Simple Class of Race using Beyers

[nomadpat]

I am laid up for a few weeks (broke my leg skiing :mad: ) and work said to not be in a hurry to get back, so might as well explore some handicapping projects.

My question to everyone is how would you quantify how hard/class of race using Beyer figures?
Disclaimers to my question:
1. I am not actually going to use Beyers as I make speed figs for QH races - I will be using my figs.
2. I understand all of the flaws of using Beyers (pace scenarios, horses dropping in class will have better ones in the past and may be off form, etc).

But I am looking for more of a procedural answer, e.g. "take the average of the last three races of all horses" or "average only top 3 3 most recent races as they will most likely determine if a race is hard".

So let's say in a given 5 horse race:
:1: : 56/65/60 (more recent first)
:2: : 75/75/75
:3: : 70/80/65
:4: : 75/80/75
:5: : 80/80/65

So horse :1: may not be used (depending on the methodology) as the other horses represent tougher competition. So whatever figure the :1: runs in this race can be compared to the calculated class of the race to understand that this figure would have been run against better horses. In his subsequent start, it might be interesting to note he ran against better horses than he is facing then and can understand that his fig from this race can be improved.

Thanks in advance to any ideas out there!

[bobphilo]

Brisnet seems to do something like this when they caculate a races Class figure, which they call the Race rating. I think they average the expected times of the first 4 finishers. The winner's Class rating is usually 3 points above the Race rating. I'm not sure whether they do this by full dress handicapping or just averaging each horse's last 4 races. It does give a more accurate rating than just the race conditions or purse value.

This looks a lot like what you do except you use the Beyer ratings and the whole field. I would concentrate on just the top 3 or 4 finishers. Beyond these the others are hopelesly beaten and/or off form and not running a typical race for that class.

Bob

[Light]

Reminds me of a race about a year ago at Aqu for 6f 4^ clm $10K on Feb 18 2008.

The favorite's figs:

E2 LP SR Class Date

91 105 100 10K Feb 1

90 92 88 14K Dec 30

Now look at the winner's fig @6-1

E2 LP SR Class Date

84 93 84 14K Dec 30

These 2 horses came out of the same Dec 30th race. If the favorite increased his SR by 12 points with the drop then the 6-1 had to be projected to increase his SR by 12 points as well with the same class drop. Instead what most people are handicapping by is that the favorite ran a 100 and this 6-1 shot ran an 84 for slighly higher and probably will still come up short to the favorite.

My reasoning was that the drop is the wake up call.How "hard" a horse ran for the higher tag is not that big of a deal to me. I watch replays of horses losing under strong urging,dont bet them cause I think they're "used up" and watch them romp next out.Why? Cause today there is something new and a class drop certainly is a wake up call. BTW the eventual winner was obviously well rested since its last start with some really nice workouts. Your concept of "hard race" has to incorporate the days since raced.

Since the favorite already dropped,he had his wake up call and stuck at the same level would only regress since he needed the previous drop in the first place and he's not going anywhere today.Sure enough he finished 3rd regressing to his previous SR and was given an 89. The 6-1 was definetly going to wake up since we have direct proof from another horse that did that out of the same race (the fav).When you combine the improvement expected for the 6-1 shot and the regression for the favorite,the 6-1 shot comes out on top. The 6-1 shot did improve exactly 12 points and ran a 96 in victory. This was my reasoning in real life and I did cash on this one.

I'm sure you are aware of most of what I said. But I still find it fascinating how much horses do improve with a class drop. I offer this explanation as one possible answer. I think the main answer varies with horses individualistic idiosyncrosies and thats why this game is so challenging. Even their trainers and owners dont really know 100% of the time how their horses will run. But where there is something concrete offered in the form as in the above example,I'll take it. And surprisingly I've seen trainers and owners who will miss out on clues like these on their own horses improvements or patterns of impending failure. Dutrow is still scratching his head on BB,yet I wasnt as mystified.

