Useless Beyer Figure

[Steve R]

2015 Derby Speed Figures (http://www.chef-de-race.com/images/2015 Derby Figs.jpg)

For PF, lower is better. For BRIS and RPR, higher is better.

[Capper Al]

2015 Derby Speed Figures (http://www.chef-de-race.com/images/2015 Derby Figs.jpg)

For PF, lower is better. For BRIS and RPR, higher is better.

I'm not following. One or two races don't make a statistic.

[Tom]

Useless when do not know how to use them.

[rgustafson]

2015 Derby Speed Figures (http://www.chef-de-race.com/images/2015 Derby Figs.jpg)

For PF, lower is better. For BRIS and RPR, higher is better.

Beyer figure for Derby revised to a 95. Does that make it more or less useless?

[Saratoga_Mike]

Beyer figure for Derby revised to a 95. Does that make it more or less useless?

Did they really revise the number already? I actually agree with the lower number, but what changed?

[DeltaLover]

2015 Derby Speed Figures (http://www.chef-de-race.com/images/2015 Derby Figs.jpg)

For PF, lower is better. For BRIS and RPR, higher is better.

Before you conclude about whether Beyer figures been 'useless', can you please explain how you use them and what exactly you expect from them?

[ArlJim78]

Beyer figure for Derby revised to a 95. Does that make it more or less useless?
Do you have a link for that, or are you just speaking hypothetically?
I don't see any mention anywhere of the figure being revised.

[classhandicapper]

I think the OP was pointing out that American Pharoah has a higher Beyer figure than CA Chrome and he doesn't agree with the figure for last year.

The complications for making that figure were widely discussed last year.

[keithw84]

I'm not an expert on figure making by any means, but it seems when numbers get adjusted or are controversial, it's because we want to believe something different than what the evidence suggests.

"How could California Chrome only run a 97? Surely, there must be a variable we missed and the number should be different."

Every race is different. Just because a G1 winner ran a 95 and an Allowance winner ran a 100 doesn't mean the G1 horse would lose head to head. It just means the G1 horse only had to run a 95 to win. It doesn't say much for the competition, but it also doesn't mean the horse that ran a 95 isn't capable of much better.

[Robert Goren]

I'm not an expert on figure making by any means, but it seems when numbers get adjusted or are controversial, it's because we want to believe something different than what the evidence suggests.

"How could California Chrome only run a 97? Surely, there must be a variable we missed and the number should be different."

Every race is different. Just because a G1 winner ran a 95 and an Allowance winner ran a 100 doesn't mean the G1 horse would lose head to head. It just means the G1 horse only had to run a 95 to win. It doesn't say much for the competition, but it also doesn't mean the horse that ran a 95 isn't capable of much better.I do not see how "Pharaoh" could have run much faster. It looked to me like he was all out.

[classhandicapper]

"Numbers" are impacted by dozens of things that are sometimes hard to quantify. Different figure makers will come to different subjective conclusions and include different things. That's why there is a wide variance.

[rgustafson]

Do you have a link for that, or are you just speaking hypothetically?
I don't see any mention anywhere of the figure being revised.

Sorry, my mistake, but I thought it was changed on the Derbytrail website when I checked earlier this morning. Looking at it now it is listed at 105. :blush:

[ArlJim78]

Sorry, my mistake, but I thought it was changed on the Derbytrail website when I checked earlier this morning. Looking at it now it is listed at 105. :blush:
Yeah I saw the same thing and was wondering if you got it from the same place I did. I looked high and low and could find no other mention of a 95 Beyer so I guess it was just a typo by Byk.

[Capper Al]

Correcting numbers is necessary given all the variables and variants put into making the numbers. It's not all science with Beyers. There is some art in it. It could have been the wind?

[Tom]

Now don't tart that again! :lol:

Does anyone know if they watered the track during the 8 hours between the Derby and the last dirt race? :rolleyes:

[boys at tosconova]

i saw a lot of numbers and figures this year. more than any other year.

i like when the new rogue numerical guy shows up with color charts. if we're going to make a mockery of such things,..i say go all out

[Steve R]

I'm not following. One or two races don't make a statistic.

Those are the respective Derby figs for four figure makers. Three of them have California Chrome's Derby as good or better than American Pharoah's. Beyer says American Pharoah's Derby was 5 1/2 lengths better than California Chrome's. Also, if you consider the annual par times published by Dave Schwartz, the other two-turn races on Derby day this year were significantly faster than par while the other two-turn races on Derby day in 2014 were slower than par. This actually makes California Chrome's adjusted time faster than American Pharoah's.

