Database Handicapping Software- JCapper

JCapper Message Board

          General Discussion
                      -- Quick Update - OddsProxy New Factor

Home Register
Log In
By Quick Update - OddsProxy New Factor
jeff
7/15/2025
8:12:56 AM		I wanted to post an update about a new factor I've been working on.

The new factor is called OddsProxy.

It's going to be in the coming new full program update (and creating and optimizing this new factor is the reason I haven't yet published the new program update.)

 -- Out of sample testing 06-17-2025 fwd:


     query start:         7/2/2025 8:30:04 AM
     query end:           7/2/2025 8:30:10 AM
     elapsed time:        6 seconds

     Data Window Settings:
     Connected to: C:\JCapper\exe\JCapper2.mdb
     999 Divisor  Odds Cap: None
     SQL UDM Plays Report: Hide

     SQL:  SELECT * FROM STARTERHISTORY2025
           WHERE [DATE] >= #06-17-2025# 
           AND [DATE] <= #07-01-2025#


     Data Summary          Win         Place          Show
     -----------------------------------------------------
     Mutuel Totals    16830.30      16887.06      16955.64
     Bet             -22350.00     -22278.00     -21490.00
     -----------------------------------------------------
     P/L              -5519.70      -5390.94      -4534.36

     Wins                 1593          3162          4465
     Plays               11175         11139         10745
     PCT                 .1426         .2839         .4155

     ROI                0.7530        0.7588        0.7971
     Avg Mut             10.57          5.34          3.80



     By: SQL-F11 Rank (BUCKET_A_ODDSPROXY)

     Rank         P/L        Bet        Roi    Wins   Plays     Pct  Impact  AvgMut
     ------------------------------------------------------------------------------
        1     -365.48    3178.00     0.8850     525    1589   .3304  2.3178    5.36  
        2     -663.18    3168.00     0.7907     335    1584   .2115  1.4836    7.48  
        3     -517.30    3172.00     0.8369     261    1586   .1646  1.1544   10.17  
        4     -619.10    3154.00     0.8037     185    1577   .1173  0.8229   13.70  

        5     -794.86    3024.00     0.7371     138    1512   .0913  0.6403   16.15  
        6     -897.48    2558.00     0.6491      71    1279   .0555  0.3894   23.39  
        7     -786.42    1836.00     0.5717      44     918   .0479  0.3362   23.85  
        8     -366.36    1108.00     0.6694      22     554   .0397  0.2786   33.71  
        9      -25.62     646.00     0.9603      13     323   .0402  0.2823   47.72  
       10     -307.20     350.00     0.1223       2     175   .0114  0.0802   21.40  
       11     -108.00     108.00     0.0000       0      54   .0000  0.0000    0.00  
       12      -36.00      36.00     0.0000       0      18   .0000  0.0000    0.00  
       13        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       14        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       15        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       16        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       17        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       18        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       19        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  



     By: Morning Line Rank

     Rank         P/L        Bet        Roi    Wins   Plays     Pct  Impact  AvgMut
     ------------------------------------------------------------------------------
        1     -425.70    3196.00     0.8668     555    1598   .3473  2.4364    4.99  
        2     -665.60    3232.00     0.7941     347    1616   .2147  1.5063    7.40  
        3     -893.40    3318.00     0.7307     253    1659   .1525  1.0698    9.58  
        4     -852.46    3344.00     0.7451     183    1672   .1094  0.7678   13.61  

        5     -513.52    3198.00     0.8394     132    1599   .0826  0.5791   20.34  
        6     -746.50    2450.00     0.6953      71    1225   .0580  0.4066   23.99  
        7     -591.80    1808.00     0.6727      38     904   .0420  0.2949   32.01  
        8     -319.62     976.00     0.6725      11     488   .0225  0.1581   59.67  
        9     -168.40     506.00     0.6672       6     253   .0237  0.1664   56.27  
       10     -238.00     238.00     0.0000       0     119   .0000  0.0000    0.00  
       11      -58.00      58.00     0.0000       0      29   .0000  0.0000    0.00  
       12      -14.00      14.00     0.0000       0       7   .0000  0.0000    0.00  
       13        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       14        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       15        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       16        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       17        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       18        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00  
       19        0.00       0.00     0.0000       0       0   .0000  0.0000    0.00




My intent here is to create a Proxy for the Odds - a factor we can use for betting decisions during live play without (over) stressing about where the late money is going to land.

