Beyer Figs/Beyer Turf Figs

[Macdiarmadillo]

I'm picking up a discussion line started in the "Beyer vs. Brown vs. Friedman. . ." thread. So it really starts about page 3 over there; I'm trying not to duplicate posts, that's all.

I wrote:
So IMO a horse running 115 has confirmed its ability at a certain level.
marc at drf replied:
Incorrect. Please see "Picking Winners," "My 50,000 Dollar Year at the Races," "the Winning Horseplayer, "Beyer on Speed," all by Beyer, or Len Ragozin's "The Odds Must Be Crazy." The inspiration for speed figures was the desire to get away from calculations such as the one you are relying on.
You mistake my comment for old-style class handicapping. The rather awkward construct trying NOT to say "class" is me trying to say 115 is a 115. I have all those books and have re-read them many times, thank you. Maybe you should go back and read those books since a foundation of Beyer figs is the par time chart, created by determining average times for various levels of races. Maybe this was not necessary after the first par time chart was constructed, but the eigen-chart, the ur-chart, involved correlation of figs to various levels of racing.

There are still races IMO, maidens for one, that are won based on exceeding a barrier fig. That might be interpreted as class indications to you. That barrier is the par fig. It is easy and successful to bet appropriate first-timers in situations where no other horse with a record looks to improve to the barrier level.

I wrote:
I doubt that anyone would find it acceptable that 2yo stakes colts would have pars at 80 on dirt but 50 on turf.
marc at drf replied:
Please read any of Beyer's books. He makes it painfully clear in every one of his books that his figures are based on the times of races and the previous FIGURE histories of the horses. Assuming that the times for various classes will be the same is a strategy right out of 1965.
By definition, there is a dirt/turf figs discrepancy between you and I and therefore also our par time charts. I calculated them from scratch: GGF, Bay Meadows (calculated twice and essentially the same twice), HollyPark and SA . Turf figs calculated separately. Proved to me with exception of the downhill turf races that everything lined up consistently and worked.

There are many horses that perform in the same range of figs on either surface, confirmed in Ragozin and TG numbers, at least. So too in my Beyer figs. Not yours. Confirmation for me was in two speed horses of the 1980s in N. Cal., Quick Twist and Valentine's Lad, the latter probably unrateable and therefore without pace considerations clouding his performances. Same numbers on turf and dirt for each. And the list continues on and on to this day.

You consider dirt and turf as apples and oranges, I believe Beyer contended that, not to be compared. But then most people (I do. Yourself?) have no hesitation in extrapolating 6f dirt Beyers (however they do it) to 8f dirt races. So, dirt and turf pars should have some correlation and I am convinced they do. You can think what you will.

There are still valid strategies out of 1965, by the way. And since Beyer figs are a strategy of the 1970s, what about that?

I wrote:
10 points is a huge difference -- 5-6 lengths slower for a grade 1 turf winner vs. a grade 1 dirt winner over a distance.
marc at drf replied:
Incorrect. Please see page 160 of "Beyer on Speed." Beyer treats beaten lengths differently on turf. By reading (or re-reading?) this book, you should be able to clear up any confusion.
I read it when the book first came out and was dismayed to see this. This is using the 6.5f beaten lengths chart for longer races. I am stunned you still think 11 years later that this is a valid approach.

The book notes it was determined that "grass races are indeed tighter". We see this all the time. So the finishes are tighter and so some horses in one race are going to earn close to the same number. Using the 6.5f beaten length chart certainly separates finishers' numbers over a wider range but makes an individual horse's figs vary a lot, too much for my taste.

The beaten length chart is mathematically calculated for each distance based on lengths. A turf length is different from a dirt length to a specific chartcaller? No way. So it should not be to us IMO.

