Pace handicapping

[andicap]

On another thread, Ranchwest posted this:

"What clued me in that War Emblem would win the Derby and Ten Most Wanted would not, both coming off the Illinois Derby, was that Ten Most Wanted slowed significantly in the 3rd fraction and then accelerated in the 4th fraction. This told me he had a soft race. WE had a smoother line, thus a more genuine closing fraction."

-------
This got me thinking. Are there any books on pace handicapping that explain what to look for in assessing internal fractions. Brohamer's book didn't really deal with that. Nick Borg's "Off the Charts" did so in a crude, rudimentary fashion that I did not find convincing. (In sprints he gave a universal par as 22.3 for 1st qtr and 24 for all other qtrs -- way too simpliistic).

Jim Lehane wrote a little bit about this in Calibration handicapping talking about the "Golden Eighth." Horses that can go faster than 12 seconds in the next to last furlong are good bets.

Any other resources for learning about how to evaluate internal fractions?

[GR1@HTR]

Pace Makes the Race by Huey Mahaul (sp) something like $2.99 at Gamblers Book Club. Exactly what you are looking for. Fine 20 pages on technical indepth pace analysis.

[andicap]

I'll have to read it again. I read it several years ago and found it useless -- a lot of theory that couldn't be applied. Hope I still have it.

Thanks

[Tom]

This is why my favorite all-time program is still K-Gen (Sartin).
The deceleration graphs, broken into 1/2 furlongs, are a powerful visual. You can see where the leaders fall apart. Problem is, it is all manual entry and very time consuming, from the days of 1 track at a time. But in terms of really seeing the pace of the race.
it was dyno-mite.
I agree with Glen, the Huey book is very good.

[alysheba88]

Just a word of warning to people. Pace/fractions are important for sure. But don't fall into the trap of just looking at the fractions. They don't tell the whole story.

Pressure is as important if not more so.

A horse running lose on the lead in 45.1 can hold up far better than a horse involved in a three way duel in the same time.

On the opposite end, a closer sitting behind a insane speed duel will have a much better chance of getting up than if chasing a loose leader.

Those are two just very basic examples. There are a ton more of course. In short pace is important, but you have to know the full trip story. Blind numbers alone will lead you down more false trails than anything.

[midnight]

I've read the book, and Lehene is wrong in my opinion. Horses that overrun the quarter to eighth pole in dirt races usually hang at the end because they've made their run too early and used up their energy. I'd be interested in opinions to the contrary.

[GR1@HTR]

If I remember correctly, Huey's book had no pace info for the fist 80 pages, them boom 20 pages of pace stuff and how to detect possible trip mishaps via pace figs (accelation/deceleration) w/o looking at comments.

[andicap]

Maybe I got the wrong book??

[Pace Cap'n]

Originally posted by midnight
I've read the book, and Lehene is wrong in my opinion. Horses that overrun the quarter to eighth pole in dirt races usually hang at the end because they've made their run too early and used up their energy. I'd be interested in opinions to the contrary.

What I understood Jim Lehane to be saying was that, although the horse with the good time in the golden eighth in his last race may have hung, that it is perhaps a sign of good form for the upcoming race. The theory is that the horse now has "derived energy" from that strong showing in that portion of the last race.

If he hung in the last race, all the better for the odds this time.

[Figman]

Huey Mahl
Anything that you see that was written by Huey Mahl is worth getting. I'm sure Lefty will echo my sentiment. Before personal computers were available to all, Huey was writting programs and actually giving them away for the old TI-59 programmable calculators. Pace handicapping, wagering, record keeping, Huey covered it all well.

[Fastracehorse]

We all bet pace to some degree.

Class handicappers are betting pace.

Beyer and other similar speed fig players are betting pace.

Key race players are betting pace ( class within a class ).

I think to dissect fractions is taking time away from the art of handicapping, ie, picking winners.

I use to be a surgeon - but not anymore - and I swear things are better.

:) :) :)

fffastt

[Doug]

IMO dissecting Running Style and Pace Shapes is as good as doing the same with fractional times.

Doug

[SAL]

You know, evaluating fractions was an exercise that was covered in an excellent harness book I once read. It mentioned that horses that made up ground in a fast quarter were the ones to watch for. Horses that make up a ton of ground in a slow quarter was discounted. Ten Most Wanted made up only 1 length in the 3rd quarter which was run in 25.0. So depending on the variant for the day, ranchwest may be right. I don't think that 25.0 is quite that slow, the 4th quarter was much slower.

SAL

[modred]

Huey Mahl put out 2 "The Race is Pace" books. The first was copyrighted 1975 and contained 65 pages; the second copyrighted 1983 and had 126 pages. These two different editions may have been the cause of confusion to Andi.

