clock menu more-arrow no yes mobile

Filed under:

The computers saw the Cleveland Indians coming

What would WOPR think of this?

Jack Taylor/Getty Images

You may or may not recall that, before the 2016 season began, while fans and human prognosticators were all over the place in their expectations for the Indians this season, computer projections were all quite optimistic. PECOTA, run by Baseball Prospectus, and possibly the most famous of the projected standings, had the Indians winning 92 games and taking the AL Central by ten games. At FanGraphs, they had the Tribe with a more modest 87 wins, but expected that would still be enough to win the division.

The big story at the time was both sites having the Kansas City Royals with a losing record, at least nine games behind Cleveland. Each of the two previous seasons the projections had been down on Kansas City, but the Royals had gone on to win the American League pennant, and many expected the computers to be badly proven wrong again. Well, with a little more than two weeks left in the season, the Royals have a winning record, but they find themselves nine games behind the Tribe. Score one for the computers.

At FanGraphs, the projected AL division winners were the Boston Red Sox, Indians, and AHouston Astros. Two of those teams are currently in first place, and all of them are at least five games over .500. The Toronto Blue Jays, New York Yankees, Seattle Mariners, and Detroit Tigers were projected to be the top Wild Card contenders, and all four of those teams are still very much in the hunt for a postseason spot.

With 18 games to play, the Indians are sitting on a six-game lead over Detroit and are on pace to win 93 games. They're almost certain to exceed FanGraphs' projected win total, and if they win more than they lose between now and the end of the regular season, they'll best the even loftier PECOTA projection too.

Their biggest miss was the Texas Rangers, who they projected to win 79 games. Instead, Texas has the AL’s best record, but it’s worth mentioning that their success has come on the shoulders of an eye-popping 33-10 record in one-run games. if they were 23-20 in one-run games, they wouldn’t be in first place. (The Rangers have a staggering eight one-run wins against the Astros this season.) Now, they deserve credit for winning so many close games, but FanGraphs wasn’t as far off about them as it might seem if you just look at wins and losses.

If anything, FanGraphs' projections did an even better job in the National League. The Chicago Cubs were projected to be the best team (they are), and the five teams projected to make the postseason were the Cubs, Los Angeles DodgersNew York Mets, Washington Nationals, and San Francisco Giants. If the season were over right now, those are exactly the five teams that would make it.

The projections rarely do quite as well as the FanGraphs system has done this year, but they also rarely miss as badly as they did with the Royals a year ago. Projections are about probability, not exactitude. They play out the season and postseason thousands and thousands of times, and the projects amount to an averaging out of the results from those trials. And just as you sometimes roll snake eyes, sometimes a team is going to win a dozen more (or a dozen fewer) games than a computer system projects them to.

With almost 90 percent of a season's worth of new data, FanGraphs now projects the Indians to have a 96.1 percent chance of winning the Central, and a 3.1 percent chance to win one of the Wildcard spots. In all, 97.8 percent of the simulations FanGraphs runs have the Indians playing in an American League Division Series three weeks from now. They're advancing to the ALCS 46.6 percent of the time, which means they lose in the ALDS simulations slightly more often than they win. This is because their most-likely opponent in the ALDS is Boston, whom the computers view as the American League's best team. They're given a 24.1 percent chance of winning the ALCS, which means they win in the ALCS simulations more often than they lose. This is because if they reach the ALCS, they'll likely be facing someone other than Boston, and the computers view the Indians as being better than any AL team other than Boston.

The Cubs are 36.5 games ahead of the Atlanta Braves right now, with a run differential that is 379 runs better than Atlanta's, yet the last time the two teams played, the Braves won, and the two teams split their seasons series, three wins apiece. No one thinks that 3-3 record means the two teams are evenly matched, and no one who knows baseball finds that 3-3 record surprising. We know that sort of thing happens all the time in baseball. Every postseason series involves two teams who are much closer to one another in talent than this year's Cubs and Braves are, so it should never be a big surprise when any particular team wins a postseason series, or even wins three in a row to become champions. That's what people are getting at when they refer to the postseason as a crapshoot.

That said, the better a team is, the better its chances of surviving that crapshoot. The Indians are where they are despite getting next to nothing good from Michael Brantley or Yan Gomes, expected before the season to be two of their key players. They've survived shorter-term injuries to a lot of good pitchers. They've built on their lead in the second half even though All-Star Danny Salazar has been injured and/or ineffective almost the entire time. You can look at the roster and see whatever you want. The computers still see a very good team, one with a better chance of winning the pennant than almost any other team.