Dave Litfin

[andicap]

In this week's Simulcast Weekly, Litfin writes that Saturday's Belmont track played kind to speed then notes how the winners of the two route races on the day overcame the bias.
Here's why Litfin is a moron:

Any halfway knowledgable person knows that biases for sprints and routes often differ. A track can be playing extremely speed oriented but off the pace horses will be winning in routes. I don't know why, but go back and look at days when speed was winning everything at one distance but not at another.

Litfin gives extra credit to Sunday Break and the other route winner on the day because they came from off the pace. Now I know why these guys write about racing for a living instead of betting on it. I'd put half the people on this board up against Litfin any day of the week.

the track was more than just kind to speed. it was the biggest bias I've seen in a while. Check out the charts in the three sprints.

In the 5th race, the horses that were 1,3,4 at the 2f finished 1,2,3. The horse that was 2nd was 35-1.

in the 6th race the horses that were 1,2,3 at 2f finished exactly that way. I haven't seen Indian file harness races with results like that! The favorite, a closer finished out of the money. A 9-1 horse was 2nd and a longshot 3rd.

in the 9th race, the horses that were 1,2 at the 2f mark finished that way (albeit they were the two chalk, but still.)

The evidence is overwhelming that Belmont had a HUGE speed bias on Saturday.

[karlskorner]

Andicap;

In your mind Litfin may be a moron, but you have to admit he is a well paid moron. I guess he has a contract that requires that he produce so many lines and he hopes that people like you won't take the time to prove him wrong. He falls into my opinion of most the "writers" that the public are so hungry to sop up.

Karl

[andicap]

I think "moron" was a bit too strong. I apologize to dave for that -- he does put himself out there every day which i do respect and he writes well.

I dont criticize these guys for their picks..who can pick every single race a day or two in advance? Before scratches, without seeing them in the paddock, changing weather conditions, etc.

But not to delineate between a sprint and route bias is just idiotic. So Dave is not a moron, but what he wrote in that instance certainly was.


And how well paid is he??? I wonder how much these DRF guys get paid? Ever see how they dress?

[karlskorner]

Andicap;

I have seen Beyer just about everyday at the past GP meet, you would think he was dressed to cut his lawn, between him and his wife they pull down a good piece of change. I think it's a "code" amongst press box people to see how shabby they can look.

IMO being a race track writer is not the easiest job in the world, other sports have many facets, but in horse racing all you can write about is some dumb animal. I have a friend who writes for the local and is a stringer for several publications, many is the times I have walked in on him and he is just staring at his computer, got to come up with something. I know that if I had a job, writing about the race track would be the last on my list.

Karl

[Marc At DRF]

>Any halfway knowledgable person knows that biases for sprints and routes often differ

>But not to delineate between a sprint and route bias is just idiotic. So Dave is not a moron, but what he wrote in that instance certainly was.

So, he notes a May 25 speed bias on the main track. On other dates in Simo Weekly, he notes biases specific to just sprints, but on May 25, he just says "speed held well on the main track."
The implication is perhaps not as clear as it should be, but what's inferred is that on May 25 the speed bias was apparent in both routes and sprints.

"Idiotic"? Did somebody wake up on the wrong side of the bed today?

I often don't defend our guys because I often disagree with them.

But I think this is unduly critical of Litfin.

[JustRalph]

Hey Marc
Whats the joke around the office over at DRF. I am wondering what your guys were doing after the KY. Derby. Did you have to take a gun away from anybody? Just kidding.........

I love the fact that the only DRF person to pick War Emblem was the Web Master. Does he work out of the DRF offices? If so,
I bet he was having a ball.................

[Marc At DRF]

The internet department is really small; it's like 6 of us and only 4 who are horseplayers on any sort of regular basis. Anyway, 2 of the 4 (Webmaster and our Formulator "guru") absolutely crushed the Derby.

The internet department, we've been crowing ever since!

[andicap]

Yeah, Marc, I was feeling kinda feisty this morning -- probably catching what Boxcar has.

But I think it was irresponsible not to note that sprints and routes often carry different biases. Dave's supposed to be an expert and to know that. People are betting good money based on his observations and they should carry some sophistication.

