Litfin on Figure Making

[cj]

The full article can be found on the DRF site, it is for members.

* Always make a separate pace variant. I've seen some quick-figure methods that call for applying, say, two-thirds of the daily variant to come up with a pace figure, but such numbers usually do a poor job. Far more often than not, the pace and final-time variants will be different, often markedly so. My home-made concoctions for last Saturday's Fountain of Youth card at Gulfstream, for example, produced a "Fast 4" (F4) final-time variant for the day's six sprints, but sprint pace-call times were Fast 7. Two days later, at Aqueduct, routes were Slow 5 final, but Slow 9 to the pace call courtesy of a blustery headwind down the backstretch.

I think most people get in trouble not making a separate variant of using a fraction of the final variant. Its also why I don't think much of BRIS numbers. I believe (though they don't say one way or another) that they do not make pace variants separately.

Why I make figures...

* The payoff for bothering with these details comes in several ways, the most obvious being the times - few and far between as they may be - when your figures differ significantly from published figures.

[JimG]

Craig,

Thanks for posting. That was an interesting read. Made alot of sense.

JimG

[Exactaman]

Originally posted by cjmilkowski


* The payoff for bothering with these details comes in several ways, the most obvious being the times - few and far between as they may be - when your figures differ significantly from published figures.



That's probably the payoff for any kind of handicapping.

[delayjf]

Craig,

With you use of win pars, how often does the situation that Litfin describes occur, (Pace variant faster than the final variant).
I can see how the wind he mentioned at AQ would cause this, but absence something like that.

The reason I bring this up is that MVMCKEE posted that his study have lead him to conclude that the race segment that the track variant would have the biggest impact on would be the final fraction due to the impact of not only the track speed but the endurance of the horse it's self. This makes perfect sence to me,
but flys in the face of what Litfin describes in the article. Not having read the article, makes me wonder if he's using normal pace pars times.

What were your variants for those same tracks/days?

[cj]

I will check those specific days and get back here, I didn't even think to check to be honest. When you ask how often there is a big difference, I would say it depends on the track. I could probably never make a pace variant for SoCal, and it wouldn't affect very much.

But with other tracks that are located in windy areas, it makes a huge difference. Remington, Aqueduct, Turfway, and CharlesTown come to mind right away.

As for MV, I think he is saying the variant for the entire race would have the biggest impact on the final fraction. It makes sense, but I think we are talking two different things.

[JimL]

CJ,do you make class or projected variants? I do class variants now and I know at times they are simply not accurate enough. Dont know if I could ever do projected variants. JimL

[cj]

For the first day Litfin mentioned:

He had Gulfstream sprints on 2-14 as F7 F4, I had F21 F16. He uses the Quirin scale I believe, I use Beyer, so they are almost identical.

Aqueduct hasn't been finished yet for 2-16, it will be in the Simulcast Weekly coming out in a few days.

(Since I've been in Belgium, Buddha has been doing all the variants, quite well I might add!)

JimL,

I don't make many final time variants at all. I usually use the Beyer variant which I extract from the figure of race winners. I do change them from time to time, maybe 5 to 10% of the races, when I feel they erred with one of their "projections." I make variants for the pace segments, the variants that are not public knowledge.

[Valuist]

I would definitely include Hawthorne into the "windy" tracks; Haw also has the longest stretch in N.A.

[delayjf]

Craig,

I'm sure your right with regards to various tracks characteristics. What made me questioned his GP variants was the lack of an explaination as to why a track would be faster over one portion (in this case, the backstretch and turn) than the home stretch.

I also don't necessarily agree with Litfins warning that using a % of the final variant as a pace variant does not work. According to MV one can get a fairly accurate pace variant thru PROPER segmentation of the final variant. Specifically his research of 6 furlong races with final times 1/5 second apart showed about 45% of the difference was at the 1/2 mile call. Others conductiong simular research concluded verified the same %. It would then make sence to apply 45% of the variant to the 1/2 mile call as opposed to the 66% Litfin used for example. Now, MV does not claim that the above will give you the most accurate variants but they would be accurate most of the time.

