Flu-Like Symptoms: Why PECOTA Hates Your Favorite Team (2019)

Image credit: USA Today Sports

Baseball Prospectus first published PECOTA projections for players and teams in 2003. In the ensuing 16 seasons, through last year, we projected that the Pirates would win 1,202 games. The Pirates, since 2003, have, in fact, won 1,201 games. Good job, PECOTA!

That overall number, though, smooths off a lot of rough edges. In 2006, PECOTA thought the Pirates would win 79 games. They won 67. It overshot again in 2010, expecting 72 wins, getting just 57. But it missed the boat the other way, expecting just 79 wins in 2013, when the team won 94. And PECOTA saw Pittsburgh as an 80-win team in 2015, when it went on to win 98.

So take the following graph with a grain of salt. It shows the cumulative error that PECOTA has made over its 16 seasons for all 30 franchises. A positive number indicates that PECOTA was overly optimistic, a negative number means it was overly pessimistic. Keep in mind that these totals smush together positive and negative outlier years.

While Cardinals fans might feel aggrieved, it’s not that PECOTA projected fewer wins for their team than it should have in every season. Similarly, as recently as 2017, the Diamondbacks were hardly PECOTA’s darlings, with the system predicting 78 wins instead of a 93-win Wild Card season.

One big miss can throw off the totals when you’re talking about only 16 seasons. Show me the prediction system that didn’t miss badly with last year’s Nationals (PECOTA was six wins too high) or Athletics (PECOTA was 21 wins too low) and I’ll ask for the timestamp on that spreadsheet, if you don’t mind. Additionally, PECOTA has been updated, as have been the rosters and characteristics of the teams. So an evaluation of whether PECOTA is fair or not—whether it favors some teams over others—needs to look at just the recent past. I’ll do that for the rest of this article.

In the PECOTA loves/hates discussion, three teams have stood out. Year after year, PECOTA has been accused of loving the Rays. They play in a lousy, inaccessible ballpark and they spend less money on their major-league roster than just about anybody, but year after year, PECOTA seems to project that roster near the top of the AL East, despite all the big scary monsters along the Northeast Corridor (and in Canada). Between 2014 and 2017, PECOTA expected the Rays to win 45 more games than they actually did. That’s over 11 games per season.

And the Orioles and Royals fan bases seem perpetually aggrieved, with each year’s PECOTA projections providing bulletin-board, nobody-respects-us material for the team and its followers. As noted in this space last year, maybe PECOTA has a blind spot for teams like those two, that don’t get on base a lot or have impressive rotations but feature shutdown bullpens.

So, how did PECOTA’s most extreme biases play out last season?

• Tampa Bay: Projected 83 wins, actual 90 wins
• Baltimore: Projected 70 wins, actual 47 wins
• Kansas City: Projected 65 wins, actual 58 wins

That flipped the narrative, didn’t it? PECOTA missed two teams by 20 or more games in 2018: The overachieving A’s and … the underachieving Orioles. If anything, PECOTA was guilty of O’s/Royals-philia and Rays-phobia in 2018.

I’m going to concentrate on the last half-dozen years. I looked for teams for which PECOTA was wrong in one direction or the other five or more times. If we figure the odds are 50/50 that PECOTA will be high or low, the odds of PECOTA overshooting four times and undershooting twice are about 23 percent. That’s not very unusual. Going high or low five or six times, though, that’s only an 11 percent probability. Maybe there’s something going on.

I also wanted to see the magnitude of the error. If PECOTA thought a team would do worse than it actually did, but the difference was only a game or two, that’s not a big deal. To avoid getting thrown off by one big outlier, I took the median difference of the misses.

Here are the teams that PECOTA “loves” over the 2013-2018 timeframe:

Personally, I think PECOTA is more down on the White Sox than I’d be this year. But it’s certainly been more optimistic about the club in the past six springs than what they’ve put on the field. As for the Reds, maybe PECOTA has a hard time believing a pitching staff can be that bad year after year. PECOTA has overshot the Dodgers, but not by a lot, and ironically, the year in which we got the most grief for our Dodgers projection (What? 97 wins in 2017? More than the Cubs? Are you guys nuts?), PECOTA, it turns out, was seven wins too conservative.

And here are the teams that PECOTA “hates”:

As noted, the 2018 Royals and Orioles ended PECOTA’s streak. As for the Cardinals, I’ll repeat what I said last year: PECOTA can’t account for Cardinals Devil Magic. Aledmys Diaz? Paul DeJong? Miles Mikolas? It seems that every year, St. Louis gets half a dozen or so WARP from guys you have to look up in the Annual to find out who they are. PECOTA lacks a gold-from-dross function.

Besides, a quick perusal of this year’s projected standings should tell you that whatever howls we hear from St. Louis, they are going to get drowned out by the cacophony from the North Side of Chicago. The 82-80 Cubs, only a game ahead of Cincinnati and Pittsburgh? Sorry, let me restate that: THE 82-80 CUBS, ONLY A GAME AHEAD OF CINCINNATI AND PITTSBURGH?!

OK, let’s look into this exemplar of PECOTA hatred. The case for the 82-80 Cubs, only a game ahead of Cincinnati and Pittsburgh:

• What PECOTA’s projecting is a really, really tight NL Central, with eight games separating the top from the bottom. One game out of last place in the Central isn’t like being a game out of last place in, say, any of the American League divisions. Two or three wins here or there—that’s a variation of less than two percent over 162 games—can alter the order a lot.
• There are still some free agents available (I can think of a few in particular) whose addition could add three or four wins to a team’s projection. Not that the Cubs seem likely to pursue them, but still.
• Let’s look at the Cubs’ division. The Brewers signed Yasmani Grandal. The Cardinals signed Andrew Miller and traded for Paul Goldschmidt. The Reds have added Tanner Roark, Alex Wood, and Sonny Gray along with Yasiel Puig and Matt Kemp. Even the toss-around-nickels-like-they’re-manhole-covers Pirates signed Lonnie Chisenhall and Jordan Lyles. The Cubs? Well, they swapped out Tommy La Stella for Daniel Descalso and signed Brad Brach for a year.
• Meanwhile, they Cubs had the NL’s oldest pitching staff last year. The rotation looks like Jon Lester (in his age-35 season), Jose Quintana (30), Cole Hamels (35), Kyle Hendricks (29), and Yu Darvish (32). The closer will be … Brandon Morrow (34)? Pedro Strop (34)? Steve Cishek (32)? Brach (33)? PECOTA, fully cognizant of aging curves, is projecting these old hurlers surrendering 730 runs in 2019 compared to 645 in 2018, over half a run more per game. It sees the team scoring 739 runs, less than a tenth of a run fewer per game. The pessimism is based on those aging pitchers.

I get it, we’re going to hear it from Cubs fans. But not only has PECOTA proven to be pretty accurate over the years and particularly recently, it’s also, as illustrated above, got its reasons. But take heart, Cubs faithful: Even if PECOTA’s gloom turns out to be justified, you’ll be able to bask in the warm glow of knowing your team avoided a higher payroll in 2019.

For full access to PECOTA’s player and team projections, including downloadable spreadsheets and detailed player cards, become a Baseball Prospectus subscriber today.