The evolution of a system

[formula_2002]

Part 1

572 million calculations.

We start with a data base of 143,351 horses.
We set aside 20% of those horses as our ‘hold out’. We will use the ‘hold out’ to test any conclusion we make upon analyzing the remaining 114,680 horses.

We will exclude any horse that was listed as entry prior to any scratches.

Each horse in our data base is qualified by 78 factors, consisting of various ratings for speeds, pace, class, form, running style etc. Where two or more of the same ratings exist, they have been calculated by different authorities. For instance, we may have several pace figures prepared Bris, Brohammer, Hall etc

Each factor has a ranking of 1 to over 8. We will look at the top 8 ranked horses.

We begin by calculating the win%, profit or loss, actual wins, expected wins (1/(odds+)), average winning final odds, average final running odds, fair value and number of horses for each factor, for each horse, for each ranking.

[formula_2002]

Part 2

After exploring the hundreds ways to proceed from this point (the subject of another paper), we will explore the following;

We will look at the ranked factor returning best, dollar profit or least dollar loss.
We find a ‘X’ factor that returned a dollar loss of but 7 cents on the dollar in over 14,000 races.

Now we analyze that single factor by incremental odds ranges.
The first three odds ranges 0 ,1 and 2 produced losses of 10 cents, 13 cents and 15 cents respectively.

Eliminating the favorite still produces a loss of 13 cent.

There appears to be some attainable profits in odds ranges 3 to 11 in various degrees of profits and losses.

There are almost 7,000 horses in the odds range of 3 to 11 that return a loss of but 1 cent on the dollar, at a win rate of 18%. (we are left with but 6% of our 114,000 horse sample)
------------------------------------------------------------------------------------------------------------

[Larry Hamilton]

Apparently, I am missing the point here as you are appear to be describing a methodology that doesn't work to a group with 95% of it's members capable of losing on their own. There are only two scenarios horseplayers need: how to win or how to quit playing. Anything else is self-indulgent.

[formula_2002]

Larry, stick around, eventually, the point will become quite clear.


Joe M

[Larry Hamilton]

no thanks, you just made my ignore list

[hurrikane]

I believe the point will be 'Joe is a loser so everyone else is too'.

Joe...78 factors? Way to back fit buddy! You are wasting your time.

[formula_2002]

Larry, first open your mind, then your eyes.
I'm 'holding out" 20% from test sample.

[formula_2002]

Part 3, down to 2619 horses

How do these horses perform (for the bettor) at various north American tracks?

We find that 2619 horses racing at New York and California tracks returned a profit of 3% and a win % of 19%. (that’s 2.3% of our entire sample before we obtained a profit).

Time to test the system on the ‘hold out’

And the results are….see Part 4

[formula_2002]

Part 4

First the good news. In the’ hold out’ sample, the ‘X’ factor horse lost 8 cent compared to the tested sample of 7 cents.

Not so good news, in the ‘hold out’ sample, the ‘X’ factor horse in the odds range of 3 to 11, produced a loss of 5 cent son the dollar, compared to the tested sample loss of 1 cent.


The disappointing news! In the ‘hold out’ sample, the ‘X’ factor horse in the odds range of 3 to 11, at the New York and California tracks produced a loss of 14 cents on the dollar, compared to the tested sample profit of 3 cents.
There were 1035 horses in the ‘hold out’ that qualified.

Part 5 “What next ?”

[Fastracehorse]

The Extinction of a System.

By Formula 1902.

fffastt

[formula_2002]

fast, try this. click on listen now.. it may help you through this difficult period of reading my post]

http://www.wqxr.com/cgi-bin/iowa/index.html

[Fastracehorse]

1902 - Come on: You are talking to a handicapper.

Not someone who speacializes in meaninglessness.



fffastt

[hurrikane]

here's an idea. Joe...go to your website!

[penguinfan]

For what it's worth Formula, it seems you are proving you can't beat the game by playing every race, but I can win by picking my spots.

Penguinfan

[Show Me the Wire]

Joe:

I do not understand what or why you are sharing your research with us. I particularly don't understand the significance of using 78 factors or the purpose of using so many factors.

If I was researching (and I am not) winning factors, I would limit the factors to maybe five factors, including distance, class, fitness (find a factor that indicates such), some sort of speed rating and weight.

Distance, class, and speed rating are fairly simple to identify, but fitness is the elusive factor. Maybe that is why there are so many systems based on angles such as layoffs i.e. 2nd off a freshening layoff, trainer intent, class drops, out of the money finishes, etc.

Speed ratings are popular because a good speed figure can indicate fitness and by definiton measures pace, because final time is a function of pace. Sheet type figure usage is based on predicting how the fit the horse is coming up to today's effort, through interpreting patterns of recent prior efforts and it appears this ideology is being incorporated with Beyers.

I similarly believe the fitness factor is the most significant issue that makes it difficult to automatically select a representative pace line for velocity type ratings. In order to help identify fitness, Brohamer in MODERN PACE HANDICAPPING recommended using turn-time as the important factor to measure current form (fitness). Therefore, fitness is one of the important factors that needed to be used for velocity handicapping.

More than likely the investors that repeatedly beat the take out are the better predictors of fitness. If a valid factor(s) measuring fitness were lacking in the factor mix, then the research would be unreliable in predicting if wagering on horse racing could be profitable in the long-term.

Regards,
Show Me the Wire

Perception is reality