[Rick]

At the risk of being declared a moron, I'll have to say that I've never really understood why a track bias could be so much different at routes than sprints. Now, I can understand why different distances with different lengths from the gate to the first turn should give different amounts of advantage to early speed horses, but I can't see why that would break down into two easily discernable patterns for sprints and routes. And, I can't fathom why longer races might favor closers when shorter races favor front runners, or vice versa. I would think that they would tend toward the same direction in the same day, albeit possibly to different degrees.

Now, I don't want any opinions from idiots or imbeciles, because they fall into lower IQ ranges than us morons. Only you really smart guys with IQ's of 70 or over.

[cj]

In my opinion, accurate determination of whether a bias existed or not takes watching the horses run back and see how they perform next out. Usually, it seems, by the time you can determine this it is too late to capitalize.

As for different distances having different biases, I have my doubts. The sheer extra distance of a route race is going to give closers a better chance. Every track I've ever studied shows statistically more winners on the front end in sprints than in routes.

Also, unless you have handicapped the races yourself, guessing whether a bias existed merely because longer priced entries won or hit the board is a dangerous business...remember Derby Day! Many proclaimed it a HUGE speed bias day, but already Snow Ridge and War Emblem have come back to equal there perfomances of that day, if not improve. It would be safe to look back now and say it was a fair track, but if you were backing closer types from that day in the interim, lookout bankroll!

Last thing...Andicap, Belmont is almost always favors early speed types, meet in and meet out. Also, there is NO TURN involved in the route races, so why would the bias be any different for routes and sprints?

CJ

[andicap]

CJ, I used to play track models. Only track models. This closely tracks short-term trends using Sartinesque figures. Often times, in fact usually, at Belmont and other tracks, the sprint model and the route model were different.
I don't know why. I used to think too that one turn routes should play the same as sprints, but they don't. Yes, Belmont generally favors speed, but c'mon horses going around the track Indian File like that -- that is unusual.
But you're right, we will see when the horses that couldn't close come back on a fairer track. If they still can't win (or finish a close 2nd), you might have a point. But what good is it for Litfin to even point out a bias of any sort if you can't exploit it later? If you are right then just ignore biases altogether.

[cj]

Andicap,

I pretty much do ignore them, but I'm not saying it can't be of value. Just a very tough art to master IMO. In addition to following subsequent performances, I feel you have to have handicapped the card in advance to get a good feel for things. Just because a horse is 20-1 and wins, maybe as a handicapper you knew he had a shot in the race.

To add to your original point, I subscribe to DRF Simulcast Weekly, and the larger tracks give a recap of each racing day. The days the "experts" note biases are usually a joke, Dave Litfin included.

CJ

[andicap]

Originally posted by cjmilkowski
Andicap,

To add to your original point, I subscribe to DRF Simulcast Weekly, and the larger tracks give a recap of each racing day. The days the "experts" note biases are usually a joke, Dave Litfin included.

CJ


Ah, we finally agree. If only India-Pakistan could find common ground so easily.

[Dick Schmidt]

Damn,

This seems to be "Open a can of Whup-Ass on everybody day."

Now you boys play nice.

Dick

[turfspec]

I think I can give at least two reasons why a "true bias" can be different for sprints vs routes. One is prevailing wind direction. Imagine, if you will, a 30 mph headwind gusting down the backstretch at BEL. Sprinters break from the gate relatively close to the turn, the speed therefore runs only a short distance into the wind and are then aided by a tailwind all the way down the home stretch. Routers run the entire, or nearly the entire, backstretch into the headwind. Closers tucked in behind, running under cover, do less work than the speed horses on the backstretch, swingout in the stretch where they are aided by the tailwind thereby improving their chances. Number two is attributable to the fact that virtually no track surface is perfectly uniform. Some sections may be gently uphill or downhill or deeper despite the best efforts of the maintenance staff, with varying drainage rates where moisture tends to accumulate. Again, wind direction may cause one part of the track to dry faster than another. If the affected part is run over by routers and not the sprinters then a "bias" may show up. Having said this I believe a " true bias" is much more rare than is generally assumed.

[Rick]

Every track I've looked at lately has an early speed bias (on dirt) for both sprints and routes. Someone please give me an example of a track where the bias is really different for sprints than routes over the entire meet or even for a week. I just don't see it happening for more than a day or two, and in that time period it's just as likely to be random variation. Just because I tossed heads 7 out of the last 10 times doesn't mean my coin is biased.

