Can computers evaluate improving/declining form

[bettheoverlay]

For me, the most difficult, if not crucial, element of handicapping is trying to predict if a horse will move forward, regress or duplicate his current form. By the time I'm done a full card, and sometimes just one race, I feel braindead.

I have played around with some computer software, but it always seems like I'm getting alot of information about what happened before. I will look at a list of contenders and find horses with good recent races dropping in claiming price, 6yo fillies whose last 3 races had taxing finishes, horses establishing new highs in speed figures, things I would normally be skeptical of. Conversely, they don't seem to spotlight signs of improving form.

Are there programs out there that cover these types of questions? Or do these questions have an infinite number of variables, and are better left to the intuitive reasoning of a human brain?

[pmd62ndst]

Computers can't evaluate improving/declining form (i.e. Horse A will improve 30% today, Horse B will decline 5%...)

Computers can validate improving/declining angles. Some things I've learned:

- Horses with good trainers that drop in class shows declining form and are usually overbet.

- Horses with good trainers that step up in class shows improving form and are usually underbet.

Trainer intent is probably the single most important factor in determining improving/declining form and unfortunately, it's not quantifiable enough for a computer to evaluate.

PMD

[GameTheory]

Originally posted by pmd62ndst
Computers can't evaluate improving/declining form (i.e. Horse A will improve 30% today, Horse B will decline 5%...)Of course they can. A database query can't produce evaluations like that, but a computer can be programmed to predict improving/declining form.


Computers can validate improving/declining angles. Some things I've learned:

- Horses with good trainers that drop in class shows declining form and are usually overbet.

- Horses with good trainers that step up in class shows improving form and are usually underbet.Yeah, those are generally true.



Trainer intent is probably the single most important factor in determining improving/declining form and unfortunately, it's not quantifiable enough for a computer to evaluate.Trainer intent is certainly important, but form is also somewhat independent of trainer intent. And again, of course a computer can evaluate it. Can a computer run around and do "nuance handicapping" to see if the trainer or the groom is saddling the horse? No, but it can analyze trainer patterns that signal negative or positive performance changes. It won't always be right, but then neither will a person. If you want to say that the best evaluation you can possibly get is going to come from a human expert, I'll agree, but a computer can certainly do a good enough job for it to be valuable...

[mgardens@BTW]

Bettheoverlay, I'd suggest not playing the whole card. Find out what races you're best at handicapping and narrow your focus on them.

I do all the figs for those races and look for the best #'s. Some horses have a bunch, some don't have any, some haven't had any in a looooooong time.

Once I got narrowed to those few, I look for consistency with those #'s or improvement towards a top.

I use my own figures, so I can't give you an example outside of that. But I found that huge class drops raise red flags; if I find one with a matching figure that isn't dropping, I'll play that over the plummeting one & I'm usually right.

It goes pretty quick for me. For layoffs or switches, I look at the horse's history & maybe that of the trainer.

But I'd DEFINITELY say focus on playing the type of race you're best at handicapping.

[alysheba88]

The winner of a race tends to improve his Beyer by an average of 8 points (compared to his prior race).

As far as the underbet/overbet on droppers I used to be under the same misconception. Its actually not true. Horses rising in class are overbet and horses dropping underbet. One disclaimer though., When talking of dropping I am focusing on a single drop. From like 40k to $32. Not from 40k to 10k. Or multiple drops like 40-30-20- and now 10.

[VetScratch]

I think predictive form analysis could dramatically be improved if the software downloads included the most recent veterinary activities billed to each horse! :)

However, even then, we could debate trainer intent. I guess a charge for X-rays followed by a double drop could be interpreted two ways if such information was always published. :D

[Valuist]

IMO, the answer is no