In addition, there have been 24 Derbies run on a fast track since 1983. Eleven winners ran a faster first half mile than did American Pharoah. Ten of those also ran a faster last half mile. Eight winners ran a faster six furlongs. Every one of those also ran a faster last half mile.

The point is that the BSF had to be based on a projection based on opinion and not on physics. I do agree with Beyer to the extent that this is a very good crop of three-year-olds. In my view it is the fastest since 2006 (Barbaro, Lawyer Ron, Brother Derek, etc) with the caveat that it is the fastest up to nine furlongs. IMO, none of the horses in this year's Derby actually stayed ten furlongs in the traditional classic sense.

[RXB]

On the whole, I'd trust Beyer figures over BRIS or RPR.

I didn't really pay much attention to the Derby day card or the times until just now. Seems to me that the dirt variant needs to be split for R1-R5 compared to the later races. I doubt that BRIS or RPR did this; I'm guessing that Beyer did.

Also can't help but notice that in R12 (the 8.5f race after the Derby), the Equibase chart claims that the final sixteenth was covered in 5.64 seconds-- essentially impossible.

[RXB]

Mystery solved for R12; they've incorrectly entered the mile time a second slower than what it really was. So the final sixteenth is really 6.64.

[cj]

Mystery solved for R12; they've incorrectly entered the mile time a second slower than what it really was. So the final sixteenth is really 6.64.

This happens a lot more often than people realize at all calls, including the finish. I mean, this happened on the biggest day of racing in the country. What do you think happens at Happy Park on a Tuesday? (just fill in the track, not trying to pick on any one)

[RXB]

I guess Equibase needs to pay more to get reliable data-entry workers.

What did you get for a final-time figure for the Derby? Did you split the variant after R5?

[cj]

I guess Equibase needs to pay more to get reliable data-entry workers.

What did you get for a final-time figure for the Derby? Did you split the variant after R5?

I don't have the answer for Equibase, I just know I find a lot of errors, mostly in transcribing. Seems like in 2015 there should be a way to have the tote times entered automatically, but what do I know?

Yes, clear dividing line for me after Race 5. The last two races are a bit tougher. Going to look at them a lot deeper today and tomorrow.

[Thebigguy]

Yeah I saw the same thing and was wondering if you got it from the same place I did. I looked high and low and could find no other mention of a 95 Beyer so I guess it was just a typo by Byk.

brks a saint. control yourself chi town smuggler

[Capper Al]

Let's see what happens at the Preakness and if there is another Beyer adjustment before then.

[Tom]

i saw a lot of numbers and figures this year. more than any other year.

i like when the new rogue numerical guy shows up with color charts. if we're going to make a mockery of such things,..i say go all out

Who would that guy be?

[ArlJim78]

brks a saint. control yourself chi town smuggler
saints can make a typo no? just stating a fact.

[rastajenk]

It doesn't really matter much what number Chrome earned last year, since he's become an international runner and isn't likely to face any of this year's bunch anytime soon. It's value as a handicapping tool this year has decreased greatly. All that matters is whether Pharoah's 105 (and all that follows from it) is reliable going forward. I suspect it will be.

In these troubled times, do we really need another trash-Beyer-Speed-Ratings thread?

[mountainman]

Correcting numbers is necessary given all the variables and variants put into making the numbers. It's not all science with Beyers. There is some art in it. It could have been the wind?

Like they have somebody there measuring the wind??? Or to calculate the effect of between-race harrowing??? Correcting numbers is necessary to avoid non-sequiturs, and to foster the false impression that speed figs invariably mirror performance. And it's a practice I HEARTILY disapprove of. Just serve it to me straight, without self-serving 'adjustments' intended to flatter the fig-making process-or the race in question.

Should the weather change , or it be OBVIOUS that the track got faster or slower throughout the card, then split the variant.

I do respect your posts and obvious knowledge, sir, and this is simply my opinion. But if 2:03 is a legit 105 on the beyer scale, I'm Frank Sinatra.

[f2tornado]

The time was 3/5 second faster than Chrome with a very similar track so I'd expect a better number than last year. That said, Orb was even faster in the slop and only got 104. Super Saver was pretty slow and also got a 104. Beyer might call it a science but it's effectively just an error prone educated guess. I still like looking at speed figures for a quick and dirty handicapping tool but we all know they are not the be all end all.

[Tom]

Raw times have no correlation to the Beyer - the Beyer is adjusted for track speed. A 2.09 could end up as a 105 if the track was that slow. A 2.03 gets a 108 figure on the chart.

