Pace & Rating

[Beetlebomb]

How can all these different software programs accurately project/predict pace lines for a horse if the race shape dictates a slower pace or a horse is being rated.

It seems it would totally invalidate the software's # crunching mechanism and make pace handicapping very suspect to each individul race??

[Capper Al]

We just beat this topic up in the thread "What's the case for Pace?" at the following link:


http://www.paceadvantage.com/forum/showthread.php?t=39585

[Tom]

What makes you think they all project/predict pacelines?

[Jeff P]

How can all these different software programs accurately project/predict pace lines for a horse if the race shape dictates a slower pace or a horse is being rated.

It seems it would totally invalidate the software's # crunching mechanism and make pace handicapping very suspect to each individul race??

IMHO, the question posed is a valid one.

In my own case, for JCapper, I created comprehensive pace ratings after some serious study of historical results of large data samples.

Instead of being "very suspect" (as you put it) the selections of the CPace and Pace Model Index algorithms that I developed for JCapper have proven to be pretty reliable. Neither factor is flat bet profitable on its own. But both can be included in UDMs (user defined models or spot plays) with other factors. For the most part I have found that UDM performance (as well as individual player performance) can be improved upon significantly when pace (defined as including both a horse's ability and running style along with the makeup of the race) are added into the factor mix.

My own calendar year 2006 database broken out by both PMI and CPace shows the following results:

All Starters in the Database:

Data Window Settings:
999 Divisor
Surface: (ALL*) Distance: (All*)
From Index File: C:\2007\Q1_2007\pl_Complete_History_06.txt


Data Summary Win Place Show
Mutuel Totals 309882.70 305170.00 303211.00
Bet -404684.00-404684.00-404684.00
Gain -94801.30 -99514.00-101473.00

Wins 24997 49798 73448
Plays 202342 202342 202342
PCT .1235 .2461 .3630

ROI 0.7657 0.7541 0.7493
Avg Mut 12.40 6.13 4.13


By: PMI Rank (2006)

Rank Gain Bet Roi Wins Plays Pct Impact
1.00 -3279.90 49722.00 0.9340 6335 24861 .2548 2.0627
2.00 -7019.30 49682.00 0.8587 4538 24841 .1827 1.4787
3.00 -8754.00 49804.00 0.8242 3592 24902 .1442 1.1676
4.00 -10107.80 49716.00 0.7967 2825 24858 .1136 0.9199
5.00 -14347.80 49094.00 0.7077 2207 24547 .0899 0.7278
6.00 -16764.30 45968.00 0.6353 1517 22984 .0660 0.5343
7.00 -34649.90 110226.00 0.6856 3961 55113 .0719 0.5818
8.00 141.40 414.00 1.3415 17 207 .0821 0.6648
9.00 -18.00 18.00 0.0000 0 9 .0000 0.0000
10.00 20.10 12.00 2.6750 4 6 .6667 5.3964
11.00 -3.80 10.00 0.6200 1 5 .2000 1.6189
12.00 -4.00 4.00 0.0000 0 2 .0000 0.0000
13.00 -8.00 8.00 0.0000 0 4 .0000 0.0000
14.00 -6.00 6.00 0.0000 0 3 .0000 0.0000
15.00 0.00 0.00 0.0000 0 0 .0000 0.0000
16.00 0.00 0.00 0.0000 0 0 .0000 0.0000
17.00 0.00 0.00 0.0000 0 0 .0000 0.0000
18.00 0.00 0.00 0.0000 0 0 .0000 0.0000
19.00+ 0.00 0.00 0.0000 0 0 .0000 0.0000


By: CPace Rank (2006)

Rank Gain Bet Roi Wins Plays Pct Impact
1 -3887.30 50402.00 0.9229 5442 25201 .2159 1.7437
2 -8310.90 50298.00 0.8348 4250 25149 .1690 1.3646
3 -8864.10 50344.00 0.8239 3720 25172 .1478 1.1933
4 -10202.10 50298.00 0.7972 3166 25149 .1259 1.0166
5 -13914.90 49820.00 0.7207 2622 24910 .1053 0.8500
6 -13153.70 46756.00 0.7187 2222 23378 .0950 0.7675
7 -13780.10 39054.00 0.6472 1605 19527 .0822 0.6637
8 -9921.90 28840.00 0.6560 1029 14420 .0714 0.5762
9 -7619.60 19580.00 0.6108 638 9790 .0652 0.5262
10 -4685.20 12616.00 0.6286 370 6308 .0587 0.4736
11 -2751.90 6834.00 0.5973 173 3417 .0506 0.4088
12 -1681.80 3394.00 0.5045 66 1697 .0389 0.3141
13 -113.20 484.00 0.7661 12 242 .0496 0.4004
14 193.00 200.00 1.9650 6 100 .0600 0.4845
15 -2.00 2.00 0.0000 0 1 .0000 0.0000
16 -2.00 2.00 0.0000 0 1 .0000 0.0000
17 -2.00 2.00 0.0000 0 1 .0000 0.0000
18 -2.00 2.00 0.0000 0 1 .0000 0.0000
19+ -4.00 4.00 0.0000 0 2 .0000 0.0000


-jp

.

