JCapper Message Board

JCapper 101
-- Actual v. Expected Report

Home	Register Log In

Actual v. Expected Report

joe_muller
12/2/2016
4:48:23 PM

I have a couple of questions about the new Actual v. Expected report/view that we now have available in the data window.

1. What is the expected measuring? I know it's measuring expected wins, but what are those expected wins based on? betting pool?

2. If I have a UDM and put an odds contraint in it, would the expected wins result be a betting pool expected win measurement as a result?

Example of UDM with no odds constraints:

A/E (Actual Wins: 7665) / (Expected Wins: 7509.56) = 1.0200

Example of same UDM with odds constraint of odds <=1.80

A/E (Actual Wins: 2891) / (Expected Wins: 2923.19) = 0.9900

If the odds constraint isn't getting me to the betting pool expectation, do we have that measurement somewhere?

Thanks!
Joe M.

jeff
12/3/2016
2:32:22 PM

--quote:

"1. What is the expected measuring? I know it's measuring expected wins, but what are those expected wins based on? betting pool?"

--end quote

--my reply:

Expected wins are based on a prob estimate for each starter given its odds.

Suppose for example, that the odds for the first starter returned by a set of query results are exactly 1.20 to one or 6/5.

The algorithm goes to a lookup table - finds the odds range for the current starter in the query results - and gets decimal win pct for horses at that (exact) odds range.

That decimal win pct is the prob estimate given the odds for that starter.

(FYI, the lookup table itself contains win pct based on a large data sample that I ran for every possible incremental odds range.)

In this case the exact odds range in the lookup tale would be: Odds >= 1.20 And Odds < 1.30 -- and the win pct or prob estimate after looking in the table is: 0.3824.

At this point only one starter has been returned by the query results and the number of expected wins for the entire sample is: 0.3824.

The Data Window then moves on to the next horse in the query results and the same process (What is the win pct given the odds of the current starter?) is repeated.

When the query is finished executing in the Data Window - expected wins reported by the Data Window is the sum of the decimal prob estimates for each horse in the query results.

I can illustrate this by using a top command to restrict the query results to the first 100 horses and by further restricting the query results with the following odds constraint:

AND ODDS >= 1.20 AND ODDS < 1.30

Under these circumstances, the decimal prob estimate for every horse in the query results is the same (0.3824) and expected wins for the entire sample is simply the number of starters in the query results (100) multiplied by the decimal prob estimate for each horse:

38.24 = 100 x 0.3824

Here's what that looks like in the Data Window with the results broken out by Actual Vs. Expected:


query start:         12/3/2016 12:05:27 PM
query end:           12/3/2016 12:05:27 PM
elapsed time:        0 seconds
`
Data Window Settings:
Connected to: C:\JCapper\exe\JCapper2.mdb
999 Divisor  Odds Cap: None
SQL UDM Plays Report: Hide
`
SQL:  SELECT TOP 100 * FROM STARTERHISTORY
      WHERE ODDS >= 1.2 
      AND ODDS < 1.3
`
`
Data Summary          Win         Place          Show
-----------------------------------------------------
Mutuel Totals      163.50        168.50        177.60
Bet               -200.00       -200.00       -200.00
-----------------------------------------------------
P/L                -36.50        -31.50        -22.40
`
Wins                   37            59            75
Plays                 100           100           100
PCT                 .3700         .5900         .7500
`
ROI                0.8175        0.8425        0.8880
Avg Mut              4.42          2.86          2.37
`
`
A/E (Actual Wins: 37) / (Expected Wins: 38.24) = 0.9700

Hope I managed to type most of that out in a way that makes sense.

-jp

.

joe_muller
12/4/2016
11:24:44 AM

Perfect!

Thanks Jeff!

Joe M.

JCapper Message Board

JCapper 101 -- Actual v. Expected Report

JCapper 101
-- Actual v. Expected Report