 | |
I recently completed a study of how well EquiSim's Profiler features have
been working. The conclusion of the study is that EquiSim's Profiler
feature predicts race winners with very good accuracy.
As with the Simulation Winner, and with any other aspect of horse racing, I
still contend the success of any particular factor is dynamic over time,
different from track to track and race type to race type.
The following results were recorded. They were obtained by using all
available races from 2002 held at the respective tracks. In order to do this I
added additional, developmental features in some instances. The following notes
explain the process I used.
- I started with a new, empty database.
- I selected to Store Selected Folders in EquiSim -- this typically puts
race information (simulation winner, Profiler Data) into EquiSim's Profiler
Database. Here I modified the program so that it would load each race card
in chronological order. When a card was loaded, it was simulated and then
Profiled automatically. This data -- including the Profiler Rank -- for each
horse was stored in the Profiler database. Important! There are
literally thousands of ways that the Profiler can be configured to produce
the Profiler value. For the most part I used the default settings. However,
I specified that:
 | Races would match if races came from the same track, same distance
category, were run on the same surface, and were not run on an off
track. |
| At least 20 races had to be found in a Profile. If 20 races were not
found, then the Profiler Rank was not stored in the database -- and thus
such races would not show up in the statistical analysis. This means
that not every race processed was recorded as one with a
Profiler Rank. Such is the nature of dynamic data. |
| No more than 20 races could appear in a Profile. |
| There was no regard for date range. That is, if a race occurred in a
spring meet, then that race would likely appear in the Profile of a race
early in the fall meet if it were the case that a track had two such
meets. |
| It should also be noted that there is an abundance of opportunity for
TIES to exist for the best Profiler Value. This almost never occurs with
the Simulation winner that is stored in the database. For that matter,
the statistics I use are always based on the number of starters,
not the number of races. For example, there could be 15 horses with the
Top Profiler rank in just 10 races. The win percentage would be derived
by looking at the total winning races divided by the total number of
starters. |
So far I have only tested using the previously mentioned
settings. I doubt that I found the optimum settings on the first try.
Given the initial positive results the future is very bright indeed for the
Profiler.
With the Profiler Rank of each horse in the database I could then perform
the following analysis, some of which is detailed below.
These statistics are representative of the differences between the Profiler
Value and the Simulation Winner in each race. In an analysis of 15 tracks every
single one of them showed that the Profiler Value had a better win
percentage than the Simulation Winner. At times the Simulation Winner did
produce a higher ROI. It appears that the the Simulation Winner is more likely
to pop gigantic (166$ in one case) winners than the top Profiler Horse is.
Statistics are given for the Simulation winner, the horse(s) with the Best
Profiler Value, and when the Simulation Winner is the same horse with the Best
Profiler Value. Note - statistics are determined based on the total number
of horses. For Sims, this essentially means the total number of races. But for
the Profiler Value, ties are deducted from overall statistics because the stats
use the total number of horses.
Belmont, All
| Type |
Races(starts) |
Wins |
Win% |
ROI % |
| Sim |
742 |
169 |
23 |
-25 |
| Profiler |
572 |
176 |
31 |
+2 |
| Sim+Prof |
332 |
106 |
32 |
-7 |
Belmont, TURF
| Type |
Starts/Races |
Wins |
Win% |
ROI % |
| Sim |
258 |
42 |
16 |
-38 |
| Profiler |
192 |
55 |
29 |
+6 |
| Sim+Prof |
96 |
23 |
24 |
-19 |
Delta Downs, All (even the scrub tracks work)
| Type |
Starts/Races |
Wins |
Win% |
ROI % |
| Sim |
99 |
26 |
26 |
-4 |
| Profiler |
25 |
8 |
32 |
+12 |
| Sim+Prof |
16 |
4 |
25 |
+3 |
Hollywood, ALL
| Type |
Starts/Races |
Wins |
Win% |
ROI % |
| Sim |
541 |
150 |
28 |
5 |
| Profiler |
406 |
125 |
31 |
-8 |
| Sim+Prof |
249 |
75 |
30 |
-9 |
Hollywood, TURF
| Type |
Starts/Races |
Wins |
Win% |
ROI % |
| Sim |
164 |
38 |
23 |
+12 |
| Profiler |
111 |
32 |
29 |
0 |
| Sim+Prof |
53 |
15 |
28 |
+1 |
LAD, ALL
| Type |
Starts/Races |
Wins |
Win% |
ROI % |
| Sim |
676 |
161 |
24 |
-18 |
| Profiler |
585 |
199 |
34 |
+5 |
| Sim+Prof |
358 |
117 |
33 |
-1 |
LAD, Dirt
| Type |
Starts/Races |
Wins |
Win% |
ROI % |
| Sim |
531 |
132 |
25 |
-13 |
| Profiler |
469 |
168 |
36 |
+13 |
| Sim+Prof |
289 |
99 |
34 |
+5 |
|
