07/21/2005 12:00AM

Software yielding some hard results


DEL MAR, Calif. - The festive opening week of the Del Mar summer meet may seem an unlikely occasion to address the cold reality of handicapping by computer. After all, the best action is out on the racecourse, not buried on some anonymous hard drive.

Blame it 77-year-old retired engineer Rubin Boxer. He is the creator of CompuTrak, a software program that handicaps an entire card in only seconds. It does not sound like much fun. How could it? Just download past performances from the Internet, hit a few computer keys, press the "print" button, and presto - the day's selections.

The whole process takes about a minute, and most self-respecting horseplayers might be reluctant to use a program as fast and mechanical as CompuTrak. Chances are, the quick-and-easy process would provide ordinary results. But in a five-week study during the 2005 spring-summer meet at Hollywood Park, the results were anything but ordinary.

From late April through the middle of June, in a survey over 22 racing days, CompuTrak produced results that would be the envy of any flesh-and-blood handicapper. Using Daily Racing Form past performances, the program selected 54 winners from 168 races (32 percent) at an average win payoff of $7.30. A $2 win bet on each selection ($336 total) produced total return of $394.40, a $2.34 return on investment for each $2 bet.

Are the short-term results a fluke? Boxer admits "my knowledge of statistics says it would be more meaningful if there were many, many more races."

In March 1992, Phillips Racing Newsletter reviewed CompuTrak over 17 days at Santa Anita. The program produced 36 percent winners and a $3.93 ROI that is almost too high to believe.

CompuTrak is not the brainwork of a mad scientist or degenerate horseplayer. It was created by a respected family man who had never been to the races until the late 1980's. Although engineering is based on black-and-white mathematical computations, Boxer had an idea that "you could use engineering for living things."

Boxer created a software program that predicted what a person's weight would be on a week-to-week basis. The program, called "Cut the Baloney," was distributed to relatives and friends who proclaimed that it worked. The idea did not catch on commercially, but it gave Boxer confidence that engineering principles had biological applications.

A chance outing to an offtrack wagering facility piqued Boxer's interest in racing. He wondered if it was possible to apply principles of engineering to Thoroughbreds. He recognized his shortfall going in: "We're not horse racing handicappers."

For nine months, Boxer worked on a research paper using mathematical principles as they related to the performance of a horse. As Boxer said, the idea was to determine "whether an engineering approach would be useful for analyzing Thoroughbred horses." Boxer's intention was merely to publish a research paper. Marketing a software program was the farthest thing from his mind.

The paper - "Engineering Analysis of Thoroughbred Racing" - was not published in the trade publication for which Boxer originally intended. But he said the method worked, even in the archaic days before CompuTrak became Windows-compatible. "The original paper, you'd have the Daily Racing Form on the table with you, hand-enter all the times, weights, and variant, and then the program would run."

Boxer used "Ainslie's Encyclopedia of Thoroughbred Racing" as his racing reference while writing the original CompuTrak program. One premise of the program is the concept of "friction" between a horse and the ground it travels over, based, among other things, on the weight the horse is carrying. The program also considers energy reserves, rate of deceleration, and track variants.

CompuTrak works on any track in North America, and was designed to appeal to a wide spectrum of horseplayers - from those who buy selections of published handicappers to sophisticated bettors who make their own selections. Reports generated by the program vary from a simple odds summary that includes an entire card on two pages, to extensive multi-page reports with detailed examination of individual races.

Most bettors do not believe handicapping can be pigeonholed into a computer-generated selection. And there is doubt that any program can consistently hit winners at 30 percent. But for 22 days at Hollywood, CompuTrak illustrated the value in considering ideas such as those "engineered" by Boxer.

The program picked at least one winner each day of the study, and achieved profitability on the third day (May 5) by picking five winners in seven selections on the card. The one race it did not handicap was a race in which half the field was first-time starters.

Early this year, Boxer went worldwide with his website, www.revelationprofits.com. CompuTrak currently has about 1,300 users. If the program continues to have success and catches on with more bettors, is it possible that the law of diminishing returns will kick in?

Boxer chuckled at the idea.

"I'm wishing for that day," he said, observing that that bettors "are so widespread, over so many tracks," the scenario is unlikely.

Yet who knows? CompuTrak already has proven an unexpected source of high-odds winners. The study at Hollywood ended June 16. A month later, CompuTrak picked up where it left off. On July 17 closing day, the program picked only one winner, in race 1.

She was Bright Design. The top choice by CompuTrak, Bright Design paid $93.80.