View Full Version : Quirin's magic formula
Alan Wight
11-27-2005, 04:24 PM
In the last chapter of Quirin's last book, Handicapping By Example, "Toward More Comprehensive Speed Figures," Quirin mentions that Beyer had expressed an interest in having a formula that combined pace and final time into one all-encompassing number.
He goes on to outline the beginning of such a number, "a magic formula," but says that he has "no data base that contains accurate pace and speed figures on which to run a test."
Does anyone know if this method has been tested on a database?
Has anyone since Quirin come up with a "magic formula." I think Quirin identifies the main problem here: "One's first inclination would be to add a horse's pace and speed figures, but this approach unfairly penalizes come-from-behind types whose pace figures as a rule come up relatively weak."
Quirin, along with Beyer, are the two who got me hooked on this game. I think I've improved on Quirin's stuff doing precisely what you mention. If that sounds cocky, it isn't meant that way, but I honestly believe it.
I have never read the Quirin book but I have always assumed that the various power numbers that are out there combined a final speed number with pace analysis. One could throw in other factors if so inclined.
Nowadays lots of folks have the sort of database needed to test such a number. I would imagine that most people who developed an adjusted speed figure would use a database as a development tool.
Certainly no question in my mind but that a number incorporating more than speed can be more predictive than a speed rating on its own.
I tried that formula for a while, but never had any real great results with it, taking into account the amount of work that went into it ( no puter used)
Would be an interesting thing to recreate using my HTR database - should be a simple thing to do and compare the results to the HTR fig and the Cramer figs - Beyers not in my db.
I think I'll do that...thanks for the idea.
And yes, CJ had done some things to speed figures that I rank as significant as the introduction of Beyers in the racing papers.
Quinn should do third versin of the New Expert Handciappers and include guys like CJ, Ken Massa, Dave Schwartz, Nathan, Jeff, fffastt, and others who have shred ideas here and gone out and did something new.
I know I cut and paste a lot of stuff from here and have one dyno - mite handicapping book in My Documents.
toetoe
11-27-2005, 06:13 PM
Tom,
Your forgetting the 'a' in 'shared' has inspired a slogan for my new political party --- the shredding of ideas. It has a ring to it that I can not resist. The podiatrist party? The Iamb Somebody Party? Toetoe wants YOU.
46zilzal
11-27-2005, 06:37 PM
Quirin, along with Beyer, are the two who got me hooked on this game. I think I've improved on Quirin's stuff doing precisely what you mention. If that sounds cocky, it isn't meant that way, but I honestly believe it.
When I was on THE FIFTH ESTATE (the Canadian equivalent of 60 Minutes) the production staff called Dr. Quirin to see if he had ever heard of me (had corresponded many times with him at Adelphi University) and it was on his recommendation that they eventually came out and did a show on how a group of us used the computer to evaluate pace back in the mid 80's....his was a good foundation of information that provided the basis of all MY subsequent use of the computer. I just found that the Sartin variety of pace evaluation worked the best for me.
the come from behind horses should be downplayed on dirt.
...the come from behind horses should be downplayed on dirt.
To a point yes, but how to do it was the hardest part. It took me many, many trials to come up with something that produced results.
46zilzal
11-27-2005, 06:53 PM
To a point yes, but how to do it was the hardest part. It took me many, many trials to come up with something that produced results.
Simple...establish what style wins at what distance at an oval and then if a closer has enough total energy to overcome that style, it is a choice. Also using a number (called the early/late balance) will tell you when the closer has no chance......same ideas, from different directions
46zilzal
11-27-2005, 06:57 PM
also, tracking the e/l numbers through a program will give a strong idication of a bias, if there is one.
of course the cold weather helps the front end
Well, yes, but the guy was asking about one number, not a printout with about 15 for each running line :)
46zilzal
11-27-2005, 07:18 PM
Well, yes, but the guy was asking about one number, not a printout with about 15 for each running line :)
you don't look at all the extaneous stuff...I know I don't A strong believer in K.I.S.S.
classhandicapper
11-27-2005, 07:52 PM
" I think Quirin identifies the main problem here: "One's first inclination would be to add a horse's pace and speed figures, but this approach unfairly penalizes come-from-behind types whose pace figures as a rule come up relatively weak."
