Horse's money box

[imofe]

I wanted to get some thoughts on what I consider to be the weakest part of my handicapping. The horse's record of wins, places, and shows for the current year or last year if the year is early on. The reason I consider it weak is because I really pay little or no attention to it. Current form and what class the horse is entered in are much more important to me. The other day a guy at the track told me he would prefer a horse that was 20 2 0 0 as opposed to one that was 20 2 1 2. Is running second or third a negative sign? Does winning percentage really matter that much? If a horse just got his first win of the year in 20 starts does he rate worse in his next race than if he had won one or two times earlier in the year at the same class? Just thought I could get some feedback on how to better use the money box in my handicapping.

[Tom]

I look, generally, for "secoditis."

15 2 6 7
20 1 9 8

The horse gets close but fails to seal the deal. good for exotics, though.

I also look for the record at today's track.

Lifetime 32 2 2 3 Finger Lakes 5 2 1 2

[thaskalos]

IMO...the horse's money box has drained more horseplayer wallets than any other stat on the past-performance page. There is nothing necessarily positive about a 2-0-0 record in 20 starts...nor is there anything negative about a 2-1-2 record, respectively. These are just silly superstitions which are responsible for keeping those who believe in them forever bound by the chains of ignorance and immaturity.

What is the horse capable of...and can he run to that capability today; these are the major questions confronting the horseplayer as he plies his trade, as far as I am concered. Yearly and lifetime records and earnings are meaningless, IMO...because they confuse us about the horses' current state.

[imofe]

I look, generally, for "secoditis."

15 2 6 7
20 1 9 8

The horse gets close but fails to seal the deal. good for exotics, though.

I also look for the record at today's track.

Lifetime 32 2 2 3 Finger Lakes 5 2 1 2

Okay. But let's say the 20 1 9 8 just won his last race in what you consider a dominant performance. He now steps up a little and goes off at 5-1 in a race where you think the favorites are vulnerable for whatever reason. Will this money box get you off of that horse?

[Overlay]

Yearly and lifetime records and earnings are meaningless, IMO...because they confuse us about the horses' current state.
I think that average earnings-per-start can still be a useful rating tool with regard to both class and consistency, as long as it's employed in combination with other form-related metrics that indicate whether the horse is in condition to run to its potential in today's race.

[jerry-g]

There are so many ways to slice this cake, it seems. I ran across an authors idea of Average Purse Value using the Money Box. I will try to recreate it here:

How to calculate the Average Purse: HORSE TOTAL RACES WINS PLACE SHOW

Aces Limit 24 2 6 4

TOTAL WINNINGS: $30,000




Normal Percent

Share of any Purse Number of Purses

2 Wins x .60 = 1.2

6 Place x .20 = 1.2

4 Show x .10 = 0.4_

TOTAL 2.8_

$30,000 Total Winnings divided by 2.8 = $10,714 average purse.

I'm not sure what the value, if any, of this average purse per horse,

would be in handicapping a race.

[jerry-g]

I should mention that the Author for the Average Purse was
David L Christopher in "Winning at the Track."

[thaskalos]

I should mention that the Author for the Average Purse was
David L Christopher in "Winning at the Track."

This calculation predates David L. Christopher by a wide margin. And it has very little handicapping value on its own...since it considers a 3rd-place finish to be equal to a win.

[Johnny V]

If a horse has a high percentage of wins it must be looked at as "who did he beat?" If he has been beating lesser and is in today with a horse who has a much lower win % but has some decent in the money finishes against much better then that % stat is not so meaningful.
I do not use EPS or APV in any calculations any longer in my handicapping. If you desire to use them however, and they do have some predictive value IMO, I would suggest dividing the EPS, for example, by todays purse to get an "index" number so to speak to compare with the rest of the field on how they stack up with todays purse.

[classhandicapper]

The problem with the earnings box is that the purses themselves are less efficient than they used to be because of state bred races and the horse's win record is also a function of the class in which he's been competing. It also tells you little about current form.

All that said, I do think "consistency" is a positive attribute and exceptional winning records are sometimes an indication that a horse that may have some extra racing reserves that are not apparent in its figures or winning margins. It's a lot easier to suck up for 2nd or 3rd with an easy trip, or make a bid and hang, than it is to win.