[f2tornado]

Raw times have no correlation to the Beyer - the Beyer is adjusted for track speed. A 2.09 could end up as a 105 if the track was that slow. A 2.03 gets a 108 figure on the chart.

Not directly correlated, true (see 2015 FL Derby for a prime example), but definitely a strong impact.

We take all the times on a specific race card and go through them race by race. Back in the day, it was all done on paper. Now, we have a chart-like computerized printout that shows the last five Beyers for each horse in a race, in addition to that race’s par time, what that class of horses has traditionally done over a period of time.

Then, we note the difference between what was expected (par times, recent Beyers) and the reality. For each race, we assign a difference. Then, we average those differences over an entire card and assign a variant. We take the variant and either add (if the times were slower than expected) or subtract (if the times were faster than expected) from the raw times for each race. Then, we have a Beyer figure for each race on every card in America.

The BSF team can call it science but it's really just an educated guess using mathematics based on expectations. Ample room for error but better than nothing.

[Capper Al]

Like they have somebody there measuring the wind??? Or to calculate the effect of between-race harrowing??? Correcting numbers is necessary to avoid non-sequiturs, and to foster the false impression that speed figs invariably mirror performance. And it's a practice I HEARTILY disapprove of. Just serve it to me straight, without self-serving 'adjustments' intended to flatter the fig-making process-or the race in question.

Should the weather change , or it be OBVIOUS that the track got faster or slower throughout the card, then split the variant.

I do respect your posts and obvious knowledge, sir, and this is simply my opinion. But if 2:03 is a legit 105 on the beyer scale, I'm Frank Sinatra.

The wind references was a joke to others who claim to use wind.

I have been reviewing actual times in the Derby PPs. Take a look at horse #6 (Mubtaahij) his second race, a 1:39.4 for 7f on the grass. I show that race was run about 17 seconds slower than a normal 7f turf race. It caught my eye because I rated at 1.8 on a scale where 100 is normal. What formula could possibly calculate this race?

[Saratoga_Mike]

The wind references was a joke to others who claim to use wind.

I have been reviewing actual times in the Derby PPs. Take a look at horse #6 (Mubtaahij) his second race, a 1:39.4 for 7f on the grass. I show that race was run about 17 seconds slower than a normal 7f turf race. It caught my eye because I rated at 1.8 on a scale where 100 is normal. What formula could possibly calculate this race?

Al, his second race was at a mile on the turf.

[Tom]

Oh, that formula! :D

[mountainman]

A good projection variant applied to the raw time would suffice with me. Beyond that, I'd just rather they didn't tinker.

[RXB]

To me, there is an obvious gap in track speed on Derby day for the early dirt races and what happened thereafter on the dirt.

Race 1 was N2X 3-up males; Race 12 was N1X 3YO males. Both 8.5 furlongs. In both races the winner won by several lengths. The final time for R1 was 1.51 seconds faster than R12; that is double the difference that I'd expect over the long haul for those class conditions.

Race 2 was N1X 3YO males; Race 7 a Grade 1 for older females; Race 9 a Grade 2 for older males; R13 for MSW 3-up males. All at 7f. The final time was faster for the N1X than for either of those two major older stakes (which were fairly close together in final time, as would be expected for those class/gender levels). R2 was also 1.26 seconds faster R13, where normally I'd expect a difference of about 0.5 between those conditions.

Race 5 was a G3 mile restricted to 3YO's. The raw final time for R5 on a comparative scale is significantly faster than the aforementioned G1 older female and G2 older male stakes at 7f, which seems unlikely under normal circumstances.

And then there is R11, the Ky Derby. A bunch of horses coming into the race with Beyer figures in the 104-110 range, and suddenly none of them can hit 100? That's the case if you believe that the track didn't slow down in mid-card.

I would say that the main track was close to a tenth of a second slower per furlong for Race 7 and beyond compared to Races 1, 2, 3, 5.

[cj]

To me, there is an obvious gap in track speed on Derby day for the early dirt races and what happened thereafter on the dirt.

Race 1 was N2X 3-up males; Race 12 was N1X 3YO males. Both 8.5 furlongs. In both races the winner won by several lengths. The final time for R1 was 1.51 seconds faster than R12; that is double the difference that I'd expect over the long haul for those class conditions.

