BRIS Prime Power

[GMB@BP]

I stumbled upon an article when reading about the Dubai race recaps about using BRIS Prime Power to find contenders quickly. It is linked below

http://horseplayersassociation.org/apr18issue.pdf

On page 15 there is an article about using the BRIS prime rating to eliminate the field down to 5 contenders (or as I state below). I wanted to run my own samples, but tighten it up a bit.

I only used non maiden races. If a race had the following fields I limited it to the number of Prime rating contenders that I considered when going through the results

5 or less = 2
6=3
7=3 or possible 4 if the fourth contender was withing 2 points of the third horse
8=4
9=4 or same as 7 for including the 5th contender
10+=5 contenders

With no other handicapping the win% is 84% over 57 races going back about 8 days. I am going to keep updating the numbers but that is a pretty strong sample.

[deathandgravity]

I always look @ the BRIS Prime Power & it figures into my handicapping.
My data set (DB) is a little stale & hasn’t been update much the past couple years other than spot plays on occasional weekends.
18,000+ races in my DB (NYRA & California) & the best ROI I have for Prime Power is:
Sprint Turf with Prime Power +5 points nearest contender:
Races: 1705
Win %: 40.65
ROI: -6.46

[Dave Schwartz]

With no other handicapping the win% is 84% over 57 races going back about 8 days. I am going to keep updating the numbers but that is a pretty strong sample.

84% is actually kind of low. It will go up higher.

This is HDW's Projected Speed Rating. It is their equivalent to BRIS Prime Power.

174-rPSRWIN BETS
Field1 Field2  Starts   Pays    Pct   $Net     IV
1st             5,441  1,655   30.4  $1.60   2.27
2nd             5,241  1,102   21.0  $1.72   1.57
3rd             5,176    741   14.3  $1.63   1.07
4th             4,935    529   10.7  $1.58   0.80
5th             4,954    421    8.5  $1.55   0.64
6th             4,489    264    5.9  $1.39   0.46
7th             3,592    149    4.1  $1.25   0.35
8th             2,491     95    3.8  $1.53   0.35
9+              2,876     53    1.8  $0.93   0.19

Total          39,195  5,009   12.8  $1.50   1.00

1st-5th        25,747  4,448   17.3  $1.62   1.29
                       88.8%

[Speed Figure]

I noticed that HDW stopped giving daily updates on how PSR was performing.

[JustRalph]

I used to have a BRIS PP UdM in Jcapper that flagged top 3 prime power horses that were over 4-1 morning line, and ranked outside the top 3 Jcapper picks. It kind of flagged a horse that maybe had a bad last race or something that would turn the public off, but pointed out an over-reaction to that factor.......

You didn’t get a hit very often, but letting a top 3 PP horse creep up on the odds board above 5-1 actually happen enough to allow for some pretty nice tickets.

[Dave Schwartz]

Originally Posted by Speed Figure

I noticed that HDW stopped giving daily updates on how PSR was performing.

Didn't know that and no idea why.

It still hits right at 30-31%.

Cramer's older version - what we call "Cramer Power" - is about 2% back but returns a little more money because it doesn't correlate quite as much with the tote board.

I have one that has been around for years called (simply) RTG. It also hits 29-31% but returns more money than either one because it correlates even less.

None - including BPP - will get you close to being even. In fact, they almost guarantee that you will be on a low-priced horse.

Here we look at PSR rank=1 cross referenced with Public Choice Rank. Nothing exciting to see here. $Nets are very flat.

174-rPSR+1-rPubCh 
WIN BETS
PSR    PubCh   Starts   Pays    Pct   $Net     IV
1st    1st      3,139  1,232   39.2  $1.65   2.90
1st    2nd      1,267    282   22.3  $1.56   1.67
1st    3rd        557     96   17.2  $1.53   1.30
1st    FH         361     37   10.2  $1.53   0.83
1st    RH         117      8    6.8  $1.35   0.49

Total           5,441  1,655   30.4  $1.60   2.27

I thought you might like to see the two fields reversed.