[andicap]

I guess we'll see what happens when the horses run back from last Saturday at Belmont. that's about the best clue -- although it's not an absolute answer -- of the prescence of a bias. But I will be looking for the closers to run better, unless of course the bias is still prevalent.

Of course the bias wasn't the only factor. Dutrow with a 39% win rate won the 9th rate wire to wire.

OK, OK, I take it back. Litfin's not an idiot, but I still think he was wrong. And will search to the ends of the Earth to prove my point -- somewhat like O.J. Simpson has been doing. :D

[delayjf]

If you get a chance to read "Bet with the Best", Steve Davidowitz does a good piece on track bias and how they can develop. (His chapter would be the only reason I would consider buying the book) Being a SoCal player I can tell you that DeLmar can develop a closers bias from time to time, Davidowitz gives what I believe is a reasonable explaination as to why the bias develops. Personally, I love closers bias because most handicappers are so tuned into speed, the prices are outstanding.

I would also agree with the previous post concerning wind. I was at Belmont Park for the Breeders Cup and witnessed it's affects. That flag in the infield was as stiff as a board almost the entire day, and it's a big flag.

Has anybody here ever attempted to profile the track maintenance and it's affect on the race results???

[Rick]

I'd be careful about the idea of a wind favoring closers. In Southern California at least, a good wind can dry out the track so much that all of the frontrunners win because it's so hard and dry. It's a very complicated subject.

[turfspec]

Rick, we need to define terms here. It seems you're using "bias" and "track profiles" interchangeably. As generally used, a track profile is usually a long term model of track tendencies, while a bias is usually a phenomena of shorter duration brought on by weather conditions or track maintenance. There can be a rather long term bias, but at what point does it become part of the profile? Nearly all N. American dirt tracks favor speed in both sprints and routes.

It's not the wind alone but rather the direction and force that either favor or disadvantage closers or speed for that matter. Delayjf is correct in citing Breeder's Cup day @BEL as an example of wind effect on bias.

You have to be careful in declaring a bias because you are right about a short term statistical anomaly. If 5 favorites win with a rail trip is that a bias? But, if 5 double digit longshots win on the rail, closer to the pace than is normal for them, while the favorites seem to struggle in the middle of the track, you may be onto something.

In addition to the reasons for a difference between sprints and routes I mentioned in my first post, run-up to the turn from the gate in routes can account for a difference.

Rob

[cj]

I still sumbit, IMO, that there was no bias on BC day. I know it was windy, and that pace times were slow, but exactly what horses were disadvantaged by the bias? What have they done subsequently?

CJ

[Rick]

turfspec,

Yeah, I'd agree that a short-term bias could be significantly different than the usual longer-term track profile, but it's so hard to pick up on it using only a few races. My point is that something like wind could go either way. It could be aiding the stretch runners by blowing in the same direction but it could also be helping the frontrunners at the same time by drying out the track. And then, in the middle of the card, the direction might change and mess up everything. Also, my impression is that track maintenance crews attempt to eliminate any significant bias that develops on a day-to-day basis. So, after you carefully identify a bias one day, overnight they may be creating the opposite bias the next day.

At any rate, when I studied this at Southern California tracks, I found almost zero correlation over the short term of a few days. The most significant results were for the entire current meet, but even there it takes about two weeks for a definite pattern to emerge. Using the previous meet or year was not a very wise thing to do since they frequently dig up the track in between meets. I'm starting to favor the tracks with the longest meets now for this reason.

If anyone has any solid data showing the effects of track bias, not in identifying it but in showing that it is useful in prediction, I'd really like to see it. I couldn't find it but that doesn't mean it isn't there. The really important thing to do would be to define the optimum period of time to use and it wouldn't necessarily need to be a fixed period. As you can guess, I'm scientifically oriented and therefore don't really like subjective answers to questions like this. People who "just have a feel" for things like that drive me crazy, probably because I'm jealous of their talent.

Somebody show me what I'm missing, please.

[superfecta]

the only problem once you decide there is a bias,either for early speed,or closers,is how do you quantify it?If the wind is blowing at thirty mph on the backstretch today but yesterday the rail was unusually heavy, how do you compare horses that ran on different days against different "biases" when they meet next week?
My inclination is you can't,so take their performance at face value.Maybe let the track varient rule.Doesn't always work,but at least you reduce the risk of superfluious info being introduced to your handicapping.
Also,I'm in the camp that says Litfin rode the short bus on more than one ocassion.

