Blind allegiance to the religion of analytics is harming sports.
There is nothing wrong with basic statistical analysis, in any aspect of life. But, when it is taken to the extreme and viewed as infallible, that’s when the problems begin.
It was only natural that we’d get to this point. Nothing succeeds like excess, in personal life and in sports.
Many trace the birth of the analytic movement to the Oakland A’s and the Moneyball concept, in which extreme reliance on statistical measurement allowed a small-market, limited-budget franchise to win more than its share of games.
It is important to note, however, that the philosophy never produced much in the way of postseason success and more important, ZERO championships.
A lot of the current analytics stress is on taking outsized chances, like swinging for homeruns almost every at-bat in baseball, or trying to convert fourth-down opportunities an inordinately high amount of times in football.
For a real-world example of how such things can fail catastrophically, consider the case of Long Term Capital Management, which nearly brought down the financial system in 1998.
After producing outsized gains in its first three years, Long Term Capital lost $4.6 billion in four months of 1998, necessitating a bailout.
The high-leverage strategy (investing multiples of actual capital) worked until it didn’t. That and failing to consider unlikely, but possible negative outcomes, doomed Long Term and almost the entire system.
Move to sports and the failures of analytics have been on stark display recently, if one bothers to look.
It is no mere coincidence that we’ve seen the two longest scoreless games in the entire history of postseason baseball just in the past week.
First, Cleveland and Tampa Bay set a standard when they went into the 14th inning scoreless in Game 2 of their AL Wild Card Series.
The home-run or bust strategy didn’t produce a score until the 15th inning, when Cleveland won on a solo shot by Oscar Gonzalez.
Just two days back, the Houston Astros and Seattle Mariners went into the 18th inning scoreless, re-setting the record. This one also ended with a solo homer.
That game was an excruciating viewing experience as batters continued to overswing, chase pitches out of the strike zone and generally try to end the game with one swing instead of taking a walk, hitting behind a runner, or doing any of the other things teams used to do to squeeze out a run in a tight, low-scoring game.
On Sunday, a highly anticipated NFL game between Super Bowl contenders Buffalo and Kansas City became a showcase for extreme analytics overuse.
Buffalo, trailing 7-3 in the second quarter, passed an a short field-goal attempt and instead tried to convert a fourth-and-goal from the KC 3-yard line, failing miserably.
Undaunted, with the score tied 17-17 early in the fourth quarter, Buffalo went on fourth-and-3 at the KC 46-yard line, and failed yet again.
Blessed with great field position, the Chiefs had to move just 28 yards before hitting the go-ahead field goal to lead 20-17.
Yes, Buffalo scored a late touchdown to win, 24-20, and along the way converted a fourth-and-1 from its 33-yard line with under four minutes remaining. But this last fourth-down gamble made sense because of the stage of the game, the score, and the shorter distance to make a first down.
It could be argued that the late heroics would not have been necessary if Buffalo had not passed on a field-goal try earlier, and not later handed KC a three-point opportunity with the second failed fourth-down conversion try from near midfield.
If Buffalo had resisted the analytics siren call, the Bills already would have been up based on that six-point swing.
The analytic overkill in sports these day ignores two salient points.
One, sports are contested by humans, who can ride momentum, adrenaline, emotion, to performances that defy the statistics.
Two, being an analytics daredevil is akin to walking a tightrope with no safety net. One big splat on the pavement more than outweighs any previous successes.
Sam Ross Jr's site 