[MV McKee]

I for one would be extremely leery of the golden eighth. Not because of the rationale behind it, but because the data you use to calculate it is extremely inaccurate.
This is going to sound extremely pitiful, and demonstrate clearly that I have no life.
Every year, for each track I play (there are only 15) I do the following:

I dub (video tape) 25 races at each of the commonly run dirt distances at a given track. These dubs have a stopwatch function on them. For each horse in each race (that is in the picture) I notate the time (1/100ths of a second) behind the race leader at each call, and at the finish. These time values are then compared to the chart beaten lengths, and for each race I get a value that when multiplied by the chart beaten lengths for each horse produces the smallest absolute difference from the actual time.
Example:(BL = Beaten lengths)
Position ChartBL TimeBL
1 0 0
2 .5 .09
3 3 .52
4 7.5 1.27

For this call in this race I would get the value .17, as it most accurately translates the ChartBL into TimeBL.
I then look at the Average and StDev of the values for the 25 races, and determine what the average is, as well as whether that average is consistent and reliable enough to be usable in calculating any kind of pace figure.

In the case of final times:

I have yet to find a track where these are not accurate enough to be usable, esp. since they are generally taken from the finish filmstrip. The actual time value varies widely from track to track however. The following values are EXTREMELY accurate:
Emerald Downs .17 per length
BayMeadows .21 per length
Golden Gate .20 per length
Hol, SA .18 per length
Portland Meadows .16 per length

1/4 mile (1st) calls in sprints are not nearly as accurate, but still useable, and hover around the .14 mark.

However the 1/2 mile (2nd) call in 6f races and the 3/4 mile (3rd) call in 8f races is a different story. I have not yet found a track where the time value of a beaten length at these calls does not fluctuate wildly form race to race. Values typically range from .09 seconds per length to .30 seconds per BL.

Interestingly enough, the values at the 1/2 mile (2nd) call in 6.5f races and 1 1/16 races are useable at all but 2 tracks.

When I use the word useable (??) I mean useable for what I do. I define useable as accurate to within .20 (1/5th of a second) 95% of the time. When I am calculating an individual horse's internal fraction this means that I can count on it being accurate to within .40 seconds, which is a universe as far as pace handicappers are concerned.

The difference in reliability is obviously a function of the chart callers expertise (or lack thereof), but the consistent inaccuracy of ALL beaten length calls that are 1/4 mile or less from the finish (at standard mile circumference tracks) is attributable almost solely to the angle the chart caller is forced to view that particular call from. There are a couple of compounding factors that contribute to the extreme margin of error at play here:
1) Horses tend to fan out across the track at this point
2) A horses beaten lengths at a given call are not the chart callers estimate of that horse's actual beaten lengths, but rather the sum of all estimates (and the error within those estimates) of beaten lengths for each horse preceding him.

Again, demonstrating my lack of a life, I conducted an experiment in 2000-2001. With permission (a disclaimer) I videotaped racecards at Emerald Downs, Turf Paradise, Portland Meadows and Northlands Park (a bullring). I stationed myself at the outside rail timing light for the 1/4 mile call (1/8 at Northlands) and pointed my Sony at the inner rail timing light as the horses passed. The best-fit time value for a chart beaten length was .15 (average of 31 races). Given this, I can state the following with certainty:
For all horses whose chart beaten lengths were between 6 and 7 lengths - 40% were in actuality <3.5 lengths or >9 lengths actual beaten lengths.
For all horses whose chart beaten lengths were >7 lengths - 55% were in actuality >3 lengths +- their published BL.
Even up-close horses had issues, for all horses <3 chart beaten lengths 35% were >1 length +- their chart beaten lengths.

This is not meant to question the competence of chart callers or the validity of published PP data. Making an accurate call of beaten lengths from the angle presented from the top of the turn home is quite literally impossible. And if you understand why the 1/4 pole angle presents difficulties, I am certain you have ascertained that the stretch call beaten lengths are at best a very rough guesstimate.

Now, bookend these two beaten lengths calls (as is required by the "golden 1/8th" calculation). Provided your horse is not on the lead or within a length of it at both of these calls, you will be very lucky if you accurately estimate the horse’s fraction to within 2/5ths of a second. Given that this is only 1/8th of a mile, .40 seconds is a pretty liberal margin of error.

I am not debating that fact that this method works, but estimating this particular fraction from PPs is (IMO) guesswork at best.

[JustRalph]

They will find a way to make this a premium service.....

From November 2002
Sports http://www.kentucky.com/mld/heraldleader/sports/4594245.htm

Posted on Sun, Nov. 24, 2002

HORSE RACING NOTEBOOK
GPS technology tested to enhance handicapping
By Maryjean Wall
HERALD-LEADER RACING WRITER

Racing's future came by satellite to Keeneland last week and looked like -- NASCAR?