[andicap]

Well, handicapping is a lot of qualitative data, not just quanitative. I will be looking for horses that tried to close last Sunday at Belmont to come back on a track that plays fair. Can I put it into numbers -- of course not. But as Boxcar said, I'll let my grey matter process it.

The problem with tracking biases is it takes time to look over the charts every day, look at who should have run well and who shouldn't have, etc. If you are playing every day it's easier, but my damn day job keeps getting in the way. That and the family.

BTW, what ever happened to "No Day Job?"

[karlskorner]

I am not certain if other tracks do this, but CRC puts up on their web site each morning what was done to the track that day. I think this would effect some of the theories expressed above.
Example yesterdays report.

What is the condtion of todays track. Good
What was done to the main track today:

We got another inch of rain yesterday. The floats will be used to squeeze as much water off as possible and even it out. We'll follow with the packing harrows. No water will be needed before the race. The weather will decide if we need it later.

The words "floats" "squeeze" and "packing harrows" are operative words to me. I find it difficult to compare a horse who ran yesterday with a race he ran 2 weeks ago. To overcome this I print out each report and attach it to that days charts. When I look at a horse that ran 5/17 I compare what was done to the track that day against what was done to the track today. How many times the water truck went over the track and how the track was harrowed are also noted. I am ceratin the so called bias is effected.

Karl

[Rick]

It seems like I recall Today's Racing Digest in California maintaining information on previous track bias. If they're still doing that, it might provide a source of data to test some theories. Anyone know about this?

[karlskorner]

A thought came to me while sitting at the track today about bias, while watching the tractors clean up between races.

What happens to yesterday's so called bias when this morning the tractors were sent clock-wise instead of the usual counter-clock-wise pattern they follow, which often happens. Goodbye bias. What happens today between 2 of the races when the tractors are sent clock-wise, which I was watching. Goodbye so called early bias.

Did you ever watch an early speed horse you like come out of the 7f shute in the 1 hole and most of the time lose his momentum when he hits the main track, if he stays in the 1 path. Reason being when the track crews do their harrowing the surface tends to pile to the right. Watch the front view re-run of any sprint race coming out of the shute, the jock, it he is able, will try desperately to bear to his right Track crews can be your best friend or worst enemy.

Karl

[Rick]

Karl,

I'm afraid that the truth is that track maintenance procedures have improved to the point where they can eliminate practically any bias if they want to. And track management seems to want to in most places. Where have all the cheap early speed tracks gone?

[karlskorner]

Rick;

Not to offend certain members of this Board, I also look with some doubt on "pars". Regardless how par times may be calculated or applied, they remain approximations unless supplemented with knowledge of the "actual track conditions" under which a particular running time was recorded. In other words what did the track crew do to the track that day.

Karl

[GameTheory]

Well, isn't that a major function of par times? To be able to equalize times to what they would be on a "normal" (average) track. You can't make a variant without knowing what normal is (or without making an assumption about what normal is, anyway.) Pars are useful because they are made from samples taken from many different days to cancel out effects of the varying track.

Something people seem to say a lot about these types of issues in general where measurements are not exact or unknown is, "How can you use this? How can you trust that? You must adjust for this, etc." The thing is, even if you can never know what the true situation is precisely, it is usually quite easy to find out what WORKS BETTER. Can you predict more winners after making X adjustment to Y factor? If so, it is useful, whether or not it is truly accurate.

[andicap]

When Brohamer made pars, he was always careful to factor in the average DRF track variant, at least he wrote in his book. (something tells me in real life he was a bit more precise).
Someone asked him at the Handicapping Expo years back about Gordon's par times and he said they were OK, but needed an average track variant in order to compare them by.
I.E. If the average 10K claimer at Monmouth goes in 1:11 (hypothetically), and the average variant is 13, and you're comparing it to a 10K race at Calder where the par might be 1:12 and the average variant is 16, then you would incorporate those adjustments. A horse shipping from CRC to MTH with a 1:12 time on a track that had an 18 variant would presume to be 1 tick faster than the Monmouth par. (I recall Brohamer adjusted 1 tick for two DRF pts, but I could be wrong.)

OK, have I explained this right?