Race 2 was N1X 3YO males; Race 7 a Grade 1 for older females; Race 9 a Grade 2 for older males; R13 for MSW 3-up males. All at 7f. The final time was faster for the N1X than for either of those two major older stakes (which were fairly close together in final time, as would be expected for those class/gender levels). R2 was also 1.26 seconds faster R13, where normally I'd expect a difference of about 0.5 between those conditions.

Race 5 was a G3 mile restricted to 3YO's. The raw final time for R5 on a comparative scale is significantly faster than the aforementioned G1 older female and G2 older male stakes at 7f, which seems unlikely under normal circumstances.

And then there is R11, the Ky Derby. A bunch of horses coming into the race with Beyer figures in the 104-110 range, and suddenly none of them can hit 100? That's the case if you believe that the track didn't slow down in mid-card.

I would say that the main track was close to a tenth of a second slower per furlong for Race 7 and beyond compared to Races 1, 2, 3, 5.

I looked at my worksheet after the races and concluded in about two seconds that the dirt track changed noticeably after race 5.

[RXB]

I looked at my worksheet after the races and concluded in about two seconds that the dirt track changed noticeably after race 5.

I think it took me about 2.5 seconds. My mental surface is a little slower than yours.

[Capper Al]

Al, his second race was at a mile on the turf.

Second from the bottom on October 4?

[Saratoga_Mike]

Second from the bottom on October 4?

Correct, it was listed as 7 furs in the DRF PPs. Your PPs must have the wrong distance.

[Capper Al]

Correct, it was listed as 7 furs in the DRF PPs. Your PPs must have the wrong distance.

Saratoga, BRIS had 7f. A mile would make sense for the time. Do you have another PP provider saying it was a mile?

[cj]

Saratoga, BRIS had 7f. A mile would make sense for the time. Do you have another PP provider saying it was a mile?

It was a mile, here is his Timeform page, sans figures:

https://www.timeform.com/racing/Ledger/HorseLedger?HorseId=000000386690

[Capper Al]

It was a mile, here is his Timeform page, sans figures:

https://www.timeform.com/racing/Ledger/HorseLedger?HorseId=000000386690

Thanks. Interesting. Apparently both DRF and BRIS had it wrong. The time is a mile time.

[Steve R]

My results for the day based on 2014 CD pars adjusted for age, sex and time of year.

1-6f
2-3f
4-2s
5-1f
7-3s
9-4s
11-7s
12-4f
13-1s

Average = 0.3s
StDev = +/-4.2

In race 1 the winner went WTW off very fast fractions and won by almost 5.
In race 11 the time was 1 2/5ths slower than the Derby average over the last 10 years.
Both significantly exceed the standard deviation.
Eliminate the high and the low, plot the race number vs the variant and you get a straight line, virtually flat.
Ergo, no measurable change in overall track speed for the day.

[cj]

My results for the day based on 2014 CD pars adjusted for age, sex and time of year.

1-6f
2-3f
4-2s
5-1f
7-3s
9-4s
11-7s
12-4f
13-1s

Average = 0.3s
StDev = +/-4.2

In race 1 the winner went WTW off very fast fractions and won by almost 5.
In race 11 the time was 1 2/5ths slower than the Derby average over the last 10 years.
Both significantly exceed the standard deviation.
Eliminate the high and the low, plot the race number vs the variant and you get a straight line, virtually flat.
Ergo, no measurable change in overall track speed for the day.

So you have the winner of Race 1 rated higher than the Derby winner? R12?

[Steve R]

So you have the winner of Race 1 rated higher than the Derby winner? R12?
I have no idea what you're talking about. Those are race variants, not race ratings.

Race 1 was 1 1/5th faster than par for a 3+ OCL 62.5 at 8 1/2f.
Race 11 was 1 2/5ths slower than par for the Derby.
Race 12 was 1/5th slower than par for a 3yo MSW at 8 1/2f.

[RXB]

I have no idea what you're talking about. Those are race variants, not race ratings.

Race 1 was 1 1/5th faster than par for a 3+ OCL 62.5 at 8 1/2f.
Race 11 was 1 2/5ths slower than par for the Derby.
Race 12 was 1/5th slower than par for a 3yo MSW at 8 1/2f.

R12 was N1X, not MSW.

[cj]

I have no idea what you're talking about. Those are race variants, not race ratings.

Race 1 was 1 1/5th faster than par for a 3+ OCL 62.5 at 8 1/2f.
Race 11 was 1 2/5ths slower than par for the Derby.
Race 12 was 1/5th slower than par for a 3yo MSW at 8 1/2f.

