o Analytics Don't Tell the Whole Story - Bernies Blog,
Track Times Today

Analytics Don't Tell the Whole Story
Tuesday, July 11, 2017

    Analytics is defined as the "discovery, interpretation and communication of meaningful patterns in data. Especially valuable in areas rich with recorded information, analytics relies on the simultaneous application of statistics, computer programming and operations research to quantify performance." Whew!

    One of the simplest applications of analytics in sports today involves baseball. For some reason known only to the analytics nuts, pitchers are only allowed to throw 100 pitches nowadays (give or take a few), even if they're in the process of fashioning a no-hitter or a shutout. I would love to know what the percentages are for the number of times the bullpen has blown the game for the guys who left after seven innings with the lead after having given up none, or one, or two hits. In the NFL, I would love to know how many times a kickoff returner doesn't make it back to the 25-yard-line. Those are a pair of stats we don't hear about, though. 

    Perhaps no other sport has as many analytics gurus as thoroughbred racing. One of the strongest points that possible newcomers make for their reluctance to get involved is that handicapping is too complicated. There are Thoro-graph numbers, the Ragozin sheets, Beyer numbers, Equineline figures, Equibase figures and probably a dozen more. I have always wondered, if any of these shortcuts to picking winners is so good, why isn't everybody using the same one and making millions? And if they're so good, why are the inventors giving out the info instead of just betting on the winners themselves? 

    My latest foray into the world of analytics came this week when I checked out a betting column on the Twin Spires ADW. I have always been aware of the first column, which lists the morning line of each horse. And the second column, which posts what the odds of each horse are coming up to post time. But I never paid attention to the third column, which always had numbers that may or may not have mirrored one of the other two columns, most often, not. 

    A few days ago I clicked on the third column and found that the third column was reserved for "profit line odds," which I found is akin to the people who decide that analytics can tell what a team's chances are of beating another team in the playoffs.

    Profit line odds "represent each horse's estimated probability of winning based on a rigorous computer study encompassing thousands of races. Profit line odds incorporate numerous handicapping factors (speed, class, pace, form, weight, distance, surface, trainer, jockey, pedigree, recency, etc.) The horse with the lowest profit line is deemed to be the most likely winner according to the profit line." 

    Very neat, and concise and to me, a lot of baloney. Where are the analytics that show the percentage of times the horse with the lowest profit line actually won the race. That's a number worth knowing. I haven't seen it anywhere. And, how does the profit line factor in the horse getting left at the gate, or the rider running up on the inside after getting left and burying his horse in a 21-plus quarter, or the horse being carried six wide on the turn by a bolting rival, or the fact that the horse was fouled in his last three races, or any one of a great many other problems not listed in the explanation.

    A computer can accomplish many objectives, but picking out the winner of a horse race is not one of them. Otherwise, we'd have a 1/9 shot in every race. Anyway, the use of speed, class, pace, form, etc. is just what a good handicapper determines when he goes over the PPs in the Racing Form. I would rather Twin Spires left out the profit line and used the space instead to give me the jockey and pedigree of each horse, like DRF/ExpressBets does. 

       

    

Leave a comment below
comments powered by Disqus