Mechanical Beyer System - Horse Racing Forum - PaceAdvantage.Com

PaceAdvantage · 05-04-2010, 04:51 AM

I want to open up my "Beyer Bashers Unite" thread, where I will take on all the Beyer bashers and see if they can outperform a mechanical selection method using Beyer figures.

Only problem is, I'm not sure what mechanical method would be my best option. Has anyone out there done any studies as to what method might lose the least money? I know that pretty much any mechanical Beyer method picking every race will lose over time, but that's not the point.

Any ideas?

ClassTrumpsSpeed · 05-04-2010, 05:02 AM

Quote:

Originally Posted by PaceAdvantage

I know that pretty much any mechanical Beyer method picking every race will lose over time

This is from the guy who is NOT the Beyer-Basher?

Hehe.

PaceAdvantage · 05-04-2010, 05:35 AM

Quote:

Originally Posted by ClassTrumpsSpeed

This is from the guy who is NOT the Beyer-Basher?

Hehe.

Was that code word for "I'm chicken?"

lamboguy · 05-04-2010, 06:06 AM

beyer is the greatest thing that ever hit the racing form outside of the speed varient numbers that they have. a beyer number, sheet number or pace figures all amount to be the same thing. its like a girl in a bikini, it gives you alot, but it doesn't tell the whole story!

jonnielu · 05-04-2010, 06:46 AM

Quote:

Originally Posted by PaceAdvantage

I want to open up my "Beyer Bashers Unite" thread, where I will take on all the Beyer bashers and see if they can outperform a mechanical selection method using Beyer figures.

Only problem is, I'm not sure what mechanical method would be my best option. Has anyone out there done any studies as to what method might lose the least money? I know that pretty much any mechanical Beyer method picking every race will lose over time, but that's not the point.

Any ideas?

If the intention is to showcase the power of the Beyer, it seems like the mechanical method would need to revolve around the number for the most part, to make sure that the Beyer is doing the work. You would probably want to exclude races with 30% or more first timers too. Then the Beyer could have half a chance.

How about only mechanical methods up against the Beyer mechanical method?

jdl

illinoisbred · 05-04-2010, 06:54 AM

Going from memory on a study done years ago-use the best figure out of the last 3 attempts at the appropriate distance(+/- 1 furlong) and surface.Not all entrants in any given race will have 3 proper races, so any sample would have to be "cherry-picked"-most likely from age 4+ open claiming races,alw/opt. claimer,and stakes.

gm10 · 05-04-2010, 06:59 AM

I think Beyer describes a few blind bet methods in one of his books. Why not start with those.
Something like "bet the horse who has a 3 point edge (most recent Beyer) on the others, no debutants in the race, at least one run in the last 3 months". He gives strike rate and ROI for a couple of these mechanical systems.

Tom · 05-04-2010, 07:28 AM

CJ's 4 rules concept is pretty good with his numbers.

Overlay · 05-04-2010, 09:42 AM

The widest spread and most regular progression of values that I'm aware of for a Beyer measure ranks the horses in a field based on the average of the figures from those races out of each horse's last three starts that were run on the same surface as today's race (even if that means using the figure from only one race). If none of the horse's last three races were on the same surface as today's race, use the average of all three. If the horse has raced less than three times, apply the above rules to the available one or two races.

The 1/2/3/Front Half/Rear Half IV's for the different distance categories and surfaces are as follows:

Dirt Sprints: 2.42 / 1.62 / 1.21 / .78 / .57
Dirt Routes: 2.29 / 1.58 / 1.28 / .75 / .56
Turf Routes: 2.24 / 1.54 / 1.32 / .94 / .58

(I don't have data for turf sprints or for the various types of artificial surfaces.)(Sorry.)

DeltaLover · 05-04-2010, 10:26 AM

The problem with Beyers as it is for Ragozin's and Brown's figures, relies in the fact that since they are not provided in electronic format, they are difficult to be imported in a data base which will later be used for data mining and pattern recognition.

Somebody can make the case though, that since Beyers are not black boxes in contrary to Rags and Thorograph it is possible to be created from scratch without the need of DRF to serve as the provider.

I am wondering what are the IP issues related with the creation of an Open Source code base that will be calculating B's using raw historical data.... I have to assume that even if there are such issues we probably can go around them by altering in a minimal way the standard methodology described by B in his books....

GameTheory · 05-04-2010, 10:32 AM

But they are available in electronic format, at least in the PPs. Not sure if you can get them in the charts that way.

Beyers use a standard points = lengths formula, don't they? That is, if I'm looking at a Beyer for horse X that was beaten by 5 lengths, can I figure out what the Beyer for the winner of that race was based on that, i.e. BeyerOfWinner = BeyerOfBeatenHorse + BeatenLengthsOfBeatenHorse*adjfactor.

What is adjfactor?

rrbauer · 05-04-2010, 10:40 AM

Beaten lengths value is a function of race distance. If you wanted to take into account ground loss due to the trip or weight carried vs. some norm, that IMO is where "adjfactor" would come in. How that factor would be applied is subject to debate.

DeltaLover · 05-04-2010, 10:40 AM

By e-format I am not referring to the pdf version of DRF but to parsable file that can be imported to a data base.

As far as I know something like this was not available the last time I asked for it from the DRF guys (couple of years ago) and mostly this is the reason why I took the decision to 'tightly' couple my software to the bris data instead.

rrbauer · 05-04-2010, 10:43 AM

Does Formulator cvs export provide the data?

Jingle · 05-04-2010, 10:43 AM

Pace--I don't know if this helps but the guy (can't think of his name but Overlay will know it) who did the last big study on IV's found that averaging the last 3 Bris Speed ratings and taking the top number produced the best Win % of about 27-29% and best ROI.( The ROI was in the range of .91 if I recall correctly) This showed better than the last race, best of last 2 or three, etc.

I tried it with the Beyer's and I recall the Beyers did even better in the small sample I kept. I remember calling the guy to verify the last three had to be on the same surface and he said no, he just used the last 3 races but probably should have used the same surface and would if he did another study.

As I'm writing this I recall his name is Mike Nunamaker.