[46zilzal]

How can all these different software programs accurately project/predict pace lines for a horse if the race shape dictates a slower pace or a horse is being rated.


MOST can't have their basic style of running altered too much and still be competitive.

[keilan]

For the most part I have found that UDM performance (as well as individual player performance) can be improved upon significantly when pace (defined as including both a horse's ability and running style along with the makeup of the race) are added into the factor mix.



-jp

.

Cheaper horses/fields and fillies have different pace parameters than say top level runners and those dynamics can help zero in on pace projection on a more consistent basis.

When I tried to use the same formula for all runners I soon realized the pace number/time was understated in weaker fields.

Cracking the nut for an accurate pace projection is one of the keys for eliminating low priced horses.

[cj]

Cheaper horses/fields and fillies have different pace parameters than say top level runners and those dynamics can help zero in on pace projection on a more consistent basis.

When I tried to use the same formula for all runners I soon realized the pace number/time was understated in weaker fields.

Cracking the nut for an accurate pace projection is one of the keys for eliminating low priced horses.

Very, very true. It is also usually overstated in classier fields.

Horse age is another example where pace is usually understated. With young 2yos, pace is about all that matters. With older stakes horses, it means very little. Every other race falls in between, where the younger and cheaper, the more pace matters.

[keilan]

Very, very true. It is also usually overstated in classier fields. Except in the Kentucky Derby ;) And that brings us to "size of field" in a competitive race.

Horse age is another example where pace is usually understated. With young 2yos, pace is about all that matters. With older stakes horses, it means very little. Every other race falls in between, where the younger and cheaper, the more pace matters.


Couldn't agree more about "age of the horse"

[cj]

Very, very true. It is also usually overstated (in classier fields of older horses)


Nice catch, above is what I meant to say.

[46zilzal]

Cheaper horses/fields and fillies have different pace parameters than say top level runners and those dynamics can help zero in on pace projection on a more consistent basis.


BINGO....Horses of a different color, or is it colour??

[kitts]

Success rarely comes just from interpreting the software output, IMHO. A lot of software has configuration options that allows the user to get what he/she needs. I, for one, do not make a bet after the software is done. It is merely a guide while I scan the pps for more info. The race shape might be difficult for software to interpret, but I don't bet without a comfort in my understanding of the shape of the race. The software output of Fair Odds then guides me to value bets.

[russowen77]

If there is a computer program out there that can do a better job of selecting pace lines than the average informed bettor they would really have something. That is the sort of query they are not really suited for at this time imo.

[DanG]

If there is a computer program out there that can do a better job of selecting pace lines than the average informed bettor they would really have something. That is the sort of query they are not really suited for at this time imo.
There is if you factor in volume betting. I agree a good player can isolate races and do better long term than a good automated method. Having said that HTR’s PL-5 (automated) is very impressive in large samples imo.

[Jeff P]

Cheaper horses/fields and fillies have different pace parameters than say top level runners and those dynamics can help zero in on pace projection on a more consistent basis.

When I tried to use the same formula for all runners I soon realized the pace number/time was understated in weaker fields.

Cracking the nut for an accurate pace projection is one of the keys for eliminating low priced horses.Keilan, I agree 100 percent.

The data I posted was in response to the following statement: It seems it would totally invalidate the software's # crunching mechanism and make pace handicapping very suspect to each individul race??I wanted to show that software developers do in fact have the ability to create fully automated algorithms that analyze pace scenarios in a way that produces uniform results. It goes without saying that within the overall results that I posted there are segments that can be identified and broken out... and some of those segments are better than others... and IMHO the player willing to focus on the better data segments has the ability to improve whatever he or she is doing to begin with.

-jp

.

[keilan]

Hey Jeff -- I fully realized what you were responding to, you are one of the good/great minds in the game.