[ebcorde]

I would better entertain your idea if you can provide a sample where the horses did not run against each other.

[Light]

I would better entertain your idea if you can provide a sample where the horses did not run against each other.

Fair enough

Aqu 1/10/09 5th race 6f $10kn1y 4^

Favorite was Lights of Broadway
Last 2 races looked like this at this distance:

E2 Lp SR Class Date

105 86 95 10kn1y 12/28/08

91 79 78 16kn1y 11/28/08

Winner; The Vin Man @ 8-1 looked like this last race at this distance:

E2 Lp SR Class Date

80 89 77 30K 12/31/08

Projecting the same regression on the fav due to staying at the same class,he ran 5th with a SR of 80. Again in the same range as his 2 back race with better. Exactly what happened to the favorite in my previous example.

The Vin Man won with an 85 however he steadied in that race and I would upgrade his real SR to at least an 87. A 10 point increase due to the class drop. Again almost exactly as the upset winner of my previous example.This scenario of progression/regression based on dropping for the first time vs staying at the same class for the 2nd time has a very similar outcome to my previous one despite the horses not sharing a similar race.

[Overlay]

Data that I've seen has shown a smooth flow of impact values as a result of ranking a field (1/2/3/Front Half/Rear Half) according to the average Beyer from those races out of each horse's last three starts that were run on the same surface as today's race. (If none of the horse's last three starts were on today's surface, use all three to calculate the average.) If the horse has less than three races in its past performances, use the races from its one or two starts that were run on today's surface to calculate the average, or (again), if none of the horse's races were on today's surface, use all of the available races to calculate it. There are no conditions of class or distance.
Overall impact values were as follows:

1 2.22
2 1.50
3 1.17
Front Half .93
Rear Half .57

With relation to class, doesn't the Form periodically publish a chart showing "typical" Beyer figures or figure ranges associated with ascending class levels?

[Light]

I think using an average of the last 3 is more inaccurate than a single PL.That method is totally oblivious of the class that the horse is in for today vs what it used to run for. For example, you have a horse dropping in class. Three back was for $25k,two back for $12.5 K, last for $10K and today is running for $7.5K. If the horse performed well at the higher classes, then obviously the horse cant run those speed figs for the $25k or $12.5K class any more or the trainer would place him in that class. If the figs were poor,they will muddle what the horse can do with a wake up call on a class drop. So why would you use data that will give you a figure that is totally bogus one way or the other. The closest accurate fig would be to just use his last PL,note any trouble and project his rating. Occasionally you get horses that can fit the scenario of 3 races at the same class level, and distance all within a reasonable time frame. But in reality,in most cases,that is impossible.

[CincyHorseplayer]

Top Beyers and favorites are a fairly common occurrence.They win less than 40% of the time.Bet against them unless they make it impossible on you.That's how I go about it and even though the odds might not reflect it I believe in Mark Cramer's "Low Priced Overlay".I bet Tampa Saturday and Hit 3-1,5/2,9/5,and 19-1.3 favorites lost in those races.Top Beyers are targets IMO.

[jotb]

Reminds me of a race about a year ago at Aqu for 6f 4^ clm $10K on Feb 18 2008.

The favorite's figs:

E2 LP SR Class Date

91 105 100 10K Feb 1

90 92 88 14K Dec 30

Now look at the winner's fig @6-1

E2 LP SR Class Date

84 93 84 14K Dec 30

These 2 horses came out of the same Dec 30th race. If the favorite increased his SR by 12 points with the drop then the 6-1 had to be projected to increase his SR by 12 points as well with the same class drop. Instead what most people are handicapping by is that the favorite ran a 100 and this 6-1 shot ran an 84 for slighly higher and probably will still come up short to the favorite.