I always thought this was a load of hooey since the DRF variant are a) imprecise and b) wrong. But I presume Brohamer knew most people wouldn't do the work he does so he needed a workable shortcut for the masses.

[Handle]

I agree with Andy -- the real problem with tracking biases is the same one you see in Beyer speed figures. People here have mentioned that Beyers are "subjective". What's subjective about them? Its that, and this is according to Davidowitz, the "best" way to come up with them is to find several "consistent" horses -- ones you can count on to run their race -- and see how they run.
Davidowitz says,

"No longer will you always seek to compare clockings to par times for the class. Instead you will examine a complete day's worth of races trying to find a cluster or two of relatively stable horses who appear to have performed as anticipated."

There's an awful lot to buy into there. These stable horses could be in the same race (D. says thats good), run 1-2-3, but turn in a pretty poor -- or pretty exceptional -- effort. That is, unless there really is such a thing as a "stable" horse (that's buy-in one). The time taken from their race is then applied against all of the other "inconsistent" horses that day. Same goes for finding biases using that method, I guess.

What I want is for each track to run some mechanical creature across the surface before each race. I then want the time for that mechanical creature. The mechanical creature should have 4 legs and exert jerky, springy force to move itself -- but in a constant fashion to be measurable. In a controlled environment, the time it takes for this creature to move 4 furlongs should not differ more than .01 seconds for each run. And I don't want anyone else to have access to this information <g>. Have to go talk to some friends at NASA.

-N

[karlskorner]

Now we come to "variants" another of my favorites.

Since Brohamer was quoted lets use his definition.

Procedure
1. Determine the "approximate" par for each race.
2. "Compare" performance of each race to par
3. Evaluate "apparent" inconsistencies
4. Assign the daily variants for "sprints" and "route" usually by "averaging" the difference between "par" and actual performance. Emphasis in order of performance should be placed on (a) older male claiming races
(b) older female claiming races.

The key words in this procedure: APPROXIMATE, COMPARE, APPARENT, AVERAGING and of course "par". These words are vague and certainly not positive.

To quote the 2nd paragraph on his chapter on variants.
"A daily variant is what enables pace and speed handicappers to fully appreciate good performance and "downgrade" those under too-favorable conditions. Weather and 'TRACK MAINENANCE" activity "contribute heavily" to fluctuating track speed. BINGO. Just what I have been talking about.

Karl

[Tom]

Beyer numbers are subjective because he uses the projection method and that involves personal observation. The DRF SR and TV are objective becase they are calculated mechanically. Let me say, that after years of making variants (since the 70's) that the only reality in racing times is what you see in the PP's (baring transpositional errors, of course). Everything else is not reality - it is what we decide reality is. The better our guesses, the better our numbers. As Karl pointed out, they can do strange things to a track during a card and distort an already small sample of races.
What I do, for only one or two tracks becasue it is a big job, is to make my variants over time. Say FL runs Fri-Tues and the track condition/weather stays relatively constant. I try to make a variant that accounts for all 5 days, if possible. Maybe a couple of route races will be abberant, but the sprints appear to be the same throughout - I will then tend to think that the abberations were not due to track condition, but maybe pace, or clocking errors, or something else that I cannot mesure. Another trick I use is to add the pace and speed figure together and then see how the total compares. A recent example, ripped right out of real life, was in pace variants. I had the following deviations for sprint races: F4 F3 F4 F3 F5 F4 F4 F13 F4 F3.
What to do about the F13??? If I average it, it will distort the other races. When I look at pace/speed totals, however, it is clear that the F13 was earned at the expense of final time - it was the slowest race of the sample. Using totals, the track was consistant. Sometines you just can't explain some races. This is good, too, becasue now you know what races not to use in your handicapping - find another race to use, and if there is only one race that makes a horse a contender, maybe he isn't after all.

[Rick]

The thing that bothers me about all of these procedures for calculating variants is that there isn't (as far as I know) any real scientific evidence that one method (or none) is better than another. We just don't know what really works and what doesn't. As is so often true in horse racing, we're just supposed to trust that some guru knows the answer without ever testing it.

[PaceAdvantage]

It shouldn't be too hard to test. Those that make numbers have a choice. Use adjusted figures, or unadjusted figures. If the numbers are computerized, it can't be difficult to make one set with variants, and one set without. Test it out for a few weeks, and see which win more races. I know for a fact that I've read on this message board of someone who did just that. I seem to remember that in the end, it didn't make much difference either way.....:rolleyes:


==PA


PS. If memory serves correct, the unadjusted figures won less races, but more money.....