Yes, I understand. I obviously don't have access to your speed charts, but any of the ones I've seen would show R1 as faster than the Derby if track speed was considered consistent. Thus, I was asking about yours. I was off on R12, it is slower than the Derby. But do you have the winner of R1 rated higher than the Derby winner?

Example, Beyer chart raw ratings for the times of those races:

R1 113
R11 108
R12 100

I realize you do account for pace as well (I think), but I'm talking final time.

[Steve R]

R12 was N1X, not MSW.
Sorry. That was a transcription error. It should have been race 13 which was a 3+ MSW. Race 12 was a 3yo OCL 75, not a NW1.

[Saratoga_Mike]

Saratoga, BRIS had 7f. A mile would make sense for the time. Do you have another PP provider saying it was a mile?

Sorry Al, my first post was correct - the race was a mile, and that's what DRF had. Also, do a Google search for the race, and you'll find it was a mile. Sorry for the confusion.

[whodoyoulike]

It was a mile, here is his Timeform page, sans figures:

https://www.timeform.com/racing/Ledger/HorseLedger?HorseId=000000386690

This is disturbing for me after all these months the PP's weren't corrected. No one pointed it out to BRIS regarding their error. My concern also applies to all PP service providers.

[Steve R]

Yes, I understand. I obviously don't have access to your speed charts, but any of the ones I've seen would show R1 as faster than the Derby if track speed was considered consistent. Thus, I was asking about yours. I was off on R12, it is slower than the Derby. But do you have the winner of R1 rated higher than the Derby winner?

Example, Beyer chart raw ratings for the times of those races:

R1 113
R11 108
R12 100

I realize you do account for pace as well (I think), but I'm talking final time.
I believe those are the generic ratings as published in BEYER ON SPEED, etc. My raw numbers are track specific and change from year to year and my rating scale is different in that the lower numbers are better. And , as you noted, I incorporate pace into the final figure so the final rating has that component already built in. Using my methodology the raw figures come out to:

Race 1 -73
Race 11 -44
Race 12 -33

As you can tell, I don't think the BSF 105 is accurate. My adjusted Derby figure is -50 which, coincidentally, matches exactly the BRIS assignment of 100 based on a correlation of hundreds of my figures with BRIS figures for the same races. The correlation coefficient between my figures and the BRIS figures is an excellent 0.87. Interestingly, a similar correlation of my figures with BSFs affords a correlation coefficient of 0.88. In the case of this year's Derby I think Beyer may have relied too much on projection from shorter distances. My take on the Derby is that none of the horses were up to classic standards on the day. I saw this before in the 1995 Belmont when Thunder Gulch won in 2:32 and was assigned a BSF 101 coming off a Derby BSF of 108. My figure had it equivalent to something closer to a BSF in the mid-80s. That 2:32 was the slowest in 25 years and is still the second slowest since 1970. That's not to say that on a another day Thunder Gulch couldn't have won in 2:28 but on that day the race turned strategic, the pace was lethargic and that was the result. I don't see any justification for upgrading a performance not justified by the physical reality.

[classhandicapper]

A LOT of what Beyer published many years ago has been updated. He has changed some of the relationships between distances, especially if you look at it on a track by track basis.

I don't have time to track everything his group does, but I do track NYRA and occasional big cards elsewhere. So I've seen many of the changes he's made and can build it into my review process if I'm trying to figure out what he did on a specific day. You can't back into what he did using his published work as easily anymore.

[classhandicapper]

Correct, it was listed as 7 furs in the DRF PPs. Your PPs must have the wrong distance.

You must have been looking at a preliminary set of PPs because the final edition of both Classic PPs and Formulator have that race listed as a mile.

[cj]

I believe those are the generic ratings as published in BEYER ON SPEED, etc. My raw numbers are track specific and change from year to year and my rating scale is different in that the lower numbers are better. And , as you noted, I incorporate pace into the final figure so the final rating has that component already built in. Using my methodology the raw figures come out to:

Race 1 -73
Race 11 -44
Race 12 -33

As you can tell, I don't think the BSF 105 is accurate. My adjusted Derby figure is -50 which, coincidentally, matches exactly the BRIS assignment of 100 based on a correlation of hundreds of my figures with BRIS figures for the same races. The correlation coefficient between my figures and the BRIS figures is an excellent 0.87. Interestingly, a similar correlation of my figures with BSFs affords a correlation coefficient of 0.88. In the case of this year's Derby I think Beyer may have relied too much on projection from shorter distances. My take on the Derby is that none of the horses were up to classic standards on the day. I saw this before in the 1995 Belmont when Thunder Gulch won in 2:32 and was assigned a BSF 101 coming off a Derby BSF of 108. My figure had it equivalent to something closer to a BSF in the mid-80s. That 2:32 was the slowest in 25 years and is still the second slowest since 1970. That's not to say that on a another day Thunder Gulch couldn't have won in 2:28 but on that day the race turned strategic, the pace was lethargic and that was the result. I don't see any justification for upgrading a performance not justified by the physical reality.