The justification (in the book) of bringing lousy horses turf figs in line with their equally lousy dirt numbers is misguided IMO. Enough people contend that horses prefer turf and will extend themselves on it even though stride, conformation, etc. limits them from racing successfully to the finish line -- ever think of that? Then it is logical that even the most mediocre horses would earn 39s on turf and 7s on dirt and that apt horses can improve not only a little but a lot when they move to turf.

delayjf also wrote (directed to marc at drf):
I believe Beyer addressed this dirt-turf discrepency in his last book. He felt he was using the wrong beaten length adjustment. (Macd note: wrong, meaning the original beaten length chart)
Has his adjustment solved the problem??
The proof is in the pudding -- do Beyer turf figs work, are they a measure of past performance? Well, you wouldn't necessarily say "yes" in N. Cal. In the days before 6-horse fields were common (when I was still calculating Beyers and playing there), turf horses with mediocre DRF Beyers often popped up at generous odds when in fact they had among the highest numbers in my calculations. When I could compare, mine correlated well with Ragozin figs and to a lesser extent, TG. And those were excellent numbers for predicting winners and contenders. Go through the stands and the simulcast outlets here and you will find many people have returned to adjusting the DRF speed fig+variant for local turf races.

[JustRalph]

All this stuff about Turf races...... I think if you don't take the specific Turf course into account, you can throw all that crap out. Some courses are favorable to front runners trying to upset the field (Hollywood comes to mind) others are pretty much equal and play like dirt courses in my opine (Sam Houston) and then there are others that seem to cater to those traditional late Turf runners.

A year back I couldn't hit a Turf race for nothing. I was playing with all the figures and all the conditions mentioned in this thread and more. I finally basically stopped playing turf races and just watched. Then I started checking out "Race Shapes" in my ongoing handicapping studies and I noticed that while I was watching those Turf races I was making predictions on the race shape. Suddenly the Turf races starting coming into focus much more. In studying the different Turf courses I was playing and adding that to the mix, It became much easier. I scored some nice wins last year by predicting whether or not the big later runners were going to get up or not (Whitmore's Conn) (http://www.suntimes.com/output/horse/cst-spt-race10.html) and also making predictions on whether stone cold front runners would stay on the board. Nowadays I don't look at any of the aforementioned crap...... I look at race shape and where they are running......it serves me much better

[TravisVOX]

Figures on turf can be almost completely negated unless you properly assess the race. In other words, turf racing counts from the 3/8's (+/-) on. Calculating come home times based on that will get you more accurate turf horses. Often times, a 25 flat quarter means nothing on the turf...it's the horses late kick that counts. He can get swallowed alive come the stretch.

It's also true, like stated, that each course is different layout wise. And then each course varies a second degree with where the rail is placed.

Tricky game to get some ratings, but James Quinn provided the best analysis of it in Turf Handicapping. Calculate them coming home!

[Marc At DRF]

Macdiarmadillo,

You've written some things I agree with but some that I don't. There's plenty of people who lurk or contribute here who may have more time to speak up in response to such an exhaustingly long (which I certainly appreciate) post. If no one says boo, I'll try to get to it next week. Sorry to plead "I'm too busy," but I am, for now.

[delayjf]

Tricky game to get some ratings, but James Quinn provided the best analysis of it in Turf Handicapping. Calculate them coming home!

I've been reviewing Quinns methods and I reviewed the season at Delmar that he first test ran his figures. Here's what I've concluded.

1) His methods are more accurate when comparing horses running the same distance. Reviewing his season at Delmar almost all of his profits came when he was able to compare horses returning to the same distance. Note: he does warn about this especially when the pace pars for the different distances are not the same.

2) While I agree in theory that turf horses should get more credit for running fast longer ei: 3/8 in 36 is better than 1/4 in 24, even though they are the same velocities. I'm dubious that his turf charts are as accurate as they can be. Specifically the points he assigns per 1/5 sec in 1 mile and 1 1/8 races.


