#factors

[Derek2U]

if you were using a prediction formula, how many factors
do you think would be useful? Five? Six? What#?

[Claim digger]

My main factors:

Beyers on same surface and at similar distances (weighted according to how close the race was to todays race)
Days since last race (factor dependent on todays distance).
Post position today vs. previous race
Weight change
Trouble in previous races
No. of previous starts over similar surface

CD

[Tom]

The more you use, the lower the price probably - horse that meet too man rules will probably be obvious. I would use 2-4, different ones. In a simple one I used to use to get a rating for an odds line, I used best speed last three, days away, speed point percentage, and total of two year earnings (or BRIS class if I had it available).
None of the factors were releated, and if a horse was poor in any one, it was too penalized.

[Handle]

The one's that are performing the best in a given type of race.
But, Tom's got a point -- too many good things and the horse is too easy to spot and will be over bet.

For example, since this is one of Tom's favorite tracks. 7/23 at Finger Lakes a horse named Henry Robinson goes off at 26.5/1 in a msw route for NY breds. Now, Henry had never routed before, but still, after selecting pace lines for each horse that most reflected today's race, he had the best E2 + late rating (actually, a dominant figure), a strong late pace figure, and he won my simulated race to boot.

In the last 20 races of this type at FL, these three abilities produced winners from 27.78% to 35% of the time:
Strong Late: 27.78%
Best E2 + Late: 33.33%
Simulation Winner: 35%

Each of them had been producing a positive ROI.

However, the Best E1 and Best E2 abilities (which Henry also had, btw) only produced winners 16.67 and 18.18% of the time.
Henry wins and pays 55$ with those kind of "factors"? I love it.

My point is that, given no ability can be used day in and day out on its own to produce a positive ROI, I don't see how one could decide which one to use without some idea of how predictative that particular ability is, or, at least, has been recently. The same stats given above can radically change at different times, different tracks, different race types.

-Handle