Tuesday, February 7, 2017

Survivorship Bias In Super Bowls; It's Why in Horse Handicapping We Need to Study the Losers

There, as often happens in sports, was quite the discussion about play calling, play execution, and just about everything else after the Falcons blew their massive lead on the weekend in the Super Bowl. This is, generally, what we as sports fans do when something weird like this happens.

"If team X [ran the ball], [passed the ball], [killed clock] etc, they would have won the game!"

Maybe if one or two of those things happened, yes, this startling win might've not occurred, but this analysis sometimes has a built in error with something called Survivorship bias. This is, from Wikipedia: "the logical error of concentrating on the people or things that "survived" some process and inadvertently overlooking those that did not because of their lack of visibility"

We see the success of something (a big comeback) and start attributing, because the success is apparent. What we don't see is the times this success was not achieved (in this case, say for example, a team staying aggressive and running out the clock so the other team never even gets a chance to possess the ball twice) and weigh the other side.

Examples of this from sports are plentiful.

When a 3rd and one pass works, "great call, they caught them off guard."

When it fails: "sheesh, run it, you only need a yard!"

In the Super Bowl itself, something occurred that illustrates this pretty well.

In the Super Bowl in 2013, late in the 4th quarter, with the game on the line, coach Pete Carroll called a pass play from the one yard line. This is not uncommon (coaches have called a pass play about one of every four plays from that spot), and has a 48% success rate from the goaline. And it's often used as a set up play, to keep the defense guessing.

On Sunday, Bill Belichick called a pass play, in overtime, from the one yard line against Atlanta with the game on the line. Again, not uncommon.

Now, guess which one is remembered as "the worst call in NFL history". Yes, you're correct, it was the play that went wrong, not the one that's "disappeared from view."

In horse racing we as bettors do this all the time.

When a jockey switch occurs and we bet the horse, we say "that 8-5 was a GIFT!" for everyone to read, or hear at the OTB. However, perhaps if we run the numbers, the last six times the same switch occurred it resulted in a loss. When's the last time you've heard a positive rider switch get commented on when the horse loses?

What we need to do is look at the angle, or move, in totality, and then figure out if it's profitable. That involves, like in sports betting or Super Bowls, analyzing the losing alternative. We learn much more from what people aren't talking about, than what they are.

As for the Super Bowl, a sound argument can be made that it might've been statistically best for Atlanta to run twice and kick a field goal, but it's statistical hair-splitting. And of course, if there were two incomplete passes - in the absence of holding calls and sacks - followed by a made field goal and a Falcons victory, no one would even be talking about it.

Have a nice Tuesday everyone.

 

No comments: