Lengthy Layoffs

[Thomason]

Steve Klein of DRF did some research a few years ago about lengthy layoffs and discovered that horses that had been off a year or more did abysmally in their return to the track. He said they were the worse bets in horseracing, regardless of class level or odds range. Has anyone done any research on horses that have been off for a long time but less than a year?

In general, I’m sure that the longer the layoff, the fewer the wins and the lower the impact value, but the class of the horse and the skill of the trainer must surely have some effect on that. For instance, a modest claimer who has been off for 3 months will probably have a tough time winning its comeback race, but that might be a typical interval between races for a stakes horse. If you broke it down into periods of a month, I wonder where we might see a significant drop in impact value for claimers, allowances horses and stakes horses, and if there was any difference in dirt and turf races.

Does anyone know of such research? Or should I cozy up to a stack of Charts Weekly and have it for a few weeks?

[pmd62ndst]

Betting on every horse in every race in my database (320,448 entries) regardless of anything will get you these numbers:

12.2 Win %. - 23.5 ROI % (320448 entries)

Now, betting on only horses that have >= 365 day layoff produced these results:

7.9 Win %. -54.7 ROI %. (2083 entries)

As you can see only 0.65% of all horses entered will have a layoff for this long, but it does look like a poor bet. I don't think I've ever seen a > -50% ROI. Impressive.

[alysheba88]

There are only two real relevant factors to me when looking at long layoff horses.One is obvious, one less so.

1. The trainer. Some guys specialize in this. Everyone knows about Michael Dickinson but there are other lesser known trainers very good at this.

2. The horse. See if the horse itself has run in the past off layoffs, And how they fared. Did they win first out.

Now there are other factors like class of the race, is the animal entered at his preferred distance, post, etc, but the above two are the dominant things I look at.

[Thomason]

PMD

Thank you for the post verifying Steve Klein's research. When you have the opportunity, would you run the same query with layoffs of the following lengths?

30-60 days
60-90 days
90-120 days
120-150 days
150-180 days
180-210 days
210-240 days
240-270 days
270-300 days
300-330 days
330-360 days

Nothing fancy, just the simple query you ran before.

My assumption is that with each period, the winning percentage will decline gradually, but at some point the percentage will take a dive. I'm going to the DRF Handicapping Expo in February and I want to talk with Steve about this topic.

Thank you.

Michael Thomason

P.S. I'm envious of your database. I use All-Ways and while I have a considerable database of races, it wasn't designed to conduct this type of research.

[pmd62ndst]

<= 30 Days: $ -100868.30 (-22.9%) 27869/219824 (12.7%)
31-60 Days $ -20068.30 (-20.2%) 6476/49619 (13.1%)
61-90 Days: $ -5234.60 (-28.4%) 1012/9211 (11.0%)
91-120 Days: $ -2136.20 (-24.4%) 480/4377 (11.0%)
121-150 Days: $ -1839.40 (-26.8%) 338/3431 (9.9%)
151-180 Days: $ -2068.70 (-30.0%) 320/3453 (9.3%)
181-210 Days: $ -2116.40 (-31.2%) 310/3389 (9.1%)
211-240 Days: $ -951.60 (-17.3%) 285/2751 (10.4%)
241-270 Days: $ -863.10 (-21.4%) 198/2012 (9.8%)
271-300 Days: $ -1046.40 (-34.3%) 144/1524 (9.4%)
301-331 Days: $ -958.80 (-44.5%) 106/1077 (9.8%)
331-360 Days: $ -761.80 (-51.2%) 59/744 (7.9%)


These stats are for all thoroughbred races that DRF formulator offers from 2/1/2003 - 11/3/2003.

PMD

[Thomason]

PMD

Thank you.

[Tom]

I posted some layoff stats for various time periods a while ago. Do a search for "layoffs" or for post by me - I had data from middle 2001 through early 2003, so the dates are differnet than these here. I think they were similar in results, though.

[Dave Schwartz]

Our system does not distinguish above 251 days. Hope this helps:

This represents all the non-maiden, dirt races, from 5-9f for older (i.e. not young) horses run on fast tracks going back to June 1st, 2001 in North America.


Days
----
WIN BETS
Field1 Field2 Starts Pays Pct $Net IV PIV
--------------------------------------------------------
1-29 274,690 34,695 12.6 $1.56 1.02 0.97
30-60 68,179 8,920 13.1 $1.61 1.05 0.97
61-90 11,252 1,206 10.7 $1.48 0.87 0.91
91-120 5,093 515 10.1 $1.44 0.83 0.96
121-150 3,892 347 8.9 $1.39 0.73 0.93
151-180 3,203 248 7.7 $1.20 0.63 0.78
181-210 2,907 211 7.3 $1.04 0.59 0.74
211-240 2,376 219 9.2 $1.38 0.75 0.92
241+ 7,867 611 7.8 $1.06 0.63 0.75

[kenwoodallpromos]

The steady % wins after long layoffs I assume help indicate the trainers make a big difference instead of the type of injury.

[pmd62ndst]

Dave,

One strange similarity between our findings is the increased ROI and Win % for layoffs between 210-240 days. Why?

I was looking for a couple 99-1 longshots that always seems to be there to screw up a perfectly good linear model, but I don't think that's the case here. The increase in Win % seems to be legit but I can't figure out why.

Any ideas?

PMD

[Rick]

Looks like there's not much advantage in using recency as factor. Only about 10% have been off longer than 60 days and the 61-150 group doesn't look all that bad either. But, maybe there's some value in looking at how recent the 2nd or 3rd race back is to take into account the generally accepted idea that horses run better in 2nd, 3rd, and 4th races after a layoff.