Wish list for cj's program :)

The next and final step in the process of a program is having the ability within the program to distribute different measurements based on the present track weight.

ie: let’s say track weight has a range of 1 to 5 and 5 being the fastest. After the first couple of races the owner of the software estimates that the track weight is 2 and can then re-run his output based on this new information. Let’s call that calculation tw --- now the tw has the ability to assign a more likely finish of order based firstly on all of the other variables but will now include today’s track weight.

If a player understood the true abilities of each horse then was able to estimate the pace number with a high degree of accuracy and lastly could run the output based on current track weight………could it get any better?

[russowen77]

I don't know how you can get a real good pace line in maidens and cheaper claimers without looking at them. I go to the track many days with the pace already worked out and they won't get out of the paddock before I know that one has to be scratched. The fitness level, how rank, etc. If a maiden does the run the race in the post parade thing (piston like leg movements is best clue) they won't hang long if they even try.

In cheap claimers their fitness is so fleeting that it is really a guess in many races out here until you see which ones are fit and which are sore. That used to be where we won all our money. Then some A*****e came out with a video that was good enough to make a difference to the pilgrams. At one time there was not so many trained eyes and a lot of cheap sprints sometimes have only 1 horse fit. They don't stay that way long and it is ususally a long between times. Find one dappled at that level and he/she has to be break down to lose if none of the others are fit.

i keep forgetting that they could probably have a high enough n in this game to get good at the upper levels.

[Jeff P]

The next and final step in the process of a program is having the ability within the program to distribute different measurements based on the present track weight.

ie: let’s say track weight has a range of 1 to 5 and 5 being the fastest. After the first couple of races the owner of the software estimates that the track weight is 2 and can then re-run his output based on this new information. Let’s call that calculation tw --- now the tw has the ability to assign a more likely finish of order based firstly on all of the other variables but will now include today’s track weight.

If a player understood the true abilities of each horse then was able to estimate the pace number with a high degree of accuracy and lastly could run the output based on current track weight………could it get any better?Keilan, IMHO that's a VERY NICE idea... one that I've been kicking around for years but never really tackled.

I've always tried to measure track weight by comparing win pct of horses ranked by several different factors against established norms. For example CPace, PctE, and PMI (those are factors in my world... other users have different factors they can use for this) have known historical win pct for the top 4 ranked contenders. These can be used as benchmarks. By comparing recent race results against the established benchmarks I can get a fair idea as to current track weight. That's a pretty complicated process to do on the fly.

IMHO, the part of what you posted that is brilliant is grouping track weight into simple categories... for example TW=1, TW=2, 3, 4, 5, etc. Each of those individual TW categories could have their own benchmark ranges for whatever factors are used in the first place to determine Track Weight.

I never thought to create simple Track Weight categories before.

If track weight existed within a program in terms of simple categories... suddenly numerical adjustment to various numbers output by the program doesn't seem all that complicated of a thing to make happen.

-jp

.

[keilan]

Keilan, IMHO that's a VERY NICE idea... one that I've been kicking around for years but never really tackled.

I've always tried to measure track weight by comparing win pct of horses ranked by several different factors against established norms. For example CPace, PctE, and PMI (those are factors in my world... other users have different factors they can use for this) have known historical win pct for the top 4 ranked contenders. These can be used as benchmarks. By comparing recent race results against the established benchmarks I can get a fair idea as to current track weight. That's a pretty complicated process to do on the fly.


I use PctE to measure track weight effectively, the downside for multi-track players is that the player is required to actually handicap those races before they run, which works for me as I generally play one track at a time.

IMHO, the part of what you posted that is brilliant is grouping track weight into simple categories... for example TW=1, TW=2, 3, 4, 5, etc. Each of those individual TW categories could have their own benchmark ranges for whatever factors are used in the first place to determine Track Weight.

This is the easy part, Craig breaks the track weight out by a number. One could use variant information. From this a programmer simply stratifies race winners based on class, sex, age etc in the modelling process.


I never thought to create simple Track Weight categories before.

If track weight existed within a program in terms of simple categories... suddenly numerical adjustment to various numbers output by the program doesn't seem all that complicated of a thing to make happen.


This would be very easy to implement and within 6 months be road tested. I'd love to set up a room where Craig, Game Theory and the two of us reported for work each day until we were satisfied with it.

-jp

.

Hey a guy can dream :ThmbUp:

[Tom]

That was part of the old Power Pace method - you kept a traack profile and adjsuted your basic positional ratings by how the track was playing. Fred Davis also used someting similar in his methods. Even Sartin/Purdy had three "doctor" facots in Synergism - for early, presser, and sustained tracks.

This would be taking it to a new level, if you could get around lone speed and pace meltdown scenarios.