Beyer points out the flaw in his book by showing that according to Quinn a horse who finishes 1/4 in a slow 27 flat (hollywood park) will cover the final 1/16 in 5 and change at 1 1/8. In Quinn's defense, 27 flat is on the extreme end of the chart range were any relational inaccuracies with Quinns 1/5 second values would manifest itself. Within the normal range of final fraction times, the chart may be more accurate. I've never seen a final fracton of 27 flat in a turf mile at Hollywood.

Now it seemed to me that we could assume a relationship between the 1 mile and 1 1/8 final times. But Instead of using the 1/5 values of 1.6 (1 mile) and 2.4 (1 1/8) that Quinn uses, I think it might be more accurate to simply exchange the point values derived the way Beyer does for the dirt. In this case, the point values would be 2.0 for 1 mile and 2.3 for 11/8. But, This is just a theory.
Anybody got any better ideas???

While were on the subject of Beyer vs Quinn, after reviewing Quinns parallel time chart (dirt) and comparing it to Beyers, I think Quinns is much better. funny both are derived throught the same method.

[Macdiarmadillo]

Marc, pick this thread up any time you are able.


race shapes: Yeah, I like this approach because it's flexible. Still, when you get below stakes races, you can have all kinds in there and I'd like to know, with some kind of figs, who the contenders are to start with.

come home time: Strong method but I'm seeing too many races on speed-biased turf courses where the horses with the best figs and the best kicks just aren't getting up. Close but no cigar -- method is valid but you're not making any money. The mile turf races in California seem to negate much emphasis on late pace, less so elsewhere, it seems. How's Quinn working for mile races vs 1 1/16?

Any trip handicappers want to throw in their 2 cents?

[Tom]

I have had very good success using Quinn's turf figs, but not going to all the trouble of making charts for each course. I use one turf chart with 100= :24, :30, :36.
The courses I use it on are usually Bel, Sar, Tam, so they may not be as effective on courses that favor more ealry speed. At Sar, they work VERY well.
Raw time, BTW.

[delayjf]

Tom,
That is exactly what I do. I would be interested to know if you've experienced the samething I have. For example, if I'm looking at two horses one that closed 5 lengths into a 24 flat final 1/4 at a mile vs. one that closed 5 lengths into a 36 flat at 1 1/8. I've learned through the school of hard knocks to favor the horse from the longer distance. Which agrees with Quinns theory of running fast, longer.

With regard to front runners, at the big tracks I've found that if a front runner can't run the last fraction in 24, 30, or 36, it's unlikely they will hold on. Remember, closers will be gaining lengths into those same final fractions. For the front runner it's then a matter of how slow a pace he can set.

I've had good luck on the turf with Sartin sustained pace. Sartin Sustained pace does a good job of balancing the pace of the race with the finish. But, it gives no extra credit to horses that are able to sustain their velocity for longer distances. Anybody have any idea how to factor that into the equation??

[Tom]

Delayjf,
I use only pacelines within 0.5 furlongs of today's whenever possible. I do not like to use a 9 furlong line in a mile race, but sometimes you have no choice.
If I have to mix distances, I prefer the longer final fraction if I am looking at a 9 furlong race and the shorter one if I am looking at a mile race. I like SP and LP ratings on turf, even when horses are winning near or on the lead - you still have to have the sustained kick to ward off the deep closers.

AP EP SP LP
1 2 3 3
4 3 4 4
3 4 2 1
2 1 1 2

At Sar or Bel, I would generally like the 3 and 4, but at other tracks where early is more important, I might go with 1 and then either 3 or 4.


CJ posted hismethod for having two ratinfs for turf races and I played around a bit with it this winter. I think this might be a great way to model turf racing-which rating is winning can tell you how to lay. I will be giving that idea a big test this summer.
I love turf racing, and IMHO, you don't have the crooks and cheaters on grass like you find on dirt. And I think you find a lot more opportunites for trip excused on turf if you watch the replays and take notes. (Arrrrg! Work!)