Here we look at Public Choice cross-referenced with PSR. (I showed all the ranks here.)

1-rPubCh+174-rPSR   
WIN BETS
PubCh  PSR    Starts   Pays    Pct   $Net     IV
1st    1st      3,139  1,232   39.2  $1.65   2.90
1st    2nd      1,102    396   35.9  $1.80   2.69
1st    3rd        466    145   31.1  $1.70   2.38
1st    FH         291     81   27.8  $1.64   2.27
1st    RH         113     35   31.0  $1.80   2.13
2nd    1st      1,267    282   22.3  $1.56   1.67
2nd    2nd      1,829    391   21.4  $1.64   1.57
2nd    3rd        952    182   19.1  $1.47   1.42
2nd    FH         645    116   18.0  $1.53   1.47
2nd    RH         303     56   18.5  $1.61   1.31
3rd    1st        557     96   17.2  $1.53   1.30
3rd    2nd      1,110    170   15.3  $1.55   1.13
3rd    3rd      1,418    196   13.8  $1.46   1.01
3rd    FH       1,166    144   12.3  $1.42   1.00
3rd    RH         716    115   16.1  $1.78   1.12
FH     1st        361     37   10.2  $1.53   0.83
FH     2nd        842    121   14.4  $2.14   1.18
FH     3rd      1,487    155   10.4  $1.76   0.84
FH     FH       3,507    305    8.7  $1.62   0.76
FH     RH       2,167    183    8.4  $1.60   0.70
RH     1st        117      8    6.8  $1.35   0.49
RH     2nd        358     24    6.7  $1.41   0.47
RH     3rd        853     63    7.4  $1.85   0.52
RH     FH       2,732    121    4.4  $1.42   0.37
RH     RH      11,697    355    3.0  $1.25   0.24

Total          39,195  5,009   12.8  $1.50   1.00

Here we find one "sweet spot:" Horses in the Front Half of the field for Public Choice (but >3rd) and 2nd best for PSR. Likely an aberration caused by a couple of giant payoffs.

[Handiman]

I just ran 1434 race cards through my batch processor last night and then looked at some factors looking to improve the factor set for Handifast. The HDW prime Power hit at just over 34% winners. Now I admit that the race cards are several years old. But it was interesting as was the averaging of speed ratings. That factor hit at just over 31% winners.

Now I have a friend who shall remain nameless, (D. S. of Reno fame) lol, who has access to 4,000 factors. There is no shortage of information and combos of that information, which is really mind boggling when you think that we are watching a limited number of horses just running around in a circle and trying to determine which one gets there first. Every result is the same. Barring disqualifications, the first horse across the wire wins. Seems like it shouldn't be that hard to figure out. But that's the Pollyanna in me I guess.

Working on this puzzle never gets boring and helps me keep the ole brain juices flowing. As they say, Use it or Lose it. I am encouraged though with some of the numbers I am getting from some factors that I am looking at.... I hope to have a complete factor revision soon. Well, not a complete revision as Mike has said, we have some factors that are holding their value.

Handi

[GMB@BP]

I am only interested in the contenders, I am much better than 31% picking a winner out of 4 or 5 horses.

[Dave Schwartz]

Originally Posted by Handiman

Now I have a friend who shall remain nameless, (D. S. of Reno fame) lol, who has access to 4,000 factors. There is no shortage of information and combos of that information, which is really mind boggling...
Handi

For those who might not know, Handi and I are really good friends. He's stayed at my home a number of times and often is kind enough to house sit when we leave town (and his schedule allows). Beth, the dogs and George (the Cockatoo) just love the guy.

But he pretty much ruins my life when he comes to visit because we spend hours discussing racing, factors, and programming. I get very little done.


Why am I telling you this?

Because if you are writing your own software and need some advice or a little coaching in your software development, I can be available to help.