[cj]

Jeff,

I agree with the most of the time, but that is the same thing you will get with widely distributed pace numbers, ala BRIS, TSN, ITS, etc. Its those other times I am interested in.

I also believe noone is going to have perfect pace numbers, or even pace numbers as good as speed numbers. The data just isn't good enough. I still want to have the best that I can, and what your average Joe hanging out at the OTB with nothing but a program or DRF doesn't have. Most programs based on data provided from the resellers (DRF, Bris, etc.) don't have it either.

As for the why one portion of the track is faster, I don't know. In the end, all that matters is that it was faster. Could be sun vs shade, could be wind, could be the runup was longer or shorter than the usual for a given racecard, but why really doesn't matter in the end.

[sjk]

For what my opinion is worth, I think you get better pace numbers by segmenting the track. There are days when the segments are significantly different.

Even this happens only a few days out of a hundred, you can lose some serious money expecting the horses that ran that day to perform back to their track aided pace numbers.

[Tom]

I make pace variants too, and it is really a lesson in humility some days. I would guess that 80% of the time, the pace varinat could be made by just using a percentage of the final variant. That means 80% of the time time, I have no special insight into the pace of those races. It is the 20% ofless of time when the wind, moisture, tides, whatever affect the pace that I can hope to find something that isn't out for everyone to see (expect CJ, and apparently now Buddah- haha). when you see a horse at Aqu runs on the lead in 47 111.2 on week and then 45.3 111.3 the next, what do you do? These are the race I live for (and sometime die one) buyt when you are talking about gambling on your numbers vs Beyer and you are getting 5-1 and Beyer is getting 8-5, you don't have win too many to come out ahead.

[Shacopate]

Craig,

What % of the TV would you apply to the final fraction to make a late pace fig?

I checked out your website, impressive work. How much longer will it be free?

Good luck at NATO. Hope all is well.

[cj]

Originally posted by Shacopate
Craig,

What % of the TV would you apply to the final fraction to make a late pace fig?

I checked out your website, impressive work. How much longer will it be free?

Good luck at NATO. Hope all is well.

The site will continue to be free, I have no intention of making a pay site. A few guys have the actual program, Keilan, zafonic, buddha, and jmilkowki. They helped a lot with the development and testing. I gave it to them, but that's about all I can handle. Great bunch of guys.

Here is how I calculate final fraction, I only use it for turf figures.

Once the pace and final variant are applied, I have a pace and speed figure. So, lets assume a horse had P75 S85 for a 9f race. I know the pace is 2/3s of the race. So, the 3 segments of the race would have to equal 85 * 3, which is 255. Now, I know the first two segments were 150 (75*2), so the last fraction had to be run with a 105 figure.

75 + 75 + 105 = 255. Of course, I don't do this by hand, it is all programmed in. For a turf late rating, I add the 75P to 105L, and divide by two to get a turf figure of 90. This is a simplified explanation, there are a few more rules and such, like in a turf race that goes P60 S100 (it happens!), I have some limitations on how heavily the late fraction can be factored in.

Anyway...don't want to go to deep, but ask away if that didn't explain things.

[justin]

Craig's turf numbers are top notch. I used to use the method outlined in Figure Handicapping by Quinn. It worked well. Craig's method works MUCH better. Just this weekend the 1-2 contenders in the Palm Beach ran 1-2 to the tune of a $50 exacta for a deuce. The top pick was 3/5 but the place horse was 37/1 or something like that. The place horse ran a big figure (comparative to the field) in his last start which was the same distance as Saturday's race. He also did it with the addition of blinkers. I'll accept a number like that with no questions asked at those prices. The key I think is concentrating on the races at today's distance, which I think Craig already mentioned.

-Justin

[cj]

The race Justin mentions is still on the site, Feb 21 GP, until tomorrow night sometime.