[Overlay]

The problem with the earnings box is that the purses themselves are less efficient than they used to be because of state bred races and the horse's win record is also a function of the class in which he's been competing. It also tells you little about current form.
In the last edition that Ainslie put out of his "Complete Guide to Thoroughbred Racing" in the late 1980's (when inflated purses for state-bred races had increased significantly in number), he made the categorical statement that anyone who continued to rely on earnings for class determination was "positively doomed" in venues where state-bred races were programmed.

However, statistical studies that I have seen subsequent to that (even up to the present day) have indicated that average earnings-per-start still retains its power and effectiveness (as far as impact-value range and progression) in ranking a field from top to bottom, even where state-bred races are run.

[DJofSD]

I should mention that the Author for the Average Purse was
David L Christopher in "Winning at the Track."
As best as I can tell, the initial publication of "Winning at the Track" was 1991. PIRCO was using APV long before that.

[jerry-g]

As best as I can tell, the initial publication of "Winning at the Track" was 1991. PIRCO was using APV long before that.

Sorry, I should have phrased it that the author of "Winning at the Track",
mentioned it in his book. Didn't mean to imply I knew who devised the
method.

[pondman]

In today's world maiden winnner's win Grade 1 races, and then retire after 8 races. I check the last couple purses by reading the last few chart, before I'll bet a large single.

[jerry-g]

For class, I generally just look at the level the horse has won at or finished
2nd within one length. Then rank them against today's class and the rest
of the field.

[DJofSD]

One of the problems with APV, as devised way back when, is it was based upon a purse distribution which has changed. The weights used for the old formula were used for an in-the-money purse distribution. Purse money is now distribute to 4th and 5th place finishers, if not even further down the order of finish list.

[Robert Goren]

In certain kinds of races, the number of wins can be important. If you find a race such as NW of 3 races or 3 year olds. A 3 year old with 3 or 4 wins is worth an extra look. NYRA has gone to writing races with conditions such as NW of 3 races or NW of a race in 4 months. I also look to see if the horse wins more on the turf or on an off track than on a fast dirt one.

[Tom]

I posted a selection for a turf race at Toga this year - the winner had 7 turf wins and was facing horses who were all nw2-nw3 on turf, something like that. I'll search for that post and find out who the horse was. He was not the favorite.

[Dan Montilion]

Agreed that amount of wins in conditioned races are of value. I also glance at the money/lifetime box for situations where the amount of wins are under exact conditions. Say 7 life wins 6 at distance and 5 of those are at today's track. Hardly an end all but forces one to look deeper into the PP's.

[imofe]

The problem with the earnings box is that the purses themselves are less efficient than they used to be because of state bred races and the horse's win record is also a function of the class in which he's been competing. It also tells you little about current form.

All that said, I do think "consistency" is a positive attribute and exceptional winning records are sometimes an indication that a horse that may have some extra racing reserves that are not apparent in its figures or winning margins. It's a lot easier to suck up for 2nd or 3rd with an easy trip, or make a bid and hang, than it is to win.

I guess this is where I have the problem. What if the races the horse ran 2nd or 3rd are ones where he battled the lead in fast fractions. The horse or horses he battled against were not around at the end. A closer eventually won those races.

[classhandicapper]

I guess this is where I have the problem. What if the races the horse ran 2nd or 3rd are ones where he battled the lead in fast fractions. The horse or horses he battled against were not around at the end. A closer eventually won those races.

I think that veers off the earnings box and into trips.

I don't keep trip notes for every horse, but I do notice the extreme pace scenarios and make a bias note for every day in NY. I also watch most of the major stakes races.

If you know a horse has a few races that are better than they look on paper then you should upgrade it (and vice versa). It's hard to put it into an exact formula. I can't. I use intuition/feel.

[classhandicapper]

However, statistical studies that I have seen subsequent to that (even up to the present day) have indicated that average earnings-per-start still retains its power and effectiveness (as far as impact-value range and progression) in ranking a field from top to bottom, even where state-bred races are run.

I saw Dave Schwartz's data on that.

I think it probably works better in races where state bred and open horses don't meet each other or if you used some kind of state bred adjustment.

I have often wondered if there's a way to adjust state bred purses in NY to better classify the horses, but I don't have the data and I guess I've been too lazy to try to figure it out.

[thaskalos]

I think that veers off the earnings box and into trips.