All In MY HUMBLE OPINION!
IMO, there's little doubt running too fast early causes a slower final time.
However, IMO the Quiren method often unfairly penalizes slow paced races for closers AND FRONT RUNNERS!
Running a slightly slower pace only helps most horses run a little bit faster at best, but soon after that the slow pace starts working in the other direction because most horses can't come home fast enough to earn faster final time figures once they've traveled too slow early. IMO, this is true for both front runners and closers despite what is commonly believed.
So if the typical pace is 47 113, then maybe 47.2 yields a 112.3, but 47.4 might yield 112.3, 112.4. or 113 working in the opposite direction or not at all.
Whatever the exact formula is, I doubt pace figures can be measured accurately and consistently enough to capture it all the time because the point at which the effect switches over seems very narrow to me. It's probably not even the same for all horses because they have different closing abilities even when they are similar in overall ability. The Sartin guys probably have a good grip on that.
I am much more apt to downgrade a horse that set a slow pace if he also had a loose lead and/or earned his best lifetime figure than I am just because his pace figure is a little slower. That way there is more evidence that the slower pace actually helped him run faster and didn't start working in the reverse direction. Otherwise, IMO you will often be downgrading horses for their slow pace figures when you shouldn't be (front runners too!). That's often a bigger mistake than not looking at pace at all.
IMO, another thing that Quiren had wrong was that he only looked at the second call fraction (4F in sprints and 6F in routes). There is little doubt in my mind that using just the second call pace figure gives closers too much pace credit compared to the horses that were battling in fast fractions during the earlier 2F of a sprint or 4F of a route.
In other words, 21.3 44 is a lot better than 22 44, and 45 110 is better than 45.3 110.
There are just loads of problems creating 2F pace figures because wind and starting gate runnups, beaten lengths and other things complicate matters dramatically for the earlier fractions. When I tried to make 2F figures, I was constantly running into difficulties. It may be better to simply look at the second call pace figure and see if the horse was on the lead battling early, off the pace but being used hard, or off the pace relaxed during the earlier parts.
The other thing Quiren never looked at was middle moves.
How many times have you watched a route race and seen the first 4F go slow and then the next 2 furlongs blaze as the front runner starts to repulse the first challenges?
It happens a lot in dirt routes and on the turf.
Even though they often reach the 6F point in average time (more or less) after a very slow early part, the horses making big moves into that hot middle are being used up the same way as front runners that battle in the early part, but there's no way to capture that with a single pace figure. For routes, 2 pace figures are better than one. (sprints are just too tough)
GaryG
11-27-2005, 07:59 PM
There is value in looking at the 5th furlong in 6f races. This will often show a horse that was making a winning move but was just a bit short. The horse must be in reasonable contention all the way.
46zilzal
11-27-2005, 08:01 PM
There is value in looking at the 5th furlong in 6f races. This will often show a horse that was making a winning move but was just a bit short. The horse must be in reasonable contention all the way.
many a horse is easing up by that point so it can very well be a MIS-CONCEPTION
GaryG
11-27-2005, 08:04 PM
Works fine for me 46....a horse doesn't run an :11 furlong just because the others are easing up.
classhandicapper
11-27-2005, 08:06 PM
There is value in looking at the 5th furlong in 6f races. This will often show a horse that was making a winning move but was just a bit short. The horse must be in reasonable contention all the way.
I like that too.
I even like looking at the 6F call of 7F races and things like that for sprinters, but you have to make sure you know if there was a big wind because it can really have an impact on the ratings sometimes.
IMO, another thing that Quiren had wrong was that he only looked at the second call fraction (4F in sprints and 6F in routes). There is little doubt in my mind that using just the second call pace figure gives closers too much pace credit compared to the horses that were battling in fast fractions during the earlier 2F of a sprint or 4F of a route.
