengths of the leader?
The good race definition from Ken Massa on the HTR board that touches on early speed.

On and on.

[leader]

This is my first post on this board as I'm the new kid on the block!
But the most outstanding race factor is in one word[FRESHNESS]
or [CONDITION] is the [KEY]!
[1] DAYS SINCE LAST RACE
[2]DAYS SINCE LAST WORK
Then all other racing factors fall into place!
If you check any of your past races where you got beaten by another horse,Ithink you will find that condition is what beat you!
How many times have you and I seen this happen or where did he come from at the finish line!
Well,that is my throughts on this subject!
Lefty,did you get my e-mail!
GUS

[so.cal.fan]

Leader:
Welcome to the board.
You are right about condition!
There are two factors that absolutely CONTROL horse races.
CLASS and CONDITION.
I look forward to more of your posts.
;)

[Dick Schmidt]

SoCal,

Of course, being able to run really really fast helps too.

Dick

[Show Me the Wire]

Dick:

running fast helps only if they can run fast for more than 5 1/2 furlongs.

Show Me the Wire

[so.cal.fan]

Show Me The Wire:
Would you define class as "the ability of a horse to carry it's speed over the distance of ground nature intended"?

[Show Me the Wire]

so.cal.fan

Yes that and more.... the desire, instinct or whatever a horse feels to make it want to be in front of the pack.

Regards,
Show Me the Wire

Nature acts without intent therfore its acts are without malice or kindness.

[so.cal.fan]

"Nature acts without intent therfore its acts are without malice or kindness".

Clever stuff, that Tao. I'll bet you're a very good handicapper, SMTW!

I still think that class and condition go hand and hand, you can't seperate them, so I guess there are TWO main factors.

;)

[Jeff P]

For me, condition is very important. Trouble is, I don't think that it is all that easy of a thing to nail down with factors you can pull from a database. I've had better success using database factors to identify horses that can't be in condition than identifying those that are in condition. When you throw out the ones that can't be in condition what you are left with are those that might be in condition.

To identify horses that are in condition I've had the most success trying to judge condition from the way a horse looks and acts prior to the race.

Sadly, last time I looked, BRIS didn't have an in 'condition today' (Y/N) field in any of their past performance files...

[Dave Mark]

Wire:
Commend you on,
Great reply to software dilema.

Horses run as fast as they can because of FEAR!

And depending on that degree OF Fear, be it real or perceived.

Once that loud opening bell rings in the horses ear,the gate opens the horse uses its powerful rear legs to spring forward and run from what it believes to be danger. First of all the Gate represents danger to a horse, and it Wants to flee as quickly as possible.

Second, other horses are running away from this perceived danger, and horses are heard animals.
This is instinctive.

Of course the jockey is supposed to take some control once the race begins, and hopefully rate his horse when he can, and if he is supposed to.<G>.

Only when hitting the stretch, a jockey will use the whip, the horse is frightened by this, and will try to run faster to escape this "danger" whether real or imaginery.

If you note, jockeys are NOT allowed to carry whips when racing 2 year olds!

[Show Me the Wire]

Dave Mark:

Thank you for your comments.

I agree humans attempt to control horses using fear to stimulate the horse's flight instinct to force the horse to the finish line. We can rationalize horses are spurred on in this matter because this is our frame of reference. But horses may have a different frame of reference and win races for other reasons.

If you have not had the opportunity take the time to watch horses school at the starting gate in the morning. Many experienced horses will stand calmly for extended periods of time in the gate and back out peacefully. On the other hand if the front door opens the horse will break out powerfully, even without the bell, the hooting and all the other commotion. To me this may indicate racehorses respond to something more than the fear factor instigated through the use of loud noise.

All I understand is the more I experience the less I know.

so.cal.fan.

I believe class is comprised of tangible and intangible elements. I define tangible as criteria that may be objectively measured. Conversely intangible criteria are not readily measured objectively. Adding to my definition of the tangible element of class, I include pace as an ingredient. I believe tangible class may be defined as the ability to carry speed over a distance of ground while coping with the pace of the race.

Since, I do not understand why a horse acts the way it does I cannot explain why a horse tries to run to the front of the pack. I do not know if its instinct to be a natural leader, its instinct associated with fear, its competitiveness, or simply its enjoying the attention it receives when it races that makes the horse strike to the front of the pack.

But I believe tangible class is enough to win a majority of races. The horse able to set or cope with the pace without serious competition and carry its speed over the distance of ground for that particular race will most likely win the race even if all the intangibles are not aligned perfectly.

My belief is there is not one single predictive factor for handicapping. My opinion is there are two interrelated factors tangible class and the condition book. Tangible class and the condition book together dictate the ability needed to compete in a race and shows whether the trainer understands the type of ability needed to compete.

I hope this is useful.

Regards,
Show Me the Wire

Contentment comes at no cost

[so.cal.fan]

SMTW
YES!

[Tom]

Interesting concept. How would you explain that a horse can carry its speed over X distance against Y pace, then fail to do so
against "better horses?"

Why does Horse A run a beyer of 82 against $10,000 claimers, then lose to a horse that runs an 80 in $15,000?

Not trying to argue with your theory, just hoping to expand it.
It really is a puzzling concept,class.

[keilan]

Tom -- That's why they're $10k claimers, lots of soreness, minimal consistency etc. No one expects them to string together two or three good races. Often they do, but when they don't no-one is really surprised.

[Jaguar]

Tom,

Keilan is exactly right. The best explanation of this that I have seen is in the late, great, Ray Taulbot's book: "Positive Factor Handicapping", the work that saved me a fortune over the years by getting me away from the mutuel window when the next race is for platers.

All the best,

Jaguar

[Show Me the Wire]

Tom:

I understand your scenario and my only explanation is intangible class. The criteria we cannot measure.

Becuase, I cannot understand intangible class and I know it is present I try to minimize its impact. I attempt to select wagering opportunities in which I believe tangible class in conjunction with the condition book outweighs intangible class.

Regards,
Show Me the Wire

The more I experience the less I know.

[ranchwest]

Most of the time, when fillies go against the boys, the filly might have comparable times and earnings, but when she hits the top of the stretch reality slaps her right in the face.

Class.

[freeneasy]

well I say, you can have all the sass and class, but Secrateriat with a feather on his back couldnt win if he wasnt in proper physical condition. A little mustard on the hotdog goes a long way ;)

[Dave Schwartz]

Free,

What is the record for most cliches in 2 sentances? <G>

Dave

[freeneasy]

in the famous words of the late Jimmy Durante " hot chachachacha, I gota million of em

[Chico]

I believe the ONLY no-brainer play that shows a positive ROI year after year is a place bet on
horses that go off as odds on favorites. You can't earn a living with it because the positive expectation is under +1%.

Regards