I don't keep trip notes for every horse, but I do notice the extreme pace scenarios and make a bias note for every day in NY. I also watch most of the major stakes races.

If you know a horse has a few races that are better than they look on paper then you should upgrade it (and vice versa). It's hard to put it into an exact formula. I can't. I use intuition/feel.

Classhandicapper, forgive me if my question is inappropriate...but I am intrigued by your posts, and will ask anyway:

You are obviously a very knowledgeable handicapper who puts a lot of time and effort in this game...as your above post clearly indicates. And yet, you revealed in another thread that you wagered less money this past year than you did when you were a broke 17-year old.

It always fascinates me when I meet very knowledgeable players who do not convert that knowledge into confidence at the betting windows...

[Dave Schwartz]

APV is not (and probably never has been) a very good predictor.

However, Time-decayed APV is pretty good. (Using Benter's formula for time decay - 120 day half life.)

[mistergee]

I should mention that the Author for the Average Purse was
David L Christopher in "Winning at the Track."

wow Jerry, havent heard this book mentioned in a while but i did own it and havent seen it in years. The book also had something called the "normative time table" that could be used to adjust the speed ratings for different tracks and distances. Do you happen to know that info. Thanks in advance

[JohnGalt1]

If it had value in the past, it is practically worthless today with tracks offering similar quality of racing having discrepancies in purse sizes.

Right after Prairie Meadows got a casino their $5 claimers had about twice the purse size as Canterbury Park with the same quality of horses. A horse with a lot of in the money finishes in PrM $5k races shipping to CBY for a $5k race would have a higher average earning per race.

Or tracks like Monmouth that had the short meet with huge purses. Is a horse that won or finished second in their $90k Mdn Sp Wgt races more classy than a horse who won or finished second in Gulfstream's $47k maiden races?

If you averaged the purses in the two above examples you would say yes.

Racinos has skewed the earnings box records IMO.

[jerry-g]

wow Jerry, havent heard this book mentioned in a while but i did own it and havent seen it in years. The book also had something called the "normative time table" that could be used to adjust the speed ratings for different tracks and distances. Do you happen to know that info. Thanks in advance

What I have is a free E Book I downloaded from the web. In it he discusses
the adjusted speed ratings and shows a table but suggests since the book
is now out of print, that we check our local library for it. He also mentions
that in the future an EBook version of the book will be published but that
it was not a priority at this time. The E book I have is called Horse Racing
Secrets. He shows seven basic steps to the "shortcut" version to illustrate how the P/M Table should be completed manually.
I'm not sure where I got the E book. Perhaps google might help.

Here is the LINK (http://www.e-bookspdf.org/view/aHR0cDovL3d3dy5saWJlcnR5cHViLmNvbS9pbWFnZXMvSVNCTi 0yNjMtZnJlZS5wZGY=/SG9yc2UgUmFjaW5nIFNlY3JldHM=)
Click on the download link and save it as PDF.

[classhandicapper]

Classhandicapper, forgive me if my question is inappropriate...but I am intrigued by your posts, and will ask anyway:

You are obviously a very knowledgeable handicapper who puts a lot of time and effort in this game...as your above post clearly indicates. And yet, you revealed in another thread that you wagered less money this past year than you did when you were a broke 17-year old.

It always fascinates me when I meet very knowledgeable players who do not convert that knowledge into confidence at the betting windows...

100% honest answer.

I go through periods where my game changes from being more "class" oriented to more "speed and pace figure" oriented. I'm also always trying to blend the strengths of each into a kind of more unified approach.

It's a problem similar to using several different sets of speed figures that often disagree. You'd stop being confident enough in any of them to pull the trigger when you one set says you have an overlay and the other set says you don't. Paralysis by analysis?

When I was young, I was crazy enough to think I always knew what I was doing. Now I'm smart enough to know I'm often not sure. :lol:

Last year I was particularly paralyzed even though I netted a profit on my smallest handle in decades.

[classhandicapper]

Racinos has skewed the earnings box records IMO.

Yep. That's the other problem with classifying using purses.

It really is very difficult to classify horses properly at a single circuit, let alone across circuits.

[Sapio]

APV is not (and probably never has been) a very good predictor.

However, Time-decayed APV is pretty good. (Using Benter's formula for time decay - 120 day half life.)