My reasoning was that the drop is the wake up call.How "hard" a horse ran for the higher tag is not that big of a deal to me. I watch replays of horses losing under strong urging,dont bet them cause I think they're "used up" and watch them romp next out.Why? Cause today there is something new and a class drop certainly is a wake up call. BTW the eventual winner was obviously well rested since its last start with some really nice workouts. Your concept of "hard race" has to incorporate the days since raced.

Since the favorite already dropped,he had his wake up call and stuck at the same level would only regress since he needed the previous drop in the first place and he's not going anywhere today.Sure enough he finished 3rd regressing to his previous SR and was given an 89. The 6-1 was definetly going to wake up since we have direct proof from another horse that did that out of the same race (the fav).When you combine the improvement expected for the 6-1 shot and the regression for the favorite,the 6-1 shot comes out on top. The 6-1 shot did improve exactly 12 points and ran a 96 in victory. This was my reasoning in real life and I did cash on this one.

I'm sure you are aware of most of what I said. But I still find it fascinating how much horses do improve with a class drop. I offer this explanation as one possible answer. I think the main answer varies with horses individualistic idiosyncrosies and thats why this game is so challenging. Even their trainers and owners dont really know 100% of the time how their horses will run. But where there is something concrete offered in the form as in the above example,I'll take it. And surprisingly I've seen trainers and owners who will miss out on clues like these on their own horses improvements or patterns of impending failure. Dutrow is still scratching his head on BB,yet I wasnt as mystified.

There were 3 horses coming out of the Dec 30th race. Specchio, Golden Man, and Ojibway. Specchio finished 6th on Dec 30th behind Golden Man (4th as the favorite) and Ojibway (3rd at 6-1). Come Feb 18th 2008 Specchio comes back to win at 6-1 but don't forget the horse was now with Chris Englehart/Maggie Moss instead of David Donk. The day Specchio was beat (Dec 30th) the track was labeled Good. Specchio never ran well on any type of off track in his career. Ojibway finished 3rd on Dec 30th (14k) then he come back to run on Jan 13th for 10k and finished 7th. That day Asmussen took the horse for Maggie Moss for 10k. Ojibway comes back to run 2nd on Feb 18 behind Specchio. The ownership changed from Maggie Moss to Chrome Cowboy Racing Stable. Golden Man was the favorite for the Dec 30th and Feb 18th races. Mike Hushion lost the horse on Dec 30th for 14k to Gregory Diprima/Repole Stable. Golden Man comes back on Feb 1st to run 2nd on a sloppy track in a 10k as the 2nd choice beating the favorite Red Crusader who also ran in the Feb 18th race (5th as the 2nd choice). Specchio was 3 lengths behind Goden Man and Ojibway in the Dec 30th race at 17-1. I don't think Specchio wins just because he dropped and rested. Track conditions and trainer changes can make the difference as well.

Joe

[Light]

I don't think Specchio wins just because he dropped and rested. Track conditions and trainer changes can make the difference as well.
Joe

That's true but it does not disprove my theory. I already gave another similar example to Specchio's with The Vin Man where trainer changes or track conditions did not play a part.

One note about Ojibway. He was the horse I was most afraid of beating Specchio in that race due to the claim.He ran second. But you noted that he dropped from $14k to $10k and ran worse which may look like a knock on my theory about droppers. But he had been at the $10k level 4 back. So when I talk about this wake up call,the horse must be dropping to this new low level for the first time.Also the fact that he won 3 back for $14k and lost for $10k in his last was not a good sign. But the claim was.

Specchio was shipped to Mnr and FL after his victory and won both races for higher claiming tags there. Then he tapered off and lost his last 2 at Mnr for less than he previously won there and has been on the shelf since November. At one point he dropped at Mnr from $30k back down to $10K but I would not have considered him a true dropper at that point since he had been at the $10k level previously.

