09/30/2005 12:00AM

Examining turf front-runners at Keeneland

Email

LEXINGTON, Ky. - The bias of the main track at Keeneland is a popular topic in early April and early October. But the bias trends on the turf aren't discussed nearly as much. The reason is that most handicappers agree that early speed is usually the preferred running style on the dirt, but the turf bias trend isn't nearly as clear cut.

Turf sprints at Keeneland run at 5 1/2 furlongs have produced 15.6 percent front-running winners, with a $1.80 return on investment from a group of 32 races. It isn't surprising to see that horses with early speed win back part of the parimutuel takeout at that abbreviated distance.

The longest commonly run distance on the turf at Keeneland is 1 1/2 miles. There were only four early leaders who held on to win at that distance from a group of 63 races, for 6.3 percent wins, and a low $0.88 ROI. I wasn't surprised to see that result, since my experience in watching turf races at that distance at Keeneland over the last 11 years has been that horses with early speed usually need to get a clear lead through slow fractions to have a realistic chance of winning, and even with that comfortable trip, many of them are still caught by off-the-pace runners in midstretch.

Most of the horse racing research I study yields a fairly logical pattern of results. I don't mean to say they are so logical that the answers were easily predicted. Unexpected outcomes are common, but once they have been discovered, they tend to be consistent enough to follow logical patterns. That is a good thing, because answers that are counter-intuitive are more valuable discoveries than the answers that are easily anticipated by bettors who haven't bothered to do the work necessary to dig them up.

Some categories of results prove to be small exceptions to the larger trend, but in the end, the story the numbers tell is usually logical and informative. The tricky part of dealing with the mathematics of horse racing is trying to interpret the statistics that don't tell a logical story. That is the case when you study the races run at one mile, 1 1/16 miles, and 1 1/8 miles, which are the three most commonly run route distances on the grass at Keeneland.

There were 136 one-mile turf races in the group I studied. The win rate of the first-call leaders was a solid 14 percent, with a $1.91 ROI. The sample of 148 1 1/16-mile turf races produced a different result: just 9.5 percent winners and a $1.12 ROI. When I saw those numbers, and I recalled the results of the 1 1/2-mile turf races, I was temped to guess that as the distances of grass races at Keeneland increased, the win percentages, and ROI's of the leaders would probably decline.

That was a good guess, until I saw the results produced by the 1 1/8-mile turf races. The win percentage of the early leaders rebounded to 14 percent, the same number seen at one mile, and comparable to the 15.6 win percentage in 5 1/2-furlong turf races.

The catch is that the ROI jumped to $3.07. That was unexpected.

The most tempting explanation was that a couple of winning longshots distorted those numbers. I tested that theory by figuring out what the total payoffs would have been for the early leaders in 1 1/8-mile turf races if they had produced the same $1.91 ROI as the one-mile races did. The answer was that the front-running winners of the 1 1/8-mile races paid $273 more than they should have, much more than a couple of extra longshot winners would explain.

The matter was complicated further by the fact that the small sample of 16 races at 1 3/16 miles produced just one winning first-call leader, with a $0.31 ROI.

Sometimes the easiest solution is also the best solution. I realized that the best way to understand these conflicting trends is to set aside the 5 1/2-furlong turf sprints at one extreme and the 1 1/2-mile turf routes at the other extreme, and add the results of the one mile, 1 1/16 miles, 1 1/8 miles, and 1 3/16 miles races together. This larger sample of 443 races is more reliable than the conflicting smaller samples, and the differences between them are smoothed out.

If anyone asks you how front-runners are likely to fare on the turf at middle-route distances at the upcoming fall race meet at Keeneland, you can confidently tell them to expect about 12.2 percent winners, with an ROI close to break-even, at $1.96. Just don't give them a written guarantee.