In fact, Quirin did advocate the use of a 4f pace figure in route races.
46zilzal
11-27-2005, 08:49 PM
In fact, Quirin did advocate the use of a 4f pace figure in route races.
tried it for a long time but it does not represent a consistenly competitive part of the race so I dropped it as nonpredicitive
RaceIsClosed
11-27-2005, 10:12 PM
In the last chapter of Quirin's last book, Handicapping By Example, "Toward More Comprehensive Speed Figures," Quirin mentions that Beyer had expressed an interest in having a formula that combined pace and final time into one all-encompassing number.
That number is called a power rating, and it takes speed, class, pace, trainer, jockey, surface, pedigree, equipment and medication, distance, age, recency, earnings, and the kitchen sink into account.
He goes on to outline the beginning of such a number, "a magic formula," but says that he has "no data base that contains accurate pace and speed figures on which to run a test." Does anyone know if this method has been tested on a database?
Yes, it has.
Has anyone since Quirin come up with a "magic formula." I think Quirin identifies the main problem here: "One's first inclination would be to add a horse's pace and speed figures, but this approach unfairly penalizes come-from-behind types whose pace figures as a rule come up relatively weak."
I solved that problem a year or two ago. This is why most horses who pay 50-1 or higher win from well off the pace.
If I ever get sick of scoring a $1.30 ROI on my bets, I'll tell you more.
RaceIsClosed
11-27-2005, 10:14 PM
In fact, Quirin did advocate the use of a 4f pace figure in route races.
You can circumvent the pace problem by examining speed figures from multiple races. Races which are skewed by the pace will tend to yield figures that lay off the expected range of performance (like a horse who runs 65-70 usually but tosses in an 82 when closing against a :44.1 half, etc.).
46zilzal
11-27-2005, 11:36 PM
If I ever get sick of scoring a $1.30 ROI on my bets, I'll tell you more.
handicapping procedures and wagering techniques are exlusive to one another
RaceIsClosed
11-28-2005, 06:44 AM
handicapping procedures and wagering techniques are exlusive to one another
In theory, I suppose they are.
The original poster asked if anyone had a "magic formula number" and I was just saying that such a number is a power rating rather than a speed or pace figure, that some pros are making this type of number, and that given their ROI, they aren't likely to be sharing their methodology anytime soon.
classhandicapper
11-28-2005, 09:11 AM
In fact, Quirin did advocate the use of a 4f pace figure in route races.
My memory must be slipping. :lol:
classhandicapper
11-28-2005, 09:37 AM
tried it for a long time but it does not represent a consistenly competitive part of the race so I dropped it as nonpredicitive
I find 2 pace figures useful in routes on dirt and turf.
It helps you see which horses made big moves into a hot paces vs. those that made big middle moves against dying front runners.
It also allows you to see which horses worked hard during the entire early part of the race vs. those that were close at the 6F call but ran more evenly in getting there.
Assume PAR of 100.
If the pace figures were 80 100 and the final time was 100, I can't see how that could be a good race shape for a closer even though they reached the 6F point in PAR. When you watch races like that you can see the closers working really hard to get into contention during that middle part. They often can't get into contention if the middle is very fast and other times they hang because of the effort.
On the flip side if the pace figures are 120 100 and the final time was 100, IMO, that very fast 4F almost certainly took some starch out of the front runners even though they were able to slow it down to PAR for the second pace call.
There are obviously many possible combinations.
If the question is whether or not we can accurately formulize all these various pace combinations and various horse position adaptations related to them etc... the answer is probably no. At least I can't past a certain point. But
I would still rather have a better understanding of the race development.
hurrikane
11-28-2005, 10:34 AM
If I ever get sick of scoring a $1.30 ROI on my bets, I'll tell you more.
I smell a rat.
CH,
The shapes you mention, such as 120 - 100 -100 or 80 - 100 - 100 are rare. It wasn't worth the time and effort involved in making an additional pace number. I did it for two years and found this first number was misleading as often as it was helpful.
