Computer program key to McKeever’s success

There is a new prominent player in the contest world, and his name is Scotty McKeever.

McKeever, 52, is familiar to many horseplayers from his work on TVG. He played in his first contest in January, winning more than $90,000 in the Pegasus World Cup Betting Championship, then came back last weekend and took home prizes worth more than $20,000 in a $500 buy-in contest at Santa Anita.

“I wish I’d known how well tournaments played to my personal style,” he said. “That information would have been helpful years ago. I’m always trying to win, and I’m not afraid.”

The Newport Beach, Calif., resident uses a computer model to assist in his horseplaying, and unlike many others who use models, he is willing to talk about the process.

“Machine learning isn’t new, but it’s become more advanced,” he said. “Our algorithm gives a win-percentage chance to every horse in every race, and our top contender wins 36 percent of the time.”

He is part of a website called EquinEdge.com that provides this information to the public. Currently, the site offers a select amount of free information, with plans for a commercial release as soon as April.

“It’s designed for beginners as well as experienced players,” he said. “We want to empower players and give them a quick way to get an advantage.”

As McKeever has demonstrated recently, this could be particularly useful for tournament players. Though, of course, there is likely to be a learning curve.

“At the Pegasus, I was the only player to play every race,” he said, “which in retrospect was an error.”

After a hot start, he missed five bets in a row. Then he connected with a $500 exacta in the Gulfstream Park Turf Sprint that keyed logical 3-1 Rainbow Heir over 35-1 lonsghot Oak Bluffs and returned $60,000.

It’s fascinating that McKeever has found a lot of success in races with first-time starters, not a situation where many players would guess a computer-oriented bettor would thrive. He does this by looking at breeding stats and workout patterns.

“I hear players say all the time that first-time starters are impossible to judge,” he said, “but through machine learning, there are identifiable patterns. I like to see a fast fivefurlong work, then a slow fourfurlong work. I want to see that they started earlier with threeeighths works. First-time starters only win 9.5 percent of the time, so you’ve got to figure out which ones have a chance.”

Indeed, a firster was his key to victory at Santa Anita, with McKeever once again keying a logical over a longshot. The longshot firster was 36-1 Coco Kisses (No. 7) in the eighth race. His bankroll was down to $51 heading in.

“The 10 horse [Karmic Affinity] was the top pick, but the program doesn’t just rate the likeliest winners, it rates every horse in the race, and there were a few live longshots,” he said. “I played an ice-cold tri, 10-7-3, and it paid $10,000.”

McKeever works closely with a programmer, spending at least a couple of hours every other day with him.

“After a race, I go back and look to see why a certain horse might have won. We go over factors, he puts in the code, and we look to see, ‘Does this factor improve the algorithm?’ ”

It’s interesting to note, however, that his process doesn’t end with the model. He uses it to eliminate horses who don’t have much of a chance, enabling him to focus on the contenders.

“I use the Racing Form with my other tools, and after eliminations, I look at maybe half or one-third of the field, and that’s a huge advantage,” he said. “The Beyer numbers are great at judging past races. My win percentage allows me to judge how that horse fits in that race today. Using both, I can quickly process information and determine what I’m going to play.”

Every cynical horseplayer reading this is probably thinking the same thing: If McKeever can make money betting with these numbers, why doesn’t he want to keep them for himself?

“I’ve fought with that,” he said, “but that’s just not who I am. I love handicapping, love the game. I want to share my work and try to make a difference. If we’re going to attract millennials to horse racing, they have to be able to process information fast and have a fighting chance of winning. They’ll read the Form, but they’re not going to read it like we read it.”