OddsProxy doesn't change (unless there's a gate scratch.) It isn't a moving target like the odds.

The new OddsProxy factor is based on a CLogit (or McFadden) model I created using the Data Transformations Tool in the new soon to be released full program update called JCapper Platinum 2025 Ver 2.0.

The new Data Transformations Tool allows you to select an Export Type from a drop down (in this case CLogit) which when selected makes a pre-programmed series of data transformations geared towards the export type selected by the user (again in this case CLogit) to data as it is exported from the new StarterHistory2025 table to .csv file on the hard drive.

From there, I brought the exported data into STATA, ran a series of CLogit (or Conditional Logistic Regression) commands, noting Std-Error and P-Values for the chosen factors, as well as Log-Likelihood in the output matrix.

If Std-Error and/or P-Value looked too high, or if log-likelihood looked too low, I removed the newly added factor and replaced it with another that represented the same area of the game I was hoping to address in the Model (Class, Figs ,Form, Early, Late, Rider, Trainer, Breeding, etc.)

Once I had what looked like a workable Model based on the output from STATA, I used the new MLR Quick Start Tool and Quick Grid to write the needed entries for the new Module [Model] into the ImpactValues table in the JCapper.mdb file.

From there I tested what I'd created by running DBBuilds on out of sample data.

Fyi, out of sample means race cards not used during Model creation. This included data from prior years AND new race cards going forward in time AFTER Model creation.

All of this might sound involved (even over your head.)

After I get the new program update out I'm going to make a video series covering this process.

It might look like rocket science. (But it's really not.)



Right now as I type this, out of sample testing clearly shows the new OddsProxy factor with Win Percentages for the top 4 ranks similar to the Morning Line.

Perhaps more importantly:

Win ROI for the top 4 ranks that outperform the Morning Line.

This without using the Tote, the Morning Line itself, HDW PSR, or Brisnet Prime Power as inputs.

OddsProxy as it exists right now is based entirely on data in the (new) StarterHistory2025 table.

That said, I'm still looking to optimize this a bit.

As a proof of concept project I manually added Will Pays to OddsProxy for a handful of race cards - and noted the resulting factor was nearly as accurate as final odds -- which can't be known until the horses are in the stretch.

Because of that, after I get the next Full program update out (I still have a few more things on my do list before I can make that happen) the first thing I'll be working on after that it [is] going to be getting the Live Play Module to parse Will Pays from various online toteboards.

Once I get that fully going later on this year:

I think most of you are going to like this new OddsProxy factor.


-jp
.






~Edited by: jeff  on:  7/15/2025  at:  8:12:56 AM~

Reply
NYMike
7/3/2025
4:16:26 PM		Sounds great!

Reply
JDL
7/6/2025
9:16:54 AM		Hi Jeff: How is OddsProxy factor different in functions to a UPR designed doing the same type of thing now?

Reply
jeff
7/7/2025
9:46:55 AM		There really isn't a difference in the way OddsProxy and UPR GroupNames based on CLogit are created.

I decided to 'bake' OddsProxy into JCapper Platinum 2025 for a number of reasons:

1. I wanted to give players in the JCapper User Community a better starting point than HDW PSR or Brisnet Prime Power.

2. It dovetails nicely with a number of soon to be released features in JCapper Platinum 2025 (that I'm not ready to talk about yet.)

3. Preliminary R&D suggests adding Will Pays to OddsProxy (which I plan to do later this year) will give us something almost as accurate as the tote that does not change after the gate opens.

The winner of last night's R2 at MNR is a perfect example: 6/1 Morning Line, 8/5 at the gate load, won going away in the stretch, final odds 2/5. (The Will Pays suggested where the late money was going to land.)

If you have statistical modeling chops you have the ability to create factors like this on your own (and probably have.)

But if you don't have statistical modeling chops....


-jp

.

Reply
Reply
Copyright © 2018 JCapper Software              back to the JCapper Message Board              www.JCapper.com