Projection systems have an impossible job.
See, projection systems are always evaluated against the ideal: correctly predicting a season or player result almost exactly. It’s not really good enough that a projection system be 80% correct; we demand perfection from our computers. There’s only an eight mile distance between Kauffman Stadium and the Kauffman Center, which is not a huge split considering the earth’s 25,000 circumference, but a GPS that took you to one when you wanted the other would be unacceptable. Computers can crush the best players at chess, take us to exact coordinates on a map, and calculate pi to a million billion decimals. When they aren’t good at predicting things, it’s conspicuous.
We’ve all watched baseball, though, and we can probably all agree on one thing: perfect predictions are inherently impossible. Baseball is imminently unknowable and endlessly random, with tiny variables making outsized impacts on every facet of the game. Projection systems are doomed to fail because they are seen through this light of perfection. It is an impossible standard. It’s like expecting Taco Bell to get your order right 100% of the time; it’s staffed by high schoolers making minimum wage. It’s just not gonna happen.
So when the PECOTA predictions were released, there was tangible foofaraw in Royals kingdom. This year, the vaunted projection system predicts the Royals to win all of 71 games and cruise to last place in the American League Central. This seems patently absurd on both parts. Royals fans thought so, too; just take a look at the burning hatred for the projections and the equally searing hot takes on the Facebook comments of our article.
But there is a defense of PECOTA and projection systems in general, and it’s a pretty simple twofold. One, projection systems have a process that is equally applied to every team. That is both its greatest strength and biggest weakness, as it can’t be persuaded by personal feelings but it also can’t adapt or predict breakouts.
The second one is the kicker, though. Projections shouldn’t be matched up against the ideal, but against human projections. Predicting baseball is hard. But projections were built to try to better predict baseball, to eclipse human understanding of the game.
Human understanding of the game is usually, to put it rather delicately, bollocks. I submit to you a series of tweets from Skip Bayless, who is paid millions of dollars to analyze and have opinions about sports.
Throw up the X, Cowboy fans. This is the year.— Skip Bayless (@RealSkipBayless) July 15, 2015
The Houston Texans will forever regret it if they do not take Johnny Manziel with the No. 1 overall pick. He will haunt them.— Skip Bayless (@RealSkipBayless) May 9, 2014
Derek Fisher will prove to be a much better NBA head coach than Steve Kerr.— Skip Bayless (@RealSkipBayless) June 10, 2014
There is a double standard which exists, and that is that human guesses, especially based on ‘gut’, are never challenged in the same way as projection systems. A gut prediction can be skewered, but the process behind making that prediction never is. On the other hand, a projection system result is often ridiculed at the same time that its process or reasoning is.
So I have a challenge for all of you who have ever thrown a fistful of insults at the computer nerds and idiocy of projections: back up your criticism. Do better.
And I’m not talking about doing better for a single team’s prediction. I’m talking about doing better for a larger set of teams. PECOTA uses the same process for the Royals as it does the Red Sox.
I am therefore challenging all comers to submit their predictions for the win/loss record of each team in the American League. You think projection systems are worthless? Beat them.
You may submit your predictions one of three ways:
One - Comment in this post on RoyalsReview.com
Two - Comment on Facebook in response to this article
Three - Email firstname.lastname@example.org
I will compile your predictions. At the end of the season, we’ll review who was closer to the actual results: you, or the computer.
Just to be clear, I don’t have any skin in this game. Pecota isn’t my grandfather’s name, I don’t have stock in ESPN or Baseball Prospectus, and I don’t really care about who wins or not.
Mostly, I’m just interested in if the loudest, brashest voices in the anti-stat camp have a point, or if they’re just bullies who think that they are better and smarter fans than anybody else. Either could be true—as could both!—but we’ll never know who wins until we give it the good ol’ college try.
Happy prognosticating, peeps. May the best predictor win.