[jotb]

Fair enough

Aqu 1/10/09 5th race 6f $10kn1y 4^

Favorite was Lights of Broadway
Last 2 races looked like this at this distance:

E2 Lp SR Class Date

105 86 95 10kn1y 12/28/08

91 79 78 16kn1y 11/28/08

Winner; The Vin Man @ 8-1 looked like this last race at this distance:

E2 Lp SR Class Date

80 89 77 30K 12/31/08

Projecting the same regression on the fav due to staying at the same class,he ran 5th with a SR of 80. Again in the same range as his 2 back race with better. Exactly what happened to the favorite in my previous example.

The Vin Man won with an 85 however he steadied in that race and I would upgrade his real SR to at least an 87. A 10 point increase due to the class drop. Again almost exactly as the upset winner of my previous example.This scenario of progression/regression based on dropping for the first time vs staying at the same class for the 2nd time has a very similar outcome to my previous one despite the horses not sharing a similar race.

Yes, The Vin Man came out of the 30k with a SR or 77 but coming out of the same race was Mr. Meso who outfinished The Vin Man in the 30k. I guess with your theory about the drops, you have to include Mr. Meso in the mix. Don't forget Meso was coming off a long layoff when he ran for 30k. He drops to the 10k off a 10 day layoff. Meso ran upclose early in the 10k and from the result charts of that day, it seems like most of the winner's were coming from off the pace. Meso is bumped up to 16k (12 day's later) in his next start finishing 6th and then drops back to the 10k 8 day's later spliting the field. I guess because The Vin Man has a trouble trip in the 30k you gave him the advantage over Meso. By the way, I think Lights Of Broadway, speed figure for the 12-28 race was inflated and the track was labeled good that day.

Joe

[classhandicapper]

I do a similar thing, but by using the Class and Finish Position of the horses recent races relative to the size of the fields. I use the Beyer speed figure PAR for the class and some stats I accumulated that equalize various finishing positions in each class with each other. I don't calculate a numeric rating. I just rate races as Strong, Average and Weak depending on the quality of the horses and Deep, Average and Light depending on the number of contenders relative to average.

I am working on something new that I think will create a class consistency rating that's superior to anything that I have seen published. Unfortunately I don't have a database and the programming skills needed to both develop and test my ideas quickly. I am going to have to do it via trial and error over time. It's funny because I've been trying to develop this system for about 20 years but could never solve one of the problems. Then one day last week a light bulb went off. I feel like I have been an idiot for 20 years because a reasonable solution was obvious.

[Light]

jotb

Mr Meso ran a credible race finishing less than a length behind The Vin Man. But I still would have had to stick with The Vin Man. Its not automatic for me to bet every dropper. Some are just hopeless.

Mr Meso's form was a bit much to take. A 9YO, he had beaten better and needs a lone lead. The Vin Man was pretty much at the same level in his last 7 starts,until R. Persuad claimed him for $16k 4 back,but immediately hiked him back to $30k for 3 starts before the big drop. His last 2 races were omens (and conditioners) to an impending good race especially 2 back where he posted a 100 E2 fig in a route race.

The Vin Man certainly needed the drop cause he had run the same big E2 fig in a $30K route before and went into a sprint for $25k and was up the track.Dropping to $10k was the right thing to do.But unless they do an Ainslie big win,I dont usually like these types in their next start cause they usually bounce or as I term it:bottom out.

[Overlay]

I think using an average of the last 3 is more inaccurate than a single PL.That method is totally oblivious of the class that the horse is in for today vs what it used to run for. For example, you have a horse dropping in class. Three back was for $25k,two back for $12.5 K, last for $10K and today is running for $7.5K. If the horse performed well at the higher classes, then obviously the horse cant run those speed figs for the $25k or $12.5K class any more or the trainer would place him in that class. If the figs were poor,they will muddle what the horse can do with a wake up call on a class drop. So why would you use data that will give you a figure that is totally bogus one way or the other. The closest accurate fig would be to just use his last PL,note any trouble and project his rating. Occasionally you get horses that can fit the scenario of 3 races at the same class level, and distance all within a reasonable time frame. But in reality,in most cases,that is impossible.