You bet. The lead horse careened at 35.6 mph around the "fourth" turn. Three horses rounded a virtual Keeneland course as dots on laptop screens, individually timed and separated by the right amount of lengths.

Puts a whole new slant on pace handicapping, doesn't it? Tracking horses by GPS -- global positioning systems -- is the way that race charts of the future will be mapped. It's also the way races might be shown to viewers in the future.

Auto racing fans will be excused for yawning here, for the technology already exists in that sport.

But horse racing still clings to the old-fashioned way: tracking race participants through the eyes of a trained observer known as a chart caller. The caller watches the race live from the press box, dictating observations to an assistant or into a tape recorder. The caller estimates the running positions of horses at designated poles while also estimating the lengths separating them.

The system is not without flaws, as you can imagine. The marvel always has been that chart callers are able to do as well as they do. Chart calling is an art even if it is destined to become arcane.

In this post-modern phase of the space age, outsiders probably would be shocked to learn that horse racing still does not benefit from available technology. The difference is like typewriters to computers. There is a huge gap.

Horse racing is working on catching up, however. Sportvision, a broadcast enhancement company participating with NASCAR, is also working with horse racing with Equibase, the supplier of data to the horse industry.

The two companies are hopeful that horse racing will be able to offer patrons similar enhancements to those available in NASCAR on its pay-per-view channels. Enhancements could include a channel to focus on an individual horse during a race or a channel to identify horses with names and arrows as they race in the pack. These enhancements would move in real time.

Equibase, Sportvision and InCompass, which like Equibase is a subsidiary of The Jockey Club, were at Keeneland last week for testing during morning workouts. The technology has been tested several times at Keeneland and, according to Equibase President John Ertmann, "the technology and economics are to the point where they're getting more viable."

A problem unique to horse racing has been reducing the weight of transmitter equipment worn by the rider. Horse racing, unlike auto racing, is a weight-sensitive sport.

Antenna components are contained in a vest worn by the rider. The vest currently weighs slightly more than 1 pound. The final model will be smaller than the testing model and will go under a jockey's racing silks.

Last Wednesday, testing on several horses included three from Jeff Thornbury's stable. Information that showed up in real time on laptop computers included the distance traveled by each horse, the individual times of the horses, the distance they raced off the rail and the distance separating each horse from the other.

Trainer John Ward, who also has supplied horses for GPS testing, gave the technology two thumbs up and said that "it should give the modern handicapper who has a feel for technology some exact science for handicapping a race."

[hurrikane]

MV,
thanks for confirming what all of us know deep down inside...the data provided by equibase is poor at best.

Ralph,
the thing is...90% of the betting public doesn't care. They won't take the time to use it and they won't understand how to interpet the data(they don't know the data that exists now). 10% are just geeks that will evaluate the data to excess with no consideration for physicallity, connections, conditions or the business of racing. So, IMHO, the only thing it will do is give you a little more confidence in the numbers. I'll bet favorites still win 33% of the time and only 20% of the people really make any money at this.

Just my opinion

[tcat]

If you are serious about Pace Handicapping, you need to be reading the Ken Massa newsletter. His clients get to read it immediately, but non-clients may read it a month late. However, if you are serious, they are priceless. Why he doesn't charge more, I'll never know. There are people out there charging a bunch for stuff that doesn't come close to statisical research and straight talk that Ken publishes, especially pace handicapping.

[GameTheory]

MV --

Since you've done this research, maybe you'll be in a position to answer the age-old question, "How long is a length?" I've done some similar (but much less in-depth) research with the video replays trying to determine this.

I don't quite understand what you're getting at with your time per length values. It sounds as if you're expecting/hoping for there to be a constant time value per length for different races & the calls within. But why should this be so? Races are run at different speeds & early calls are faster than later ones (generally), so the time per length will (& should) vary. I may be misunderstanding you.

Anyway, what should be constant is the number of feet in a length. It isn't, of course, but I'm interested in what the best value to use might be. Most people use 10 feet per length, probably more for ease of calculation than anything. Charles Carroll argues for 8 feet per length, which is claims is the average length of a horse, but most thoroughbreds are at least 9 feet so I'm not sure where he come up with this. But what is important is what is the best fit for the average result chart, not the actual length of the average horse.

Have you ever tried to calculate that?

[keilan]

– Interesting to read that you spent some time at Northlands in 2001 timing races and comparing your results to what was posted in the DRF.

To bad Les Butler (GM) didn’t inform you that the clock was malfunctioning most of the year and they relied mostly on hand timers upstairs that were frequently incorrect. Don't feel bad no one else knew either.

The maintenance crew in early June accidentally severed the underground cables and it wasn’t adequately repaired until late that fall.

I did pace numbers at that time and realized within 10 days into the meet that there were timing problems, I had more than a few meetings with Northlands Management and the Federal Body that over-sees wagering in Alberta and both were well aware of the potential implications this inaccurate information had on the general public and they took months to correct the situation.