Time-decayed APV is better than non-Timed-dayed APV doesn't add much information. Replace APV without another time dependent factor and the same is true. Not relevant, unless it was ignored in the betting... using Delta Lover's Chi -Square (where expected is odds based) test or similar.

Thomas Sapio

[Dave Schwartz]

Sapio,

LOL - Well, let me say it differently. APV lacks very much punch.

It simply is not a very good factor in most kinds of races and the ones it does reasonably well in, are not the ones the public thinks.

But to be clear, many factors do not improve with time-decay. Examples are things like average gain (or loss) in the stretch.

APV with a time-decay component does have more value as a handicapping factor but it must be computed (therefore) on a race-by-race basis. Unless one gets that info in a download, it is not worth the effort to compute.

[DJofSD]

Dave, I am not familiar with the time decay version. From the use of the phrase 'time decay' is it safe to infer it is just another way of measuring recency?

[Robert Goren]

My brother had a side rule that used average purse values to pick horses around 1970.

[imofe]

By the responses I am going to assume there is nothing much to a horse one way or another with a few more place or show results in his box. But how about an extreme? How about a horse with no wins for the year or for last year if it is early in the year? Is a horse with 20 0 2 3 a throwout racing against horses that show a decent winning percentage? What if the horse's last few starts show the best current form in the race and the price is there?

[Overlay]

By the responses I am going to assume there is nothing much to a horse one way or another with a few more place or show results in his box. But how about an extreme? How about a horse with no wins for the year or for last year if it is early in the year? Is a horse with 20 0 2 3 a throwout racing against horses that show a decent winning percentage? What if the horse's last few starts show the best current form in the race and the price is there?
That 0-for-20 win statistic is a fairly large impediment to overcome, especially against other horses that have shown that they can win and/or finish in the money more consistently, and even if the price is there (which might be unlikely if -- as you say -- the horse shows the best current form in the race).

I would want to try to find any reasons for the horse's lack of wins that either do not apply in today's race, or else appear to have somehow been addressed by the horse's connections. If I can, the horse might still not be the one that I rate as likeliest to win, but might warrant some degree of support if it is overlaid, commensurate with the disparity between what I judge its actual chance of winning to be (based on how it compares to the rest of the field in all my handicapping areas -- not just consistency) versus the odds at which it is going off.

[thaskalos]

By the responses I am going to assume there is nothing much to a horse one way or another with a few more place or show results in his box. But how about an extreme? How about a horse with no wins for the year or for last year if it is early in the year? Is a horse with 20 0 2 3 a throwout racing against horses that show a decent winning percentage? What if the horse's last few starts show the best current form in the race and the price is there?

Consistency is over-bet, IMO...unless the consistent horse is outclassed in today's race. The money is in anticipating the improvement of seemingly inconsistent horses...who might be sitting on a big race today.

If I like what I see in the horse's recent form...then I won't hold his prior futility against him.

[Robert Fischer]

Look at the money box, and if you see a red-flag, then verify that with current form and thorough handicapping.

[Dave Schwartz]

Dave, I am not familiar with the time decay version. From the use of the phrase 'time decay' is it safe to infer it is just another way of measuring recency?

Recency in ALL the races.

Just consider a list of values that you want to average, like speed ratings or purses. The farther back the rating is, the less value it has. Eventually, it reaches a point of absolute diminished value, and devalues no further.

Thus a speed rating from 10 days ago might have a multiplier that makes it 20% more important than a speed rating from 40 days ago.

[DJofSD]

OK, thanks.

But to drill down a bit more. What if the race 40 days ago was a route and well intended while the most recent is a sprint. Today's race is another route.

Is a situation like that something a decaying APV is capable of catching?

[classhandicapper]

I once did a controlled study where each horse had run under the same conditions as today (within 1/2 furlong, same surface, no off tracks, no more than a minor class change, no layoffs mixed in, no trainer changes, no terrible trips, no lightly raced improving horses, fairly consistent performance in the past) over his last 3 races.

I then had some math guys do some regression analysis on the best way to weight those prior 3 speed figures to predict his speed figure today.

The weights clearly favored the most recent races over the older ones and started losing usefulness as soon as you started going back 3 races. Now granted, there may be times we have to go back further for relevant information, but there is significant decay as you go back.

