Have Beyers Jumped The Shark

[PowerUpPaynter]

This is something iv been thinking about for the past few months. Is the beyer system an old fad that has lost validity?

Latest example: Nyquist getting a 94 Beyer yesterday running two seconds faster than Materiality last year who got a 110...

Is the beyer system a flawed fad that is dying?

[cj]

I had Materiality at 119, Nyquist at 116. I thought all along the Materiality Beyer figure was too high, and nothing that happened later in the year dispelled that IMO.

That said, the track was much slower last year. If you check the other races on both days it really can't be argued.

[PowerUpPaynter]

Originally Posted by cj

I had Materiality at 119, Nyquist at 116. I thought all along the Materiality Beyer figure was too high, and nothing that happened later in the year dispelled that IMO.

That said, the track was much slower last year. If you check the other races on both days it really can't be argued.

yes i agree and your figures are more like it. but a 110 to a 94 is a much bigger difference.

[Augenj]

Originally Posted by PowerUpPaynter

This is something iv been thinking about for the past few months. Is the beyer system an old fad that has lost validity?

Latest example: Nyquist getting a 94 Beyer yesterday running two seconds faster than Materiality last year who got a 110...

Is the beyer system a flawed fad that is dying?

Pardon my cynicism, but doesn't Beyer subtract 10 points if it's a West Coast horse?

[cj]

Originally Posted by Augenj

Pardon my cynicism, but doesn't Beyer subtract 10 points if it's a West Coast horse?

Come on, that is just silly.

[pandy]

This kind of goes back to a conversation we were having on the other speed figure thread. In last year's Fla. Derby, the track was slow and the speed figures were over adjusted. As it turned out, the track wasn't the only thing that was slow, the horses were slow, too, because the Florida Derby turned out to be a negative key race. As I alluded to in the other thread, you can't trust high speed figures unless the actual time was fast.

[cj]

Originally Posted by pandy

This kind of goes back to a conversation we were having on the other speed figure thread. In last year's Fla. Derby, the track was slow and the speed figures were over adjusted. As it turned out, the track wasn't the only thing that was slow, the horses were slow, too, because the Florida Derby turned out to be a negative key race. As I alluded to in the other thread, you can't trust high speed figures unless the actual time was fast.

I've done a few queries. I might post later this week, maybe not. But with my figures at least this isn't true at all.

[DeltaLover]

I agree that the beyers as published in DRF have been degenerated and are not a reliable measurement any more.

For the history my figures look as follows:

Nyquist: 102

Materiality: 105

The 3 point difference in favour of Materiality can be translated as approximately 3 ticks in six furlongs race.

Still, we gain absolutely nothing by complaining about the low quality of the published figures as they account for most of the errors that can be found in the public's betting.. Instead of brining up their inefficiency and pointing to their errors we better stay quiet and let people continue with them...

[DeltaLover]

Originally Posted by pandy

This kind of goes back to a conversation we were having on the other speed figure thread. In last year's Fla. Derby, the track was slow and the speed figures were over adjusted. As it turned out, the track wasn't the only thing that was slow, the horses were slow, too, because the Florida Derby turned out to be a negative key race. As I alluded to in the other thread, you can't trust high speed figures unless the actual time was fast.

While what you say here might be true for some of the commercial figures, definitely does not apply globally and it is completely doable to create an algorithm that can perform even in the more extreme ranges of the variant distribution.

[pandy]

Originally Posted by DeltaLover

While what you say here might be true for some of the commercial figures, definitely does not apply globally and it is completely doable to create an algorithm that can perform even in the more extreme ranges of the variant distribution.

Not really. If there are 10 races and 8 of them are turf races, one is a dirt sprint and one is a route sprint, you may trust your dirt figures that day but I wouldn't. I did figures for all of the major tracks for years, so I know all of the pitfalls. My point in the other thread, when two horses come back to race against each other and they have the same figure over the track in the same class, but one of the horses ran a much faster race, the horse that ran the faster race has the edge. Now, this isn't fool proof. Sometimes the slower horse wins, but over the long run, the horse that ran the actual faster time finishes in front of the horse than ran the slower time more often.