I think that is a yes, you have the winner of the 1st not just faster, but substantially so. I don't agree, but I do like the boldness of it. What odds would you need to bet that horse in a match race against the Derby winner?

[cj]

You must have been looking at a preliminary set of PPs because the final edition of both Classic PPs and Formulator have that race listed as a mile.

He said that a few posts earlier.

[cj]

A LOT of what Beyer published many years ago has been updated. He has changed some of the relationships between distances, especially if you look at it on a track by track basis.

I don't have time to track everything his group does, but I do track NYRA and occasional big cards elsewhere. So I've seen many of the changes he's made and can build it into my review process if I'm trying to figure out what he did on a specific day. You can't back into what he did using his published work as easily anymore.

I'm aware of the updates. They aren't going to affect what I posted very much at all.

[Capper Al]

Here's what BRIS shows me:

04Oct14 Newmarket(GB) à 7f gd 1:39« ¨¨¬ Stk 12568 ¨¨ª

[Saratoga_Mike]

Here's what BRIS shows me:

04Oct14 Newmarket(GB) à 7f gd 1:39« ¨¨¬ Stk 12568 ¨¨ª

The race was a mile.

http://www.sportinglife.com/racing/results/04-10-2014/newmarket/result/643431/wild-duck-norfolk-woodland-retreat-ebf-stallions-maiden-stakes

[Steve R]

I think that is a yes, you have the winner of the 1st not just faster, but substantially so. I don't agree, but I do like the boldness of it. What odds would you need to bet that horse in a match race against the Derby winner?
So what you are saying is that on the day when the three-year-old Thunder Gulch won the Belmont in 2:32, despite the race being strategic, it would have been impossible for an older male running in a a high-end optional claiming race to have gotten a mile-and-a-half in, say, 2:31.

BTW, the winner of the OCL, Paganol, has a lifetime best BRIS rating of 102. The well-beaten favorite in that race, SW Irish You Well, has a lifetime best BRIS rating of 103 which, coincidentally, is American Pharoah's lifetime best BRIS rating. I'm not suggesting at all that Paganol is better than American Pharoah. What I am suggesting is that on a given day one horse may exceed his previous top performance and another may fail to reproduce his best. I think that is what may have happened on Saturday. I should reiterate that I believe this Derby was below standard regardless of how fast the field may have been at shorter distances. Even Beyer thought so, assigning a 105 against a par of 107. In fact, every figure I've seen has been below the Derby standard.

[RXB]

R1 vs. R12 - same distance. R1 was an older male N2X; R12 a 3YO N1X/OCL75K. R1 had a five-length winner, R12 a six-length winner. The difference of 1.51 seconds in final time far exceeds any expected difference between those classes.

R2 vs. R7, R9, R13 - same distance. R2 was a 3YO N1X/OCL75k, and the fourth finisher was beaten by two lengths which rarely signals an exceptionally fast race. Yet R2's final time was faster than one of the best sprinters in the country (Private Zone. who won by 4.5 lengths) in R7, and faster than the older G1 female sprinters in R9. And, 1.26 seconds faster than the 3-up MSW's in R13, which again far exceeds the expected difference between those two classes.

R5 vs. R11 - If one assumes that the track played evenly, then the G3 3YO Pat Day Mile was substantially faster on final time compared to the Ky Derby. The Beyer figure for the Derby would be in the 95-97 range, despite the fact that a bunch of horses went into the Derby with Beyers in the 100-110 range and the favoured horses with solid numbers performed well.

Whereas, if I assume that the dirt track was about a tenth per furlong slower from R7 onwards, things seem much more in line when making comparisons across the card.

[Tom]

This is disturbing for me after all these months the PP's weren't corrected. No one pointed it out to BRIS regarding their error. My concern also applies to all PP service providers.
Or, it was reported and they did not bother to fix it.

[whodoyoulike]

Or, it was reported and they did not bother to fix it.

I remember CJ has mentioned running across errors like this in other posts. Someone mentioned earlier that a preliminary DRF also indicated the 7f but, was corrected to 8f in a later version.