I understand your points, and can't say that I disagree with them, or that there aren't other effective (or more effective) ways to project speed or pace. I'm just quoting from the results of a study that used what I would consider a sufficient number and variety of races from which to draw valid conclusions, and stating what the data in that study showed. Of course, it would be possible to further refine the data by adding filtering considerations such as class, distance, or condition, but the impact values that I cited were for the sample as a whole, without regard for those elements.

[proximity]

* i'd just consider the last half dozen or so races of all the horses in the field, going back no more than say a year or so.

* then i'd make projections based on the top 3-4 horses like bob p said above.

* the winner of the race: what is his top in the last half dozen or so races? does he show a recent win? if so what figure did he run in that win? this stuff hints at his projection.

* the second place horse: does he show a recent 2nd place finish? if so, what kind of figure did he run in that effort?.....

*overall i'd stay away from averaging in favor of the above method. there will be times though that you don't have alot to work with and you'll just have to use all of your handicapping powers, experience, and sixth sense to make an educated guess.

* conclusion: do the best you can to make projections for 3-4 of the race's top finishers and then (using beyer's beaten length chart) combine these projections to give you an overall class figure for the race.

[Light]

Although I may seem to be preaching here,I'm actually learning a little something talking about this myself. I always wondered exactly how many points to adjust a class dropper and it seems 10 points (Bris numbers) is about right.(Remains to be seen). So I just scanned the free Brisnet races today and found one race that fit the scenario I talked about in my previous 2 examples: A favorite who earned a big fig off a drop and a dropper who is about to earn a bigger fig. Race 2 Tampa 6f $10kn3L 4^F&M. Follow along.

http://www.brisnet.com/bris_link/pdfs/winstar_farm_llc_139872.pdf

1) 77SR 10kn3L 7-1

2) 58SR 16Kn3L 38-1 Adjusted to a 68 SR for the drop

3) 70SR 16Kn3L 4-1 Adjusted to an 80 SR for the drop

4) 80SR 10Kn3L 1-1

5) 75SR 8Kn2L 9/2

6) 82SR 10K 7-1 off since Dec. Tr 0-16 w/shippers.

7) 70SR 8Kn3L

Both the 4 and 6 achieved their figs after dropping from a higher class and are still at this level. The 6 looks like she needs a race, which leaves the 4 bet to favoritism because of the big fig from the drop. And again if we project a little regression on her part for staying at the same class level and add the 10 points to the 3's SR for her drop we come up with the $10 winner.

[andicap]

So you make an approximate two-second adjustment in the dropper's final time? Do I have the BRIS 10-point adjustment set correctly? Isn't BRIS about one point for each fifth of a second/one length on final time?

If so, that would amount to about 16 to 24 points on the Beyer scale, I believe, depending on the distance since at longer routes 1.6 points is one-length and at short sprints its 2.4 points or thereabouts.

Am I correct here?

[cj]

I believe you are off on the BRIS scale. I think the value of a length is about 10 divided by the furlongs, so at 5f it would be 2 points, at 10 furlongs it would be 1 point.

I also think you are underestimating the Beyer adjustment. 1.8 is better if you want a basic "route" number.

[bobphilo]

So you make an approximate two-second adjustment in the dropper's final time? Do I have the BRIS 10-point adjustment set correctly? Isn't BRIS about one point for each fifth of a second/one length on final time?

If so, that would amount to about 16 to 24 points on the Beyer scale, I believe, depending on the distance since at longer routes 1.6 points is one-length and at short sprints its 2.4 points or thereabouts.

Am I correct here?

You can skip the conversion from Bris figure, to time, to Beyer by just converting from Bris to Beyer directly. My studies indicate that a change in 10 Bris figures is about equal to 15 in Beyers.

