April 12, 2016

Pebble Hunting

How You Tried To Beat PECOTA

by Sam Miller

Everything is gone. Just gone. There were no survivors. Here. There. Nobody knew who started it.

It was the machines.

A new order of intelligence. Decided our fate in a microsecond. Extermination.

But there was one man who taught us how to fight. To storm the wire of the camps. To smash those metal monkeyfighters into junk. He turned it around. He brought us back from the brink.

His name is [Here]. [Yourname Here].

***

A couple weeks ago I challenged you to Beat PECOTA. Many hundreds of you did, logging around 10,000 total picks to prove that, presented with a pool of hundreds of algorithmically derived projections, you, the human, the bottle of flesh and spit and sadness, could identify the weakest ones. As much as I look forward to seeing whether my Beat PECOTA team can beat your Beat PECOTA team, I’m much more eager to see whether we, The People, can beat It. Whether we can prove we have a Monkey-Fighting place in this Monday-to-Friday world.

To that end, let us review the tactics we chose.

The Most Popular “Over” Pick
There are exactly two ways to answer this one. The player who received the most total “over” picks was Wade Davis (3.57 projected DRA), chosen by 77 Beat PECOTA contestants*. That’s 11 more times than any other player was chosen, and 32 more than the third most popular “over” pick, Manny Machado. This is totally unsurprising. I love PECOTA, more than you do probably, but even I consider the Wade Davis projection to be its Nikki & Paulo episode. Not that I can blame it. This is Wade Davis:

• He’s a 31-year-old pitcher with a 3.11 ERA over the past three years, in a pretty good (275) block of innings.
• He’s also a 31-year-old pitcher with a 0.97 ERA over the past two years, in a relatively slim (139) block of innings.

Which, to PECOTA, must be like saying he's simultaneously a hammer and a nail.

You and I get to believe whatever we want to believe, but PECOTA’s got to follow a set of rules that will apply to all baseball players, and there’s not really a set of rules that applies to all baseball players and whatever Wade Davis appears to be. He’s a historical outlier.

And yet: Maybe we’ll still (almost) all be wrong! Greg Holland was nearly as good as Davis before 2015, and would have been worse than PECOTA’s Wade Davis projection in 2015. And there are, amazingly, four people who bet the “under”—that is to say, “worse” on Davis this year.

Those four are the reason there is another way to answer this question: 66 people chose the Over on Bryce Harper (.312 projected TAv; last year, he was at .386), and not a single educated picker chose the under. (Harper, like Davis, was one of my handful of picks.) Harper’s coming off one of the greatest seasons of our lifetimes, the best offensive season since Bonds. One year before that, he had a worse OPS+ than Matt Joyce and Justin Ruggiano. One year before that, he had a worse OPS+ than Dioner Navarro, Michael Cuddyer and Mike Carp. And one year before that—three years ago—he was worse than Will Middlebrooks, A.J. Pierzynski and retirement-tour Travis Hafner. I’m right there with you folks—corner turnt—but if Bryce Harper had been a second-round pick, we’re probably not all so cocky here.

Other popular Overs:

• Manny Machado – .277 TAv, .292 last year (45 overs, 1 under)
• Mike Moustakas – .254 TAv, .291 last year (43 overs, 0 under)
• Sonny Gray – 4.02 DRA, 2.76 last year (36 overs, 0 under)
• Addison Russell – .253 TAv, .252 last year (34 overs, 1 under)
• Zack Greinke – 3.87 DRA, 2.17 last year (31 overs, 5 under)

The Most Popular “Under” Pick
Pablo Sandoval: .272 True Average. Last year he was at .229, and the years before at .283 and .275.

I don’t know how many of our contestants thought about this, but the minimum number of plate appearances required for a pick to “count” poses a little bit of a loophole. If a player projects to be so bad that if he gets a little worse than his projection he won’t clear the minimum, then it’s almost free money to take the over. If he does hit the over, you win. If he hits his projection, it’s a push. If he hits the under, then he won’t reach the minimum number of at-bats, and that’s a push, too. We aimed to avoid this loophole by setting the minimum projected plate appearances (251) high enough to only allow betting on established players, and by setting the minimum actual plate appearances required to count low enough that even a backup catcher would clear it with ease. But about two-thirds of our picks in this game were Overs, which means we are either collectively bullish on baseball players as a species, or that we think PECOTA is too pessimistic, or that we are all gaming this little loophole, intentionally or unintentionally.