A study I would like to do is related to some that are already out there (last figure, best of last 2, best of last 3 etc...)

I think best of last 2 or 3 performs well partly because the horses that rate on top off their next to last or even an older race, also rate fairly well off the subsequent races. On other occasions there were trips or other excuses for their last race and you would have thrown the race out anyway.

What if you say best of last 2, but the last race was a bad race without excuse?

What if you say best of last 3, but the last 2 races were bad without excuse?

What if say only go back past the last race if the subsequent races were within "x points" of the back race?

[mistergee]

What I have is a free E Book I downloaded from the web. In it he discusses
the adjusted speed ratings and shows a table but suggests since the book
is now out of print, that we check our local library for it. He also mentions
that in the future an EBook version of the book will be published but that
it was not a priority at this time. The E book I have is called Horse Racing
Secrets. He shows seven basic steps to the "shortcut" version to illustrate how the P/M Table should be completed manually.
I'm not sure where I got the E book. Perhaps google might help.

Here is the LINK (http://www.e-bookspdf.org/view/aHR0cDovL3d3dy5saWJlcnR5cHViLmNvbS9pbWFnZXMvSVNCTi 0yNjMtZnJlZS5wZGY=/SG9yc2UgUmFjaW5nIFNlY3JldHM=)
Click on the download link and save it as PDF.
thanks Jerry that is what I was lookinfg for in this he calls it the Basis Tmes Table. In what I had from him which was more like a rating system he called it the Normative Time Table

[Cratos]

Recency in ALL the races.

Just consider a list of values that you want to average, like speed ratings or purses. The farther back the rating is, the less value it has. Eventually, it reaches a point of absolute diminished value, and devalues no further.

Thus a speed rating from 10 days ago might have a multiplier that makes it 20% more important than a speed rating from 40 days ago.


Aren’t you really talking about the slope of the horse’s performance curve with respect to the time of its performance? IMHO “time decay” is another one of those made up terms in horseracing.

[Cratos]

I once did a controlled study where each horse had run under the same conditions as today (within 1/2 furlong, same surface, no off tracks, no more than a minor class change, no layoffs mixed in, no trainer changes, no terrible trips, no lightly raced improving horses, fairly consistent performance in the past) over his last 3 races.

I then had some math guys do some regression analysis on the best way to weight those prior 3 speed figures to predict his speed figure today.

The weights clearly favored the most recent races over the older ones and started losing usefulness as soon as you started going back 3 races. Now granted, there may be times we have to go back further for relevant information, but there is significant decay as you go back.

A study I would like to do is related to some that are already out there (last figure, best of last 2, best of last 3 etc...)

I think best of last 2 or 3 performs well partly because the horses that rate on top off their next to last or even an older race, also rate fairly well off the subsequent races. On other occasions there were trips or other excuses for their last race and you would have thrown the race out anyway.

What if you say best of last 2, but the last race was a bad race without excuse?

What if you say best of last 3, but the last 2 races were bad without excuse?

What if say only go back past the last race if the subsequent races were within "x points" of the back race?

You did a controlled study? Please enlighten us on whether you used control charts, DOE, ANOVA, or some other statistical tools to monitor your study.

Also why did you go to “math guys to do some regression analysis” for help; “stat” guys would have been better equipped.

Furthermore, what was the typical value of the dependent variable in your regression analysis and how did it change when any one of the independent variables varied?

I am just asking to try and understand what was done, but if it is proprietary information I will understand.

[classhandicapper]

You did a controlled study? Please enlighten us on whether you used control charts, DOE, ANOVA, or some other statistical tools to monitor your study.

Also why did you go to “math guys to do some regression analysis” for help; “stat” guys would have been better equipped.

Furthermore, what was the typical value of the dependent variable in your regression analysis and how did it change when any one of the independent variables varied?

I am just asking to try and understand what was done, but if it is proprietary information I will understand.

I also bet on basketball.

Over the years I accumulated some acquaintances in the NBA advanced stats community. It's a community that has produced a few guys that have gone on to do advanced stats work for NBA teams. Others will probably follow as the movement picks up steam in the league. I told them what I was was trying to accomplish and learn, I collected the data for them, and then one of the guys did the study for me and gave me his observations as a favor. I don't have the stats background to do a study like that. Sometimes it's who you know.