I am a bit skeptical, however, about adding so many points to a horse's figure off any class drop. Many class drops are negative signs. I'd only add in cases where it fits the trainer's M.O. for an improved performance.

Bob

[Light]

So you make an approximate two-second adjustment in the dropper's final time? Do I have the BRIS 10-point adjustment set correctly? Isn't BRIS about one point for each fifth of a second/one length on final time?
Am I correct here?

From Bris website: On the BRIS Speed Ratings scale one length is approximately equal to 1 1/2 pts in sprints and approximately 1 point in routes.

So 10 Bris points equals a little over 6 lengths in sprints.That may seem like a alot for an improvement but look at the lengths these 3 winners had lost by prior to their drop.

Specchio had lost by 5 1/2 lengths

The Vin Man had lost by 11 lengths

Pyriteville had lost by 8 3/4 lengths

There is no way you are going to get any of these horses into your contender list using a moderate SR improvement projection.(Using pace figs,they can become contenders,but not on SR's). Anyrate,these are the type of droppers I like to see. A horse that lost by a length or 2 and drops is most likely a short price and/or a false favorite and I would be hesitant with that type. He's also unlikely to improve as much as a horse who lost by 6 lengths because there usually is not much room left for a form reversal nor can it be categorized as a form reversal if he wins. I dont use Beyers,so I dont know the conversions there.

[classhandicapper]

You can skip the conversion from Bris figure, to time, to Beyer by just converting from Bris to Beyer directly. My studies indicate that a change in 10 Bris figures is about equal to 15 in Beyers.

I am a bit skeptical, however, about adding so many points to a horse's figure off any class drop. Many class drops are negative signs. I'd only add in cases where it fits the trainer's M.O. for an improved performance.

Bob

A big part of the improvement is related to pace. Just as the average winning final time tends to get faster as you move up the class ladder, so does the average pace. Horses that were competitive (at least early) against higher quality horses typically were either battling or chasing a faster pace than they are likely to face against cheaper horses. That allows them to improve their final time figure when the drop in against easier horses. I think pace accounts for the majority of fluctuations in figures we see when horses take big jumps or drops in class, but there is more to it than that.

However, I agree that you have to be careful when evaluating claimers because sometimes the drop in class is a negative sign about the horse's condition.

[denneyrl]

The Vin Man running now, the 2nd at Aquduct

[bobphilo]

A big part of the improvement is related to pace. Just as the average winning final time tends to get faster as you move up the class ladder, so does the average pace. Horses that were competitive (at least early) against higher quality horses typically were either battling or chasing a faster pace than they are likely to face against cheaper horses. That allows them to improve their final time figure when the drop in against easier horses. I think pace accounts for the majority of fluctuations in figures we see when horses take big jumps or drops in class, but there is more to it than that.

However, I agree that you have to be careful when evaluating claimers because sometimes the drop in class is a negative sign about the horse's condition.

I agree on the pace issue. I believe the main reason that some horses do so poorly when going up in class, aside from the higher final time pars, is that that they are contending with faster paces. I know that many like to think of class as an intangible but most of the effects of running in a different class are really mainly matters of pace. Very often one can predict a big improvement or decline by going down or up in class by looking at the pace scenarios. This and trainer intentions make up most of the class effects.

On this basis, I would be reluctant to project big increases in speed figures based purely on class drop unless I could expect a favorable change in pace scenarios.


Bob

[fmolf]

Top Beyers and favorites are a fairly common occurrence.They win less than 40% of the time.Bet against them unless they make it impossible on you.That's how I go about it and even though the odds might not reflect it I believe in Mark Cramer's "Low Priced Overlay".I bet Tampa Saturday and Hit 3-1,5/2,9/5,and 19-1.3 favorites lost in those races.Top Beyers are targets IMO.


when i find an unbeatable favorite an their are some... i use them in the one slot in exactas with the overlayed two or three combos beneath it