[BillW]

Originally posted by Rick
The thing that bothers me about all of these procedures for calculating variants is that there isn't (as far as I know) any real scientific evidence that one method (or none) is better than another. We just don't know what really works and what doesn't. As is so often true in horse racing, we're just supposed to trust that some guru knows the answer without ever testing it.

Rick,

You bring up a great point. I see a lot of opinions about what is and what isn't good, but very little backing facts. I develop my own software and spend a great deal of my time on analysis. It does provide a lot of interesting answers. One major lesson that I have derived from that exercise is that "close is usually good enough".

Bill

[alyingthief]

that dog, that barking wolfhound, that snapping beast sartin may again prove correct: the variant is a reflection of the running of the race, the interaction of the animals, and not much track surface.

i maintain a model of the energy expenditures, by fraction, of the track i am playing, and i have found some very odd things. for instance, one year at SA's winter meet, all the sprints for older horses were won by closers, even the 7F, with a given energy figure. the next year, all of the races were won by forwardly placed horses with, oddly, the same energy patterns and figures. which would seem to indicate that percieved bias, or running style bias, even of the long term variety, may vary, but be meaningless, because it's the ability and inheritance of the animals themselves--some speed stays, some doesn't, because the genetic material happens that way. breeding sheds employ sophisticated genetic modeling to produce runners of a certain kind, and they are as prone to fashion and crazes in selecting inheritances as the bettors are methodologies. sometimes i wonder if the various popular methodologies, like speed or pace, really have the universal validity we attribute to them: speed, like dowst's consistency model, produced strong returns so long as the public was oblivious to it, as did pace. all one does when selecting speed or connections or class or form is adopt a methodology that is either one prevalent and popular, and a guaranteed loser, or ignored and perhaps lucrative. the long term statistic of winning favorites would indicate this to be so--it's just that the favorite now has a big speed figure, whereas formerly the favorite had some consistency number that stood out.

maybe as handicappers the quest is not to find the best number, but to develop numbers that reflect the crowd's inattention. a number, or handicapping method for form/condition or for class, whatever, so long as it's unpopular. and i don't think the number and method you use for these things can be what worked in some other era: racing changes, and it impacts the models we maintain, and the methods we use. maybe class as a method began to break down with the advent of year round racing, and escalating injury due to over racing became commonplace, compromising a horse's inherent class. a new class model would need to examine this kind of thing.

i have a hard time placing much reliance in variants because everytime i have set down to create par times myself, the times of races hardly shake out in these neat one second incremented class hierarchies that once they did--or quirin has said they did. take the 16k class in SC: when you realize that the (older) horses that generally win these races almost always have careers that include victories in much higher class brackets, and have fallen on physically hard times--it isn't so difficult to understand why you see these incredible times from them. 108 and change, for instance: if you create a variant on the basis of this running time, you may not have the slightest indication of surface speed. i have sat down with the drf variant as my ruler, and looked at the times that classes have run. weird, but 10k claimers average a difference of 17 ticks from the 3yr record, and 50k, 16 ticks. etc. where do you put this datum?

does anyone have a variant that excludes these problems?
and the 10k claimer hypothesis is long since compromised by inflation: track to track adjustments using this as a baseline has to be suspect. seems to me, the 20-25k horses do this now. anybody have ideas on this one?

[Rick]

alyingthief,

I've been trying to develop the kind of rating you suggest for a long time and it's difficult. Although the method I'm using now does well enough, I'm afraid that in a few years I'll have to change factors I use due to the changing popularity of different approaches. I'd really like to have some self-adjusting approach. One approach I tried was comparing average finish positions compared with the average finishing position predicted by a model based on odds alone. Although it works somewhat, it's not as good as I had hoped, probably because predicting winners is different than predicting finish position. But if you use winners only, you need an unreasonably large database. The best procedure I've found so far for predicting the effectiveness of a method based on a small sample involves using win and place payoffs for the top two rated horses. It's very frustrating to deal with such noisy data.