[classhandicapper]

Aren’t you really talking about the slope of the horse’s performance curve with respect to the time of its performance? IMHO “time decay” is another one of those made up terms in horseracing.

I think he's making the general observation that the older the information, the less significant it becomes. Many statistical studies that have looked at multiple races within a horse's PPs (for consistency, class, earnings, speed figures etc..) have tended to weight each race equally or count one that occurred 10 days ago as equal to one that occurred 90 days ago. He's simply suggesting that if you count the more recent ones more heavily, it will sometimes produce better results. That's what the figure study I had done suggested.

[MJC922]

By the responses I am going to assume there is nothing much to a horse one way or another with a few more place or show results in his box. But how about an extreme? How about a horse with no wins for the year or for last year if it is early in the year? Is a horse with 20 0 2 3 a throwout racing against horses that show a decent winning percentage? What if the horse's last few starts show the best current form in the race and the price is there?

Mark me down as one who looks at the lifetime record often. If you have observed horses like Home Plot lose race after race when they absolutely figured so many times... it's really tough to ignore the fact that some horses will only pass that last horse by sheer accident. The record by itself is only an indicator, I mean you should handicap a race, reach the conclusion of how can this horse possibly lose in here and then watch it hang the entire stretch through a slow final quarter. To various degrees, this is not uncommon behavior and you have an edge if you know who the severe hangers really are as they get bet down quite a bit.

[Dave Schwartz]

But to drill down a bit more. What if the race 40 days ago was a route and well intended while the most recent is a sprint. Today's race is another route.

Is a situation like that something a decaying APV is capable of catching?

No, of course not. How could APV possibly correct for speed rating differences from sprint to route?


Aren’t you really talking about the slope of the horse’s performance curve with respect to the time of its performance? IMHO “time decay” is another one of those made up terms in horseracing.

Dude, that is a clown question.

Yep. I just made it up. I also made up regression and the internet.

[Cratos]

No, it isn't a "clown" question and I can prove it mathematically and statistically; and it is your type of nonrsene that paint this industry in bad light with the general public.

Yes, I understand that you are a subscriber vendor to this forum and I hope that you are successful in your entrepreneurial endeavor, but what you are attempting to explain with anecdotal evidence just don't pass the test.

There many people who don't play horses, but know plenty about statistical analysis and would laugh at your conclusion.

Therefore I am not going to turn this into an argument or a p___ing contest about who is right or wrong because the members of this forum and those who visit the forum and who understand analysis will have the acumen to understand the difference between our posts.

[HUSKER55]

I am having a bad morning. Isn't "time decay" the name of the slope in queston?

I call it stamina but I suppose you could call it "the cat's meow".

If we are not talking about the name of the slope would you be so kind as to "dumb it down" what we are talking about.

Thank you guys!

[DJofSD]

I am having a bad morning. Isn't "time decay" the name of the slope in queston?

I call it stamina but I suppose you could call it "the cat's meow".

If we are not talking about the name of the slope would you be so kind as to "dumb it down" what we are talking about.

Thank you guys!
I believe the basic premise is the further back into the PPs you go, the less valuable or applicable is the data. In this case, the data is the stats that are used to create the so called money box.

This "adjusted" APV is in contrast to the original APV where all races and the data from them were treated the same.

With the original APV, you were using the life time money box. This, as it states, includes information for all races in the horses career. The time decay APV does not give the same worth or relative weighting to data from 10 lines back as it does the most recent race.

Does that help?

[HUSKER55]

thanks!

Appreciate your efforts.

[DJofSD]

N. P.

Now, here's a little something to address an issue from earlier in this thread.

The standard APV from way back when treated all races and results more or less the same. This reflects the rather static purse structure that existed.

Now we have the racinos which have radically altered the size of purses and therefore the dollars earned by each runner that earns a piece of the purse.

A possible fix to address the problem would be to not use raw dollar values. Instead, use a relative purse structure rating.

By relative purse structure I mean scale the dollars earned to a value relative to the purse structure at the specific track. So, if a racino has a total amount for purses during the meeting of $10,000,000 and a bottom of the barrel claimer whens a race and earns $20,000 then assign a value for that $20,000 that is relative to the $10,000,000.

Do this for each line in the PPs and then construct the money box stats but instead of actual raw dollar values, use the relative values.

Maybe not the best approach but it's at least a start.