Is the error source Equibase since I think I once read they're the source of U.S. pp's?

I wonder how DRF was notified of the error.

I just reviewed the BRIS pp and since all of his races were outside the U.S. there were no BRIS data calculations made. But, I just noticed their "good" comments stated "Won last race (MEY 03/28 9.44f Dirt ft MdSpWt)" when the last race was won but for a Stk 2000000 Group G2 at Meydan(Uae).

Because of my OCD, I'm going to now suspect and will probably now question all foreign races.

[clocker7]

I've said it before (and the KD confirmed my opinion) that the 3yo class of 2015 is OK but not scintillating. Better than those of 2012 and 2013, but not as good as 2014.

Consistency by a handful of its betters is its hallmark. That being said, I also believe that they are better suited for 9f, and that performance at 12f will be underwhelming.

[depalma113]

Had the race been run earlier in the day, the final time would have been 1.5 seconds faster at roughly 2:01.50, which would have been the 5th fastest time in the last 20 years.

Far too many people are not taking into account how slow that track got later in the day.

[classhandicapper]

I'm aware of the updates. They aren't going to affect what I posted very much at all.

I wasn't referring to anything you said.

It was a generic statement.

People shouldn't try to figure out what Beyer did in terms of track variants using any of his published books or charts because he's made changes in the relationships from distance to distance since back then, especially at certain tracks.

The only thing you can do with confidence is look at races at the same distance.

[cj]

I wasn't referring to anything you said.

It was a generic statement.

People shouldn't try to figure out what Beyer did in terms of track variants using any of his published books or charts because he's made changes in the relationships from distance to distance since back then, especially at certain tracks.

The only thing you can do with confidence is look at races at the same distance.

Again though, a few days worth of results with Beyer figures and it isn't exactly tough to figure out the adjustments. Someone with a year of data in a database could do it very quickly.

[Tom]

Grab a Beyer speed chart and a download of the Winner's Books and you can figure it out.

But you need to do a few days because be breaks out more than a few races
Right now Belmont is a mess - looks like random variants - I know for sure not to use and Beyer numbers from the past weekend until they get fixed.

[classhandicapper]

Again though, a few days worth of results with Beyer figures and it isn't exactly tough to figure out the adjustments. Someone with a year of data in a database could do it very quickly.

I look at his figures for NYRA every day and it's no longer that easy. He does a lot of breaking out/splitting along logical lines that confuse what his actual charts say the true relationships are. As an outsider, you can't always tell what's a breakout from what's wrong with your old version chart from when he's making adjustments for other factors.

I still can't figure out some of the relationships he's using on turf and I have very good information.

[cj]

I look at his figures for NYRA every day and it's no longer that easy. He does a lot of breaking out/splitting along logical lines that confuse what his actual charts say the true relationships are. As an outsider, you can't always tell what's a breakout from what's wrong with your old version chart from when he's making adjustments for other factors.

I still can't figure out some of the relationships he's using on turf and I have very good information.

Belmont is pretty crazy right now, I'm not sure what is going on but the times haven't made a lot of sense on several occasions.

[Tom]

For Saturday, 5/2, at Belmont, I got for 8.5 Furlong races Beyer variants of S8, F7, F7, 0.

Class, try this -

For routes, subtract the Beyer number for 83 and divide by 2.

This will give you the Quirin equivalent. For the Derby, the Beyer Par is 103.
103 - 83 =20
20/2 = 10, add to 100 and the Quirin par is 110.
Now use the 10K par time to get your variant.

It should take out most of the Beyer variances you see, other than broken out races.

[classhandicapper]

For Saturday, 5/2, at Belmont, I got for 8.5 Furlong races Beyer variants of S8, F7, F7, 0.

Class, try this -

For routes, subtract the Beyer number for 83 and divide by 2.

This will give you the Quirin equivalent. For the Derby, the Beyer Par is 103.
103 - 83 =20
20/2 = 10, add to 100 and the Quirin par is 110.
Now use the 10K par time to get your variant.

It should take out most of the Beyer variances you see, other than broken out races.


Thanks Tom.

I have the information for all the dirt distances. It's the turf data that is a problem for me. My source is either out of date or his figures are all over the map. I think it may be both. I more or less decided that tracking the turf variants wasn't a good use of my time. So I just stopped.

5/2 at Belmont was comical.

This entire project may not be worth my time, but I'm searching for some correlations hoping to find a nugget that's not common knowledge. I'm trying to build a completely different model of figures. I know this will sound totally insane, but I learned something new making my own figures for mule races on the fair circuit last year. If it's applicable to thoroughbreds, I may have something useful.