If you are ever this way again send me an email in advance.

[Pace Cap'n]

When it comes to analyzing chart calls.......

YOU DA MAN!!!!!!!!

[pic6vic]

As far as I am concerned (for what it's worth)

This info if valid is much better than picking winners, especially in today's pic3 and pic4 atmosphere.

Thanks

[Dave Schwartz]

MV (et al),

Very interesting story. Thank you.

We actually tested something years ago in a similar fashion. Recall when there were 2 "forms" for awhile in the early '90s competing. One was DRF and the other was... Racing Times(?)... I think.

Anyway, the source of one was EqB and DRF was (obviously) the other source.

One of our users had provided us with track videotapes (recorded from a TV replay show I think) of an entire meet in N. Califronia. We were interested in answering the question, "When these sources had drastically different calls for a single horse, who was right most often?"

We defined "drastically different calls" as more than an 8-length difference at the pace call. I know that sounds crazy but there were a LOT of them when you compared the two data sources side-by-side.

After spending a couple of hours per day watching videos and making notes, the guy that worked for me concluded that when there was one of these drastic differences, about 1/3 of the time one could just not tell from the video who was right or wrong! The rest of the time DRF was right about 2 times out of 3.

My real point is that we got a healthy new respect for chart callers, as I am sure you did as well. By 8 lengths or more! Hard to imagine that such a mistake could be made so often.

Regards,
Dave Schwartz

[MV McKee]

GT,

Actually, I have no idea how long a length is, but would assume it varies radically depending on what point a horse is in his stride (fully extended or recoiled).
As for the time value per length, I did expect (and found) that the time value of a race chart length does differ from call to call, I was simply trying to find what, and how accurate those values are.
So in answer to your 2nd question, all I ever really tried to do was correlate an average chart length to a time value.
I learned that finish beaten lengths are generally very accurate, beacuse they are calculated from the photo finish film strip, utilizing a device that is (in essence) a ruler to determine the beaten lengths. You get the occasional gross error on badly beaten horses, but I'm generally not creating a figure for those types anyway. But the time value of a finish beaten length varies quite a bit from track to track (which surprised me). I thought perhaps SoCal horses may have finish beaten lengths time values that were smaller than their NoCal counterparts because they do actually finish slightly faster (say 50fps v.s. 48fps). But at Emerald Downs, the time value of a chart beaten length was smaller than any Cali track (.16 Seconds), and I don't think Washington horses are smaller than Cali horses, and I know they aren't finishing faster. After asking someone who would know, I found out that the speed at which the film passes in front of the Photo Finish "slit" can (and does) vary. Doesn't have any ill effects other than to slightly compress or elongate the image of the horse that appears on the film strip.
So I don't have a universal number I can use at all tracks as a Chart Finish Beaten length time value, but do have an accurate time for each track I play.
For reasons I elaborated on in my previous post, I can't use the 1/4 mile (to run) or stretch calls to produce pace figures, so I resorted to doing something that likely seems bizzare to most. I take my pace calls at the 3/8's or 5/16's (to run) poles. The camera angles on these poles is almost dead on. In other words, the inner rail and outer rail timing lights are very closely aligned from this angle. It's drudgery, but I go to the video tape, I have a character generator and editing suite that I use to superimpose a static line and a stopwatch function onto the screen, in stop frame mode I note the stopwatch time when the 1st horse's nose hits theat line, and the behind time of each subsequent horse ( provided they are in the picture). Since I make the assumption (unless obviously incorrect) that the official race final time is correct, it is very easy to calculate the fraction of the race by subtracting the final fraction (3/8's to finish) from the posted final time.
Again, it sounds laborious, but only takes 5-6 minutes per race at most.
Caveats are that I sometimes do not get a video tape copy of a race, or the camera work is bad. In these cases I am left without a comporable pace figure for those races. It also is not a lot of fun to return from a business trip or vacation and have to do this for 10-20 racing days (yawn!)
If they ever do utilize the GPS system described in Just Ralph's posting, I would be in 7th heaven.

Keilan,

I wasn't too worried about the times at NP, since I was only up there for a brief time and not playing the meet full time. Were the handtimes pretty innacurate? I always wondered how much handtiming error was attributable to the actual stopwatch operator, and how much was due to the flagdropper.
I am in Edmonton at least once a year usually early fall, near the end of the NP meet. We are doing some research work with Royal Alexandra that I anticipate will be ongoing for at least 5 more years, so I am certain that I will be up this fall. I always combine the business trip with a week at Jasper Park Lodge. I'll let you know when I am headed that direction.

[keilan]

There’s a ton of stuff going on here during the summer, never a lack of things to do. I look forward to it; I’ll show you around.