[Macdiarmadillo]

(Brought over from Beyer vs. Brown vs.. . .)

I wrote:

I'm guessing the lower Beyers can be accounted for by allowing for no improvement in any horse when calculating new Beyers.
Lance responded:
This is untrue. One could find 20 examples of its falseness in 20 seconds. I just glanced at a stack of DRFs for No Cal. The counter-examples are endless. That Beyer does not know that racehorses can improve is truly one of the, well, strangest comments I have ever heard. Please tell me what prompted you to make it. Then, if you don't mind, we can get to your other comments on these two strings. I have many similar questions for you.
If you glanced at a stack of forms to respond, I'm not making clear to you what the concern is. Do you know the procedure for calculating Beyer figs? I'm talking about calculating daily variants, which you need to do ". . .when calculating new Beyers". And since these are based on previous Beyer figs, you will ignore a certain amount of improvement. Beyer was aware of this problem. I'm talking about potential improvement of certain horses being reduced because the daily variant is figured wrong. Keep doing this calculation over and over this way (strictly based on historical figs) and Beyer figs for ALL horses are gradually lowered.

So there is a possible loss of about 1-2 Beyer point per race for a specific horse due to not allowing for improvement when calculating the daily variant. Enough errors introduced like this (they may self-cancel sometimes) yields inaccurate Beyer figs down the line.

IIRC, Beyer's solution (can't find which book to cite from) was/is to add a point every month to every Beyer fig. That may have been selective for specific horses. It helps correct the problem but I feel it is inadequate and not rigorous. I have found a normal rounding procedure of a half point in the daily variant takes care of the problem: if a daily variant is calculated to be 16.5, then I rounded to allow the higher new Beyer fig. (how you do this depends on whether you consider the daily variant a positive or negative number [add or subtract it from a raw fig]). Anything else is rounded to the nearest whole number. There were multiple occurences of the daily variant falling on a half point in a month, therefore the blanket corrections could be cumulative, could be self-cancelling. I felt a one point correction a month was insufficient. And it should not be applied to every horse, either.

Granted, a point or two loss over several months for certain horses may not seem major but how about a horse being 4-5 points lower than he should be at that point after a year? I'm sure you think that amount is significant when handicapping.


I had great confidence in Beyer figs I calculated because they worked. And Racing Times/DRF Beyers seemed to gradually get lower against mine as would be expected if you were not allowing for enough improvement in handling the daily variant. Marc_at_drf may take issue with the above procedure, the previous comment and the notion of "a" daily variant, but we'll await his reply. But the par time table presented in "Beyer on Speed" (p. 237) matched my table calculated 12 years before, independently, and remained valid throughout the time I calculated Beyer figs. While the DRF group uses a different scale than I do or Beyer did, I still do not see a lot correlation with their numbers to other fig methods (Ragozin, T-G, for two), particularly for turf races. And there should be.

[Lance]

Macdiarmadillo wrote:

"If you glanced at a stack of forms to respond, I'm not making clear to you what the concern is."

I was addressing what you wrote, which is not necessarily the same thing as addressing what you were thinking, especially if you failed to actually type what you were thinking. If you allow for no improvement in any horse, you will get no improvement for any horse. Looking at DRFs is the quickest way to show your statement was false, which, of course, it was.

"Do you know the procedure for calculating Beyer figs?"

Answering this question would be inappropriate. One should display knowledge rather than assert that one has it. I mean, sure, once is OK, but over and over, as has been done on these two strings you split, well, I think it gets tiresome, especially when the assertion is false.

"I'm talking about calculating daily variants, which you need to do '. . .when calculating new Beyers'. And since these are based on previous Beyer figs, you will ignore a certain amount of improvement."