[Tom]

Mule races, huh?
Did you use the Finger Lakes Speed Chart for that one? :D

[ubercapper]

Also can't help but notice that in R12 (the 8.5f race after the Derby), the Equibase chart claims that the final sixteenth was covered in 5.64 seconds-- essentially impossible.

This has been corrected.

http://www.equibase.com/static/chart/pdf/CD050215USA12.pdf


If you see an error in a chart please send email to feedback@equibase.com

Thanks.

[cj]

This has been corrected.

http://www.equibase.com/static/chart/pdf/CD050215USA12.pdf


If you see an error in a chart please send email to feedback@equibase.com

Thanks.

It is great that you guys accept the feedback and get the stuff fixed. I've done that many times, but not recently. I guess I'm selfish and just fix it myself and build it into my numbers.

That said, this stuff would be very easy to catch with simplistic software that could review each days times for abnormalities. I see fat fingered fractions all the time that I am pretty sure are wrong, watch replays, and usually I am right. It wouldn't take very complex algorithms at all to find these errors. What is stopping you guys?

[Blenheim]

I believe those are the generic ratings as published in BEYER ON SPEED, etc. My raw numbers are track specific and change from year to year and my rating scale is different in that the lower numbers are better. And , as you noted, I incorporate pace into the final figure so the final rating has that component already built in. Using my methodology the raw figures come out to:

Race 1 -73
Race 11 -44
Race 12 -33

As you can tell, I don't think the BSF 105 is accurate. My adjusted Derby figure is -50 which, coincidentally, matches exactly the BRIS assignment of 100 based on a correlation of hundreds of my figures with BRIS figures for the same races. The correlation coefficient between my figures and the BRIS figures is an excellent 0.87. Interestingly, a similar correlation of my figures with BSFs affords a correlation coefficient of 0.88. In the case of this year's Derby I think Beyer may have relied too much on projection from shorter distances. My take on the Derby is that none of the horses were up to classic standards on the day. I saw this before in the 1995 Belmont when Thunder Gulch won in 2:32 and was assigned a BSF 101 coming off a Derby BSF of 108. My figure had it equivalent to something closer to a BSF in the mid-80s. That 2:32 was the slowest in 25 years and is still the second slowest since 1970. That's not to say that on a another day Thunder Gulch couldn't have won in 2:28 but on that day the race turned strategic, the pace was lethargic and that was the result. I don't see any justification for upgrading a performance not justified by the physical reality.


You wrote, " My take on the Derby is that none of the horses were up to classic standards on the day."

I"m having a little trouble understanding why 19 of 20 horses ran below their latest Roman Performance Figure. What is the likelihood of that event happening?
~
Have you considered publishing your RPF formula so horse racing fans can produce reliable performance figures?

Thanks
:11:

[Steve R]

You wrote, " My take on the Derby is that none of the horses were up to classic standards on the day."

I"m having a little trouble understanding why 19 of 20 horses ran below their latest Roman Performance Figure. What is the likelihood of that event happening?
~
Have you considered publishing your RPF formula so horse racing fans can produce reliable performance figures?

Thanks
:11:
Since so few American horses are bred for classic distances these days it doesn't surprise me that they can't match the figures they earned at 9f or below. I assume I'm not the only one who believes this. Beyer's par for the Derby is 107. The last Derby winner to match or exceed that was Big Brown in 2008. I credit American Pharoah for being able to maintain his figures throughout the entire TC trail even if my figures also have him a couple or so lengths below classic standards. Some may do better in the next two races because the Derby is unique in the sense that the horses have never experienced reaching the quarter pole after running so far only to discover they have to keep running longer than they ever have. With that experience in hand some may respond with more confidence and enhanced conditioning.

The issue with my figures is not the formula which is straightforward. Instead of using just final times I use entire pace lines for comparison with some standard, and I generate variants in a similar way except that I use regression analysis instead of averaging or clustering. The issue is that they can only be calculated using a computer and that they require pace line pars for each track, surface, distance and class of race. Frankly, if someone out there had the experience and expertise to do so I wouldn't mind trying to go national with them. For now I can only do selective race cards although I do publish the figures for all graded stakes in the U.S. and Canada.

[cj]

This has been corrected.

http://www.equibase.com/static/chart/pdf/CD050215USA12.pdf


If you see an error in a chart please send email to feedback@equibase.com

Thanks.

Check the final time of Northlands Park R6 on May 8th, that is my good deed for the day.