I'm not a huge fan of making predictions about who's going to win a given series, or the whole postseason, or what have you. I very much enjoy talking about who has a better chance of winning, and for what reasons, et cetera. But I see little reason in saying anything like "I'm picking Twins in 6." I guess maybe it's fun? (Not a big fan of fun things.)So generally speaking, I don't really care what people's predictions are.But then...once in a while...19 baseball experts from the nation's leading sports network try to predict which team will win the World Series, and not one of them even picks a team that makes it to the World Series, let alone wins it.And that, I kind of feel like, is worth pointing out.I know what you're thinking: what are the chances that this could happen, assuming even that the ESPN analysts have monkey-throwing-raisins-at-a-dartboard level of guessing ability? And by "this," I mean 19 picks for World Series champs not even making the World Series.Well, the chances that any given ("random") playoff team makes the World Series is 1 out of 4. In this case, 19 people missed what would have been a 1 out of 4 chance, if they had just simply "guessed" (at random). To measure the probability of this, we have to think in these terms: 19 people in a row "hit" a 3 out of 4 chance. What are the chances of that? .75^19 = .00423, or .42%In other words, the anlaysts could have thrown all of their baseball knowledge out the third floor window of the Bristol megaplex, picked a random team to win the World Series...and there would have been less than a 1 in 200 chance that zero of their picks would get to the World Series. (Which, just to remind everyone, is exactly what happened.)But wait! It gets better.Looking further at these picks -- and I'm sure someone pointed this out before -- not one single person picked the Tigers to even get out of the ALDS. And only one person picked the Cards to beat the Padres in the NLDS.Which means (stay with me) in the combined series of: Yanks / Tigers ALDS, Padres / Cards NLDS, ALCS, NLCS, and WS, the 19 ESPN experts went a total of 1 for 95. ONE FOR NINETY-FUCKING-FIVE.If you let 10,000 teams of 19 monkeys randomly pick winners in those series, those monkey-picking-teams would average 30.9 out of 95. (.5 for 38 DS picks + .25 for 38 CS picks + .125 for 19 WS picks)Congratulations to Enrique Rojas, the only person who picked either the World Series bound Tigers or Cardinals to win even one series. He also picked El Duque as the WS MVP in a victory over the Twins.EDIT: I'd like to take a second to address two categories of e-mails I'm getting from a lot of readers.Category 1 is best summed up by e-mailer CJ:"You're giving them too much credit. Each of nineteen guys failed to connect on TWO one in four chances. If everyone picked at random, the probability that any one guy would fail to pick either wold series team is (.75 * .75) = .5625. Raise that to the 19th power and you get 0.0000178, or 1 in 55933."
On its own, this is true. And "more" impressive. As for accusations that I was wrong, however (which were often made), I stand by my original numbers. I was looking at the chances of a different phenomenon. ("And by 'this,' I mean 19 picks for World Series champs not even making the World Series.")So, dudes who wrote -- excellent point. The chances of going 0-38 in CS Champs picks are even more astronomically difficult than going 0-19 in WS Champs picks not even making the WS. But I was right also. Sweet.On to Category 2 now, as written by the beautifully named Alessio, who is probably a dude but in my imagination is a gorgeous 23-year-old woman from like Monaco who loves baseball and statistical analysis. I quote him (her? please?) at length because it's easier than writing this all out myself:I'm no statistician, but I think you made an analytical error in your post on "picks". The chances of what happened are not nearly as distant as you calculate. In fact, you're far more likely to get such results from intelligent decision makers than from random chance.
The fact that human beings are picking will tend to "bunch" the picks a lot more than random chance. For example, let's say the Yankees are better than the Tigers, and everybody recognizes that. Everybody will pick the Yankees, so the picks could rationally be 19-0 even though their actual chances of winning might be something like 55%. When the Tigers beat those odds, all of a sudden you have 19 wrong picks, although there's only one upset in the series.
Now, when you have three or four series upsets (nothing unusual there), all of a sudden you have a whole lot more than 19 wrong picks.
When you have a consensus on the various team strengths, combined with just a few upsets, you get the seemingly anomolous result of a bunch of prognosticators going 1-for-95. A random picking system would almost never be that bad; but on the other hand, it would almost always be around 50%. The humans could just as easily have been around 90% if those series had gone the other way.
Alessio. My sweet, innocent Alessio. Alesssssio, my princess of Monaco... [daydreaming now: playing with Alessio's hair; engaging in conversation about VORP vs. WARP3 over mussels and wine...now realizing instead that Alessio is almost certainly a 45-year-old dude from Canton, Ohio or something, and on top of that, feeling the obligation to publicly apoligize to girlfriend about the whole Alessio-fantasy situation]...sorry, what now?
Oh, the numbers thing.
Yeah. Well, Alessio, you fat fucking ugly monster of a man, you make what I guess is a good point. I guess my response is: yes, of course. Of course humans will, over the long haul, be better than random-team generators at predicting who wins certain games / series / whatever. I realize why, especially in this case, the experts were especially bad at picking winners. Your point is spot on: a rational human being will pick the 55%-likely-to-win team, and, likewise, so will 19 rational humans. I'm just trying to put a scenario together that sort of points out the whole ridiculousness of "predictions" in general.
Listen: It's a cheat. I cheated. And that's the kind of thing you do when you run a blog devoted to making analysts look silly.
You take advantage of a combination of hindsight, upsets, and odd numbers, and use them in a way to make people silly. And you sort of hope, I guess, that people make their own conclusions about just how much these numbers actually mean.
To me, the overall point is not that we should let monkeys throw raisins at a dartboard instead of letting experts make their predictions. But rather, isn't it kind of silly / interesting / amusing that in this particular case, a team of monkeys would have been almost a sure bet against these so-called experts?
That's all. Interpret at your own risk.
Sorry / Thanks to Alessio, whose appearance and gender remain an absolute, delicious mystery to me.
Labels: espn, predictions, world series