Don't misunderstand... I am not offering to be on-call every time you have a question. I just cannot do that. I'd never get anything done. (There is only room in my life for one Handifast. LOL)

But, if you are coding your own software and want some time to talk about it, send me an email and we'll make an appointment.


Best to all,
Dave

[Handiman]

GMB.... I would hope you're better than 31% out of 4 or 5 contenders. But this was a run inclusive of all horses looking for validation of factor relevance. Contender selection is a whole other subject. I would propose that contender selection is done with relevant factors or some combo there of and without relevant factors to consider, contender selection would be horrible. At least ineffective at best.

Handi

[headhawg]

Originally Posted by Dave Schwartz

Because if you are writing your own software and need some advice or a little coaching in your software development, I can be available to help.

Two things entered my mind when I read that: "Be careful what you wish for" and "No good deed goes unpunished". I know that you're good for it, though, as you mentioned it in an email to me. It's a nice offer, Dave.

@Handi -- Handifast should be killer if you have access to those factors. Geez...I'm jealous.

[mikesal57]

Hey Guys...

First off, I want to apologize to Dave for being an ass on several occasions
Dave has even helped me a few times and what I did was uncalled for..
Sorry , again...

As for writing code for a software program, the one thing I tell guys is to include some sort of database that your factors can be analyzed.
You work so hard in getting ideas into software why not go the next step and validate to yourself and others that they DO work.....
Handifast was put together maybe 6-7 years ago and I finally got Handi to include a database....from there I was able to tell which factors were good and which were not with a statistical math program....
I believe with Dave's help this can be an excellent program.

A question for Dave, have you done any logic regression work on your factors?


Mike

[pandy]

It's nice that people have tested the Bris Prime Power, but it would be interesting to see how it does using odds ranges and using different rankings. For instance, playing horses that went off at 4-1 or better than had the top ranking, or playing the highest odds horse ranked in the top four Bris Prime Power. Either of these should produce higher ROI's than betting the top ranked pick.

[Dave Schwartz]

Originally Posted by mikesal57

Hey Guys...

First off, I want to apologize to Dave for being an ass on several occasions
Dave has even helped me a few times and what I did was uncalled for..
Sorry , again...

As for writing code for a software program, the one thing I tell guys is to include some sort of database that your factors can be analyzed.
You work so hard in getting ideas into software why not go the next step and validate to yourself and others that they DO work.....
Handifast was put together maybe 6-7 years ago and I finally got Handi to include a database....from there I was able to tell which factors were good and which were not with a statistical math program....
I believe with Dave's help this can be an excellent program.

A question for Dave, have you done any logic regression work on your factors?


Mike

1. Apology accepted. Thank you.

2. I assume you mean "logistic regression." I am not a huge fan
of this process for racing simply because it is what the whales use and it places me in direct competition with the best.

Don't misunderstand - the process is fine - there is a smattering of a pseudo R^2 in a couple of places. Specifically, McFadden R^2.

I've just intentionally avoided anything that puts me in direct competition with the whales.

Remember that I am not a stats geek. I am a physics geek.
(He said proudly.)


Dave

[mikesal57]

Originally Posted by Dave Schwartz

1. Apology accepted. Thank you.

2. I assume you mean "logistic regression." I am not a huge fan
of this process for racing simply because it is what the whales use and it places me in direct competition with the best.

Don't misunderstand - the process is fine - there is a smattering of a pseudo R^2 in a couple of places. Specifically, McFadden R^2.

I've just intentionally avoided anything that puts me in direct competition with the whales.

Remember that I am not a stats geek. I am a physics geek.
(He said proudly.)


Dave

Question ...Why would you believe that your in "competition" with the other guys..??

Maybe I'm misinterpreting the R stuff....but

Isn't the math based on the factors you created?

Lets use your 4000 factors ( I know you cant use em all at once )
or some other amount....

Logistic regression will give you a correlation of those factors to the winning horses..you take out the ones that dont model winning horses ...and then refine those that are left .....to a point that you can be used and weighed like the Fibonacci approach...but a better predictability...

Mike