If times don't matter, only speed figures, then methods like "velocity ratings", which are based on the actual fractions the horse ran, are worthless.

[cj]

Originally Posted by pandy

Not really. If there are 10 races and 8 of them are turf races, one is a dirt sprint and one is a route sprint, you may trust your dirt figures that day but I wouldn't. I did figures for all of the major tracks for years, so I know all of the pitfalls. My point in the other thread, when two horses come back to race against each other and they have the same figure over the track in the same class, but one of the horses ran a much faster race, the horse that ran the faster race has the edge. Now, this isn't fool proof. Sometimes the slower horse wins, but over the long run, the horse that ran the actual faster time finishes in front of the horse than ran the slower time more often.

If times don't matter, only speed figures, then methods like "velocity ratings", which are based on the actual fractions the horse ran, are worthless.

That is a different story than your first post, having a limited sample. In that case, sometimes I am very confident in the figures, sometimes not. It depends on the horses and the consistency of my projections.

I personally think velocity ratings, while maybe not worthless, are inferior to ratings adjusted for the conditions.

[Augenj]

Originally Posted by cj

Come on, that is just silly.

Nobody every accused me of being serious.

[pandy]

Originally Posted by cj

That is a different story than your first post, having a limited sample. In that case, sometimes I am very confident in the figures, sometimes not. It depends on the horses and the consistency of my projections.

I personally think velocity ratings, while maybe not worthless, are inferior to ratings adjusted for the conditions.

My Diamond System is sort of like velocity ratings, although I prefer, "pace balanced" ratings, but my formula is different because I include late pace in the final rating. Speed figures should pick more winners than a velocity style rating but the average payoffs can be considerably higher with the velocity style rating, which is one of the reasons why people use them. I posted all of the top ranked Diamond horses for Del Mar on my website last summer and the top ranked horse won 20% of the time and generated a profit for the meet. At easier meets, the win percentage can be much higher.

But, I don't think I've ever seen a profit produced over the course of a meet using a horse's best recent or last race speed figure, although the win percentage is higher.

I'm not knocking speed figures. I'm just saying, a horse's actual fractions and final time should not be ignored. Fractional times are extremely important.

[DeltaLover]

Originally Posted by pandy

I'm not knocking speed figures. I'm just saying, a horse's actual fractions and final time should not be ignored. Fractional times are extremely important.

I do not be think anyone can disagree with what you are saying here. Of course fractional times are important and the same applies for speed figures; still they are highly correlated and represent highly correlated metrics, something that is easy to overlook.

[classhandicapper]

Originally Posted by pandy

I'm not knocking speed figures. I'm just saying, a horse's actual fractions and final time should not be ignored. Fractional times are extremely important.

I think you are on to something. I just don't want to elaborate much.

IMHO, the entire model upon which pace and final time speed figures is based is broken. I'm not sure how much it matters because figure makers often back into the right answers for the wrong reasons by splitting the variant, breaking races out etc... Still, if it were all done properly (or at least better), I'd have to think it would give someone an edge. I more or less know what has to be done from a theoretical perspective to get it all right, but actually doing the work to get there would be a huge project and I'm not up to the task now. Also, there's one complication I don't have a good solution for yet other than watching the races and using subjective judgement. There's only so many races one person can handicap thoroughly and watch.

In the end, all the current time based solutions are fraught with disagreement and error. There are no easy solutions.

Horse racing is closer to basketball (my other gambling hobby) than I realized. There are loads of advanced metrics in basketball used to measure player productivity. They are generally developed by guys with sky high IQs and advanced degrees in statistics and math. Yet despite their best efforts, these metrics all have subtle flaws in different areas and usually disagree (sometimes strongly) about the same players. You eventually get to the point where you realize there isn't a "right answer" among any of them that works all the time. They are all just tools that give you different views on the same thing that you can use to divine the right answer.

The next time I express a strong opinion on a horse you should probably smack me in the head, because unlike most people, I know better.