The Cubs Vs. a Decade of Projections

Earlier, I published a post including this graph:

Of all the things that stand out, perhaps nothing stands out quite like the Cubs. Over the last 10 years, they’ve won 55 fewer games than they’ve been projected to win, and while there are some quibbles you can have with blending all the years and all the projection systems, at the end of the day, projections aren’t too dissimilar and they haven’t changed super dramatically over the decade, and no one comes particularly close to matching the Cubs’ level of disappointment. At -55, they’re separated from the next-lowest team by 19 wins.

I don’t think there’s really anything predictive, here. I’d never bet on a given team beating or undershooting its next-year projection. Certainly not based on what it just did. But -55 seems worthy of investigation. What can we find out, about why this might’ve happened? What traits or events might’ve contributed to the Cubs coming off as such a disappointment?

To begin with, an obvious step. Over the last 10 years, the Cubs have won 756 games. But it’s worth checking on their run differentials, to see if everything matches up. And as it turns out, the Cubs have undershot their own Pythagorean records by a little bit. According to the runs scored and runs allowed, then over the decade, the Cubs should’ve won 769 games, which is an increase of 13. Maybe you don’t like the use of the word “should’ve”, but you know what I mean. I don’t think you can expect a projection system to know when a team will or won’t play to its run differential. So now we’ve accounted for almost a quarter of the difference between the projected wins and the actual wins.

Yet, if you take away 13 from 55, you still have the Cubs in last, at -42 wins. So it’s clear there’s more to explain. I decided to take a look at team clutch performance, as measured by our in-house Clutch statistic. (You can find it on the leaderboards.) It’s true that Clutch isn’t a perfect stat, capturing everything. It’s true this could be partially captured already by under-performing the Pythagorean record. But I thought there could be something here, and it seems like there is.

For the 10 years, I combined both batter and pitcher Clutch score. The average, by design, is zero! One standard deviation is about 16 points. The Cubs are found at -27, ranking second-worst in all of baseball. The one team with a lower score is the Rockies, at -36, and it turns out they, also, have under-performed preseason projections over the decade. So at least by this rating, the Cubs have been the second-least-clutch team over a 10-year span, and it stands to reason that would have a significant effect on final record. It also stands to reason this can’t be predicted. I know the temptation is to think 10 years is enough to cancel out most of the noise, but Clutch seems to be random. If you know the key to harnessing it, I’m all ears, but I doubt you’re going to say anything, because if anyone actually understands how to be consistently clutch, that person works for a team. Maybe the Baltimore team.

It was pointed out in the comments that the Cubs have frequently been sellers, especially in the last few years. It seems like that could be a contributing factor — sellers lose good players halfway through the year, while team projections never count on trades taking place. Feels like a really obvious point. But here’s the interesting thing: over the decade, in the first halves, the Cubs have won 46% of their games. In the second halves, they’ve won just under 48% of their games. In the first halves alone, there’s a difference of 36 wins between projections and performance. So if trades are a factor, they seem like a really small one. The Cubs have mostly under-performed even before selling parts off.

I should note that, while the Cubs are at -55 wins overall, it of course hasn’t been even every year. They more or less matched projections in 2007 and 2014. They exceeded projections in 2008. Out of the seven seasons remaining, no disappointment has come close to the 2006 disappointment. In that year, the Cubs won 18 fewer games than they were projected to in March.

There’s a lot that goes into something like this, but I can easily identify some key factors. Derrek Lee missed most of the year with injuries, and the season before, he’d been worth seven wins. Mark Prior was limited to 44 mostly ineffective innings, because of shoulder problems. Kerry Wood didn’t even throw 20 innings, after seeming like a decent bet to bounce back. Injuries were devastating. The Cubs also under-performed their Pythagorean record by four wins, and in the summer they traded away Todd Walker, Greg Maddux, and Scott Williamson. They also dealt both Neifi Perez and Phil Nevin. Today I learned the 2006 Cubs had Phil Nevin.

So pretty much everything went wrong for that team. That’s the story of the Cubs’ biggest disappointment of the decade. There’s still a stretch from 2009 – 2013, where the Cubs undershot projections by 43 wins. I’ll point back to run differentials and clutch performance. There were also some injuries and surprising under-performances. In case anyone’s troubled by the pattern, the Cubs last season did what they were supposed to do. Which provides some hope this year’s team could be able to match its relatively lofty expectations. Again, probably the biggest point: while it’s been, overall, a rough decade for Chicago, there’s not great evidence that under-performing is predictive. Even if I haven’t completely explained the last 10 years, it doesn’t seem like it should mean anything for 2015. It just means Cubs fans would be extra appreciative of a competitive product that doesn’t fall apart.