[andicap]

Originally posted by alyingthief

i maintain a model of the energy expenditures, by fraction, of the track i am playing, and i have found some very odd things. for instance, one year at SA's winter meet, all the sprints for older horses were won by closers, even the 7F, with a given energy figure. the next year, all of the races were won by forwardly placed horses with, oddly, the same energy patterns and figures.

That's interesting, because when I was handicapping with track models (as opposed to track profiles), 10 years ago, I found the same thing. It was irrevelant how many lengths from the lead the winners came from.
Then five years ago, using Thorovision, there was an incredible stat called "DA," which calcuated a horse's energy relative to the other horses in the field. I would model this and many times would find -- over short periods of time, but sometimes longer -- a narrow range that winners would fall into. Horses that seemed like marginal contenders to me had one factor in their favor -- their DA fell into the model. I'd give my left toenail to be able to calcuate that DA today. I don't know the formula that Michael Perry and Bill Burns used.

maybe as handicappers the quest is not to find the best number, but to develop numbers that reflect the crowd's inattention. a number, or handicapping method for form/condition or for class [/QUOTE]

Check out Mark Cramer's "Kinky Handicapping" and his chapter on class. He agrees with you 100%.

[Doug]

I remember way back when Sartin produced a very crude program where the user had to input everything.

In the manual there was a way to use class as a variant, Sartins words "A variant is a variant is a variant".

Doug

[alyingthief]

you know, sartin really was a brilliant s.o.b.

and rick, i absolutely concur on the lack of scientific verification.

on another front: what i would like to know is how those betting cartels do it in hong kong? there was a really great article about them i came across recently...it appears they make a lot of use of ziemba's advanced material in efficiency of race track markets. they kill the races over there. anybody have this book for sale? anybody know anything about this?

[alyingthief]

it's funny you should mention that, andy: i once thought i'd found the holy grail when a model for sprints hit every race i played at SA winter, i made a killing--over 16 dollar average mutuel betting two horses, at something like 67% guess what? i implemented the strategy at oaktree, and had like 18 losers out of 20.

can you get in touch with those guys and find out what formula they used? im sure they sell it, if you inquired.

[Rick]

alyingtheif,

The book you want to get is "Efficiency of Racetrack Betting Markets", edited by Hausch, Lo, and Ziemba (Dr. Z). It was published in 1994 but reprinted later and was available within the last couple of years from Gambler's Book Club in Las Vegas. Warning: It's not cheap, maybe about $90, it may not be available anymore, and the whole thing has a lot of advanced mathematics. If you're still interested, get it and read "Computer Based Horse Race Handicapping and Wagering Systems: A Report" by William Benter. Then read the two previous papers by Bolton and Chapman. This is really good stuff if you understand it. I'm not being elitist here (is that the right way to spell it?). I really don't know how much math other people understand, but if you don't understand, I can translate it into simpler terms. I don't really understand things at an abstract level but people tell me I'm able to translate them into an understandable explanation. I don't know if they're right or not, I just know what they say.

[tanda]

I would suggest that anybody who determines a bias based on two or three races read "Fooled By Randomness" which was posted and discussed elsewhere on this board.

It discusses the error that many people make, particularly the media, in presuming that two or three events are non-random and grossly underestimating the possibility that they are random occurrences.

A truly neutral track will have races won in a similar style due to randomness. Just as a fair coin may flip heads 5 times in a row. To assume that two front-running victories, even by unlikely horses, makes a bias is akin to assuming that two consecutive heads makes an unfair coin.

[alyingthief]

rick, thanks. my math is fairly good, if i sometimes mistake a + for a -; which happened once when i thought my ROI indicated a series of bets at 75% of bankroll.

just kidding.

[andicap]

Originally posted by alyingthief
it's funny you should mention that, andy: i once thought i'd found the holy grail when a model for sprints hit every race i played at SA winter, i made a killing--over 16 dollar average mutuel betting two horses, at something like 67% guess what? i implemented the strategy at oaktree, and had like 18 losers out of 20.

can you get in touch with those guys and find out what formula they used? im sure they sell it, if you inquired.

Unfortunately Michael Perry has passed away. Bill Burns is still around and someone gave me his number but I lost it. I'll see if I can get his email address. He's an old Sartin guy too. Was mentioned in the acknowledgement for PMTR -- I actually wrote the manual for that software, had nothing to do with its creation. A great, great, piece of programming on Perry's part. Amazingly easy to use with dozens of handicapping factors all one keystroke away.