Alan Wight
11-28-2005, 12:16 PM
I think I've improved on Quirin's stuff doing precisely what you mention. If that sounds cocky, it isn't meant that way, but I honestly believe it.
Have you run a test such as the one Quirin said he couldn't do, that is, on a database? If you have done such a test (comparing the performance of your figs with other figs), have you published the results in such a way that they can be analyzed by others?
classhandicapper
11-28-2005, 12:27 PM
CH,
The shapes you mention, such as 120 - 100 -100 or 80 - 100 - 100 are rare. It wasn't worth the time and effort involved in making an additional pace number. I did it for two years and found this first number was misleading as often as it was helpful.
I know we've had this discussion before, but I wonder if that's because you were trying to create a formula for measuring the effect on final time for all these various pace combinations vs. just understanding the race development and result better.
I think I tend to be a little less numbers oriented than you even though I love the numbers almost as much.
I think we would agree that there are many potential inaccuracies in the track variant for both speed and pace figures. There are also a multitude of pace combinations. Then handicappers attempt to combine them using a formula that works the same way for horses that have varying levels of stamina, early speed, closing speed, etc.... even when they are similar in overall ability. After that you still have to worry about the effect of track bias on the formulas. Some days/tracks are very speed favoring and vice versa. That probably changes everything.
I love all these numbers and swear by the quality of yours, but I think they have limitations.
I get some of my understanding about the pace and performance just by watching the races and knowing the quality of the horses in the race and part from the numbers. Combined, I think I get a clearer understanding.
In the early 90s, I made 2F, 4F, 6F etc... pace figures for all the dirt races in NY (even 2F in routes for awhile, 6F call in 7F races, etc...). They were Beyer scale figures similar to yours etc... I found all sorts of weird pace combinations.
I found all those early fractions useful (even 2F in routes to some degree) for understanding the race development and performances. I stopped making them because of time constraints and a very good source for pace figures became available for NY via Logic Dictates where I was already customer. However, if I had that extra information now I would welcome it. I am sure it would help me. I just wouldn't try to make a formula out of it.
I think when you have a series of data inputs and each may have a small error, when you try to combine them all with a less than perfect formula you can get a mess if there are too many inputs. But that doesn't mean the individual pieces of information aren't isn't useful in other ways standing alone.
I post three tracks 5 days a week on my site, people are free to run any tests they like.
I can tell you for the entire month of July, I tested every race at every track in North America, and blindly betting the top last figure earned on the same surface as today (excluding 180 day and more layoff horses) and that produced a 5% loss. I am doing the same in November and the loss is 7% so far with a couple days to go. I am still compiling data from the entire year. Remember, this is no handicapping at all, none. If the top fig is ridden by a 1 for 124 guy, it counts, or if it is a maiden special with 8 of 10 firsters, it counts. If it was earned in a mile and a half race and today is 6f, it counts.
I think that is pretty damn good, others may disagree.
I should mention, the winning percentage in July was 30%, and in November it is 29%.
classhandicapper
11-28-2005, 12:40 PM
CJ,
Those stats are amazing.
If you ever find the time, I'd love to see some track specific stats. There are tons of figure oriented players at some tracks, but probably not all. I'd love to know if there are better opportunities some places vs. others. Obviously, I'm most interested in NY. In fact, I'd be happy with just the NY stats.
46zilzal
11-28-2005, 01:10 PM
I should mention, the winning percentage in July was 30%, and in November it is 29%.
that is very good, but most folks, (I DON'T KNOW about yourself) never seem to change the 'capping for COLD weather and it MUST be a factor since I do MUCH MUCH better in the cold.....
The problem is there aren't enough yet to come to a track by track conclusion. Here are the New York stats for November, I have to find July:
181 races, 51 winners, 28%, ROI +2%.
(Top prices: 30.60, 23.00, 20.60, 18.20, 17.40)
I didn't mention ties, highest odds is the tiebreaker. Not sure this is the best method. Yesterdays 5th, two top at 98 ran 1-2 for a $149 exacta, but it counts as a loss here as the 6-1 beat the 10-1.
