04/02/2014 12:34PM

Keith Gisser: Predicting the future can prove difficult

Stan Bergstein produced the Experimental Ratings, a list of the projected top 3-year-olds, for over 45 years.

When harness racing lost Stan Bergstein in 2011, we not only lost an iconic ambassador of our sport, but we also lost the developer of the most talked about predictions in the game. Stan’s Experimental Championship Ratings, which debuted in Hoof Beats magazine in 1965 and ran for over 45 years, created buzz and generated conversation among horsemen and fans like little else the sport has ever seen.

The main reason these ratings were so anticipated each year goes back to that very first set of predictions. Stan projected his best horses AND their projected best winning times. In 1965, it was no surprise that he picked the reigning freshman pacer of the year, Bret Hanover, to top the list. His 1:55 time prediction did surprise some people. That would be a world record. So, when Brett finished his sophomore season with a mark of 1:55, Bergstein became an instant genius. (Okay, we all know he was a genius well before that, but it certainly gave his new “experimentals” instant credibility).

[DRF HARNESS: Get Free Harness PPs when you wager with DRF Bets]

It’s an interesting phenomenon. Not to take away from Stan’s expertise, but many years ago when I “predicted” the NCAA brackets using harness racing angles and proclivities and nailed the Final Four, the prediction method became an annual article in my days at Northfield Park. It was rarely very accurate (and never as accurate as that first year), but because of the initial success, the local and industry media still requested it. And if I left Arizona out, I was sure to get an e-mail from Mr. Bergstein himself, reminding me that the Harness Tracks of America headquarters were just down the road from the Wildcat campus and that I had made a mistake. I began teasing him similarly when his rankings excluded one of my personal favorites.

Another good example of the “lightning strikes” phenomenon is the now inactive website Bracketology101. Craig and Chris were just two guys making NCAA predictions on a blog, but suddenly, one year, they outperformed ESPN’s Joe Lunardi, CBS’s Jerry Palm and every other expert in the bracketology world. They became media darlings and suddenly had the second most popular bracket prediction site in the world. But their performance lagged over the next few years, and by the end of 2011, citing family obligations and their inability to duplicate that one great success, they were gone.

Stan Bergstein, on the other hand, had many very good years and some that were not good. And 46 years counts for something.  Regardless of the previous year’s results, as I got more and more involved in harness racing, I could not wait for those spring issues of Hoof Beats. In my brief training career, in the late 70s, they were the two issues that disappeared from the tack shop the quickest and that also showed up in the track kitchen, accompanied by heated debate from trainers, owners, turf writers and even caretakers.

So when we lost Stan, Hoof Beats Executive Editor T.J. Burkett had a problem, as he explains, “When Mr. Bergstein died, I felt it would be a disservice to his memory to try to replicate his list each year. But I knew the Pacing and Trotting Preview issues were some of the most anticipated each year, so I decided to keep the conversation going by coming up with a new way to look ahead to the three-year-old pacing and trotting divisions.”

He contacted David Siegel at TrackMaster, who came up with a “regression analysis” using the company’s proprietary Speed Ratings.

“I edited Stan’s rankings for four years, but never did he tell me what criteria he used.” Burkett continues. “I assumed he made some phone calls to trainers and used a lifetime of harness racing knowledge to come up with his list each year. We chose Speed Ratings because they already accounted for variables such as track size, race type and quality of competition. It was objective, which differentiated it from Mr. Bergstein’s rating.”

Hoof Beats published these ratings in 2012 and 2013 and feedback was pretty negative.  These were computer rankings, and as with minor league, err, college football, they received a BCS-like response. And, like the BCS finally got around to doing, a change was made. But Burkett acted much more quickly than the big football schools. He pulled the plug.

“People didn’t like how a computer was doing the rankings, instead of a person. They didn’t think the rankings were successful in predicting the top horses; and they thought the Speed Ratings were arcane and hard to understand,” Burkett explains. So he went back to the drawing board.

He felt that the Speed Ratings were a viable part of the rankings, but that they did not paint a complete picture. So he approached a group of 37 harness racing handicappers (full disclosure, I was one of them) and track personnel, asking them to rank their top 10 by predicted earnings in an online survey. More than 200 harness racing fans completed a similar online survey on ustrotting.com, predicting their top 10 trotters by earnings. Once the polls were closed, the data was tabulated. TrackMaster had an actual  top 25, but  the experts and fans each picked a Top 10, with the top 25 point earners (10 for 1st, 9 for 2nd on down to 1 fort 10th) being ranked.

“While the best horse in the division isn’t always the richest, he often is. Couple that with the fact that the biggest races tend to have the biggest purses and the horses with the biggest wins tend to earn the most money,” Burkett explains. “We asked TrackMaster to reconfigure its analysis to predict earnings, and combined with the expert analysis and fan poll, we got what we call the Hoof Beats Comprehensive Predictive Ranking.”

While they are not perfect, I like the rankings for a number of reasons. I like the fan input. I like the fact that with increased Sires Stakes money in some states, regional horses still earned some consideration. l also like the fact that we are getting broader input from more sources, which at least in theory means a more accurate ranking.  Now, that can also be a problem with accuracy, as any fan could vote regardless of his or her level of knowledge. The TrackMaster input will only be as good as the formula they computed, which remains to be seen. However, their overall track record is pretty good.

[DRF HARNESS: LIKE us on Facebook and get timely updates on the latest harness news]

I am also glad that Burkett and company plan to track not just the comprehensive rankings, but each segment. That’s not just for potential bragging rights (although I am sure at least one of the 37 of us “alleged experts” will be touting his perfect bracket come December). I would think that (not after the first year certainly) a weighting system might become viable for each segment, or maybe the addition of a horsemen’s poll instead of, or in addition to, the fan poll.

Still, T.J. Burkett and USTA Communications Drector Dan Leary should be applauded. They had something that, by their own admission, was not working and they tried to fix it. Whether Father Patrick or the filly Shake it Cerry (who tied for first in this year’s trotting rankings, with Father Patrick winning a tie-breaker for the top spot) validate their lofty rankings remains to be seen. Perhaps one of the other top ranked horses or some unheard of, unraced-at-two star comes out of nowhere to take divisional honors. Regardless, we have a viable ranking that will be fun to argue over adult beverages. And that’s the best part. Now go cash.