This is false. Beyer figures are based on projections. If you project too low, your numbers will be too low. If you project too high, your numbers will be too high. But even if you tend to project too low, MANY horses will still improve on your figures--just not quite as much as they ought to have. This is why I found your original statement (quoted above) so preposterous. And in addition, the computer age, which Beyer entered about 15 years ago, allows Beyer to monitor his projections--and hence his figures--constantly. So these problems are a thing of the past. There are problems with the Beyers, but you never mention them. You only mention the non-problems.

"I'm talking about potential improvement of certain horses being reduced because the daily variant is figured wrong. Keep doing this calculation over and over this way (strictly based on historical figs) and Beyer figs for ALL horses are gradually lowered."

Now I see your problem!!! You are thinking of the pre-DRF days, when Beyer did his numbers by hand and for only one or two circuits. No wonder DRF Marc and I could not make heads or tails out of your posts. Please buy "Beyer on Speed." In it he explains that since he is in the computer age now, he is able to press a few buttons and immediately get numbers from horses shipping to and from different circuits. By adjusting these numbers, he is able to keep his figures in line. If No Cal horses have been running lower Beyers each year, it is because NO Cal horses have been getting spanked by shippers and spanked when they themselves ship. And it makes perfect sense, since No Cal racing has been getting more and more pathetic every year. Good trainers have left. Five-horse fields are common. Clydesdales ship in there and win by a pole. The racing there is now a godawful joke. And the Beyers reflect that--thank god. Your way, they would not.

"So there is a possible loss of about 1-2 Beyer point per race for a specific horse due to not allowing for improvement when calculating the daily variant. Enough errors introduced like this (they may self-cancel sometimes) yields inaccurate Beyer figs down the line."

See above. This paragraph is antiquated.

"Beyer's solution (can't find which book to cite from) was/is to add a point every month to every Beyer fig."

This is false. In chapter six of "The Winning Horseplayer," Beyer explains how he handled this situation. I implore you to read this book, but that will be insufficient. You will also need to read "Beyer on Speed," which updates the previous book, bringing it into the computer age.

"That may have been selective for specific horses."

Of course not. You must know better than that. If you change one number, you must change every number in the race. Please read "Picking Winners."

"It helps correct the problem"

Well, Beyer didn't do it. His approach was much more flexible, as you will see when you read the book. But let's take a look at your approach. If you "add a point every month to EVERY Beyer fig," as you suggest, a claimer will take a long layoff, get 20 points added to his Beyers, and come back as Secretariat, who, if he were still alive, would be showing Beyers of 460.

"It helps correct the problem but I feel it is inadequate and not rigorous."

And you are nothing if not rigorous.

"I have found a normal rounding procedure of a half point in the daily variant takes care of the problem"

Why resort to these silly gimmicks? Why not make accurate projections?

"Marc_at_drf may take issue with the above procedure"

If he doesn't, his ass should be fired.

[Marc At DRF]

MacD,

Lance has really covered most/all of the ground I would have tried to cover, only Lance is doing it better than I ever would have.

Lance makes passing mention of the limitations he finds with Beyers. I think it's important to point out that each and every speed figure has its limitations-- none are perfect, despite the claims of most of the figure makers.

That's why I usually find the game of "gotcha" that figure-users and makers try to play against competing figure-users and makers to be so tiresome.

[delayjf]

I don't have the book in front of me, but doesn't Cordello's book state that the Beyers are actually MORE (higher win %)accurate on the Turf?? Thought I read that some place.

[JackS]

Making figures often results in a relavent rating system more than an actual real world method of producing actual times. An example in my own handicapping , it has occured to me that I was making a mistake about half-way through a particular race. Rather than go back to correct everything, I choose to continue making the same mistake with all horses. The results should be simular or maybe exactly the same, although the final numbers are now much less related to pace and final times and are now considered by me to be a rating method only. Of course this would not be acceptable if the numbers produced are going to be placed in a computer data base

[Memogram]

I have a hard time using any figures in turf races, they just don't seem very reliable.