Top 5 tracks (minimum 100 races):
TuP + 16%, CD +7%, Mnr +3%, Aqu +2%, (Tie) Hou +1%, FL +1%
Worse 5 tracks
CT -25%, Haw -21%, Tdn, -16%, GG -16%, RP -15%
classhandicapper
11-28-2005, 01:20 PM
Have you run a test such as the one Quirin said he couldn't do, that is, on a database? If you have done such a test (comparing the performance of your figs with other figs), have you published the results in such a way that they can be analyzed by others?
I would be interested in seeing the results of a study like that for all the major figure makers using the same rules for each study regarding ties, layoffs, surface changes etc....
I think you could run into some problems with the Sheets and TG though. They include ground loss so they give a lot of extra credit to closers that almost always lose a few lengths that way. As a handicapper you are expected to subjectively take that into account depending on the likely trip. Not sure how you could correct for something like that.
What I would also really like to see is a study of TG/RAG results with and without ground loss included, but that would be impossible without cooperation from them.
classhandicapper
11-28-2005, 01:24 PM
CJ,
Maybe in the case of a tie you should just count them both. I'm not sure how the other studies on figures are done when it comes to surface changes, ties and layoffs etc....
I ran a query for ROI on top last race adjusted speed rating (52,000 races) and got an ROI of -6.8%. I would not have guessed that such a simple handicapping method would be that much better than the track take.
Directionally the same (though not quite as strong) as CJs results.
formula_2002
11-28-2005, 03:30 PM
I post three tracks 5 days a week on my site, people are free to run any tests they like.
I can tell you for the entire month of July, I tested every race at every track in North America, and blindly betting the top last figure earned on the same surface as today (excluding 180 day and more layoff horses) and that produced a 5% loss. I am doing the same in November and the loss is 7% so far with a couple days to go. I am still compiling data from the entire year. Remember, this is no handicapping at all, none. If the top fig is ridden by a 1 for 124 guy, it counts, or if it is a maiden special with 8 of 10 firsters, it counts. If it was earned in a mile and a half race and today is 6f, it counts.
I think that is pretty damn good, others may disagree.
I should mention, the winning percentage in July was 30%, and in November it is 29%.
cj, you really want to break the analysis down by odds and compare your top figure results, by odds, to the public's.
I'm sure I dont have to tell you that, but I feel I just have to mention it here.
RaceIsClosed
11-28-2005, 03:44 PM
that is very good, but most folks, (I DON'T KNOW about yourself) never seem to change the 'capping for COLD weather and it MUST be a factor since I do MUCH MUCH better in the cold.....
I know if human track meets were scheduled outdoors, I wouldn't be judging winter and summer performances equally, nor would I expect the winter races to be as true to form.
classhandicapper
11-28-2005, 03:53 PM
I ran a query for ROI on top last race adjusted speed rating (52,000 races) and got an ROI of -6.8%. I would not have guessed that such a simple handicapping method would be that much better than the track take.
Directionally the same (though not quite as strong) as CJs results.
Whose figures?
What happens in the case of a tie, surface switch, layoff?
Whose figures?
What happens in the case of a tie, surface switch, layoff?
My figures. Data was from dirt races only so the figure used was from the most recent dirt race. I am using real number values (rather than integer values) so there aren't any ties (or not enough to worry about).
I ignored layoffs in the calculation referenced above but re-ran using the 180 cut-off as did CJ with a -6.5% return resulting.
All this shows is that many of us are making adjusted speed ratings and they do have some predictive value. Of course it would be idiotic to handicap looking only at the last race adjusted speed figure (or to accept a -6.5% return).
46zilzal
11-28-2005, 04:01 PM
I know if human track meets were scheduled outdoors, I wouldn't be judging winter and summer performances equally, nor would I expect the winter races to be as true to form.
EXACTLY
...Of course it would be idiotic to handicap looking only at the last race adjusted speed figure (or to accept a -6.5% return).