I think we’re all too bullish. That fits my view of humans/baseball fans, generally.

Anyway, Pablo Sandoval will almost certainly reach the minimum (which, you’ve made it this far without my saying, is 81 plate appearances). If he’s awful, truly awful, he’ll still probably play enough for you to cash in on your under bet.

But, fat as he is, I’m a little surprise he’s our consensus under: 39 pickers took the under on him, and only one took the over. If you never saw his gut, you’d consider him a classic BABIP-bounceback case—his was way below his career average, and explains the overwhelming bulk of his offensive rot last year. But you saw his gut. PECOTA didn’t see his gut. Many a baseball mind has chosen to go with his gut over the numbers, and in this case we’ve all chosen to go with Pablo Sandoval’s gut over the numbers. Gutsy.

The most popular unanimous Under is Steven Souza, with 33 pickers. Souza was league-average last year, .259 TAv. PECOTA projected .282, which was substantially more conservative than a year earlier (.299) and still high enough that Mike Rizzo, Joe Ross and Trea Turner are sharing a laugh about it. Souza is hitting .381/.409/.857 through the first six games. I actually hope you all are wrong. The world is better when Triple-A sluggers get a chance, and Souza is the comp some GM is going to remember eight years from now when deciding whether to call up some 26-year-old outfielder who has been old for his levels.

Other popular Unders:

• Mike Napoli – .284 TAv, .261 last year (38 unders, 2 overs)
• Matt Kemp – .289 TAv, .273 last year (33 unders, 1 over)
• Jon Lester – 3.70 DRA, .389 last year (32 unders, 3 overs)
• Troy Tulowitzki – .291 TAv, .274 last year (27 unders, 5 overs)
• Kenta Maeda – 3.79 DRA, N/A last year (14 unders, 1 over)

We aren’t complicated people. We tended to choose players because we thought PECOTA was regressing them (in either direction) too much toward their career averages, while we put more faith in last year as a single, conclusive data point. And we tended to impose a steeper aging curve (for both young and old players) than PECOTA did. These two things are pretty much the reason PECOTA exists—to do the regression-to-career-averages math better than we can, and to employ a more educated aging curve than we can. Really, we’re playing right into its hands. But we’ll see, won’t we? We’ll see.

Miscellany
The most evenly split pick is Yoenis Cespedes, who had eight overs and eight unders. The players we were most divided on, of course, would simply not get picked—if we didn’t have a strong opinion one way or the other, we’d just pass, so maybe the most evenly split player was one of those who nobody picked. Seven chose the over on Adam Wainwright and six chose the under. Kevin Gausman, Kevin Kiermaier, Travis d’Arnaud, Clay Buchholz, Yu Darvish—all seem like players who could be very controversial in a game like this, and yet the picks split more or less right down the middle. J.A. Happ: Five overs, six unders. There is nothing less predictable than J.A. Happ, and so maybe we were all risk-averse enough to just defer.

Finally, 46 entrants chose exactly one player. These people were, obviously, extremely confident in their one pick. These are their 46 picks:

• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, over
• Jose Abreu, under
• Jose Abreu, under
• Matt Kemp, under
• Jorge Soler, over
• Nick Castellanos, under
• Jhonny Peralta, under
• Justin Smoak, over
• Juan Nicasio, over
• Nick Ahmed, over
• A.J. Pollock, over
• Jonathan Schoop, over
• Nolan Arenado, over
• Adrian Beltre, over
• Wade Davis, over
• Wade Davis, over
• Wade Davis, over
• Paul Goldschmidt, over
• Paul Goldschmidt, over
• Evan Longoria, over
• Fernando Abad, over
• Mookie Betts, over
• Brandon Belt, over
• Matt Adams, over
• Matt Adams, over
• Matt Adams, over
• Matt Adams, over
• Jason Kipnis, over
• Erick Aybar, over
• Josh Donaldson, under
• Matt Harvey, over
• Matt Andriese, under
• Yasmani Grandal, over

What do these tell us? That listing players alphabetically has consequences.

*All ballots that were submitted without access to any actual PECOTA projections were removed from this analysis, though they'll still be on the leaderboards. The leaderboards, by the way, are coming.

Sam Miller is an author of Baseball Prospectus. Follow @SamMillerBB
Click here to see Sam's other articles. You can contact Sam by clicking here