I actually won on a horse at Crc Sunday that had the worse last race adj speed figure, paid $23 (highly favorable pace scenario). Still, the numbers are a very good starting point.
Also, if you are betting 10,000,000 a year and getting a 10% rebate (I'm not doing either :) ) -6.5% would be pretty good.
classhandicapper
11-28-2005, 04:48 PM
Thanks.
It seems as though several different performance measurements produce similar win percentages and improvements over the take. I've seen others that were similar.
What I think some people underestimate is how hard it is to get over the remaining hump without the rebates. Everyone knows that distance, track condition, trainer, jockey, etc... are also factors, but to a large extent those things are already built into the odds quite efficiently also. As you add incremental information to the process, you sort of swap some winners out and new ones in without adding much value unless you have a superior understanding of each factor to the general public. You have to be picky or have other value oriented insights.
Alan Wight
11-28-2005, 05:36 PM
I can tell you for the entire month of July, I tested every race at every track in North America, and blindly betting the top last figure earned on the same surface as today (excluding 180 day and more layoff horses) and that produced a 5% loss.
How about a study comparing the top overall peformance rating vs the top speed figure vs the top pace figure?
formula_2002
11-28-2005, 06:41 PM
My figures. Data was from dirt races only so the figure used was from the most recent dirt race. I am using real number values (rather than integer values) so there aren't any ties (or not enough to worry about).
I ignored layoffs in the calculation referenced above but re-ran using the 180 cut-off as did CJ with a -6.5% return resulting.
All this shows is that many of us are making adjusted speed ratings and they do have some predictive value. Of course it would be idiotic to handicap looking only at the last race adjusted speed figure (or to accept a -6.5% return).
Was that dirt to dirt , i.e. last race was on dirt and today's race is on dirt?
Today's race on dirt and any turf races going back ignored so the adjusted speed figure comes from the most recent dirt race.
I ran an extensive 6 month study on CJ numbers.
Here are the resuults:
ezpace
12-06-2005, 10:41 PM
CJ, don't change a thing, except,, don't waste anymore time running studies on your figures especially the performance numbers . NO need. Their not as 'time" tested as my "1948" speed figs but on all dirt and minor turf tracks they are slightly better. The 1948's still rule on major turf courses.I've looked at a lot of software numbers .THE CHOICE. *Last* critique on CJ figs validity ... Just bet ;)
Alan Wight
12-07-2005, 01:11 PM
This thread has been sidetracked a bit. It's not about CJ's figures, or whether anyone can win money using them, but whether anyone has succeeded in developing Quirin's ideas about a single figure that combines pace and final speed figures for all horses.
CJ chose to say he had succeeded. Unfortunately, the studies he's presented as evidence are flawed: they have no control, which would probably be a study of the same group of races based on final figure alone, either his own final speed figures or another service, preferably both.
Furthermore, the win percentages his studies show fall within the range of similar studies for Beyer figures (Joe Cardello in SPEED TO SPARE presents the results for top Beyer figure: 25% to 31%, including ties.)
Based on CJ's studies, given that his overall figures perform roughly the same as Beyer figures, you have to conclude either that his own final speed figures perform worse than Beyer or that they perform the same as his overall figures. If the latter is the case, what's the point of his overall figures?
The speed figures win at about the same rate as the overall numbers. The difference is the set of winners is not the same, and the prices are better, which is what I aimed to accomplish. If I included lower priced ties, remember, I only counted the highest odds of ties, the win percentage jumps to around 33%, the ROI stays the same.
So, to answer your question, the point is value. Everyone has decent speed figures available if they want them, but they also get bet accordingly.
If measuring pace suddenly became all the rage and was being bet accordingly, I'd be busy finding another formula(s) that won at a similar percentage but with less public knowledge. I don't think there will be any method that measures pure ability be it pace, speed, etc. that will win at a much higher percentage, too many other factors are involved. Just my opinion.
vBulletin® v3.8.9, Copyright ©2000-2024, vBulletin Solutions, Inc.