Josh Reddick and the A’s hit well in high-leverage situations, but they still underperformed. (via NickB149)
If you want to look at these what ifs quantitatively, there’s perhaps no better way than adjusted standings—not what actually happened and not just what could have happened, but what should have happened (mathematically speaking, at least). And this year, quantifying those what ifs brings some curious results. The 2015 season had the most underperforming team in the decade-plus we have data for, along with the biggest cluster of underperforming teams in that period—making it, in a sense, the unluckiest year on record.
Most readers are probably familiar with Pythagorean adjusted standings, which use a team’s run differential to estimate wins and losses. Baseball Prospectus’ second and third-order winning percentages take this concept a few steps further—looking not at how many runs a team actually scored and allowed but at how many runs it should have, stripping the context away from its production at the plate. In the case of third-order wins, this is further adjusted for a team’s strength of schedule. A more technically advanced model of the same concepts is found in BaseRuns, hosted at FanGraphs.
Regardless of which method you want to use, 2015 gave us the biggest team underperformance in the years with data available—though the models disagree about which team that was. By third-order wins, it was the Indians — they finished more than 12 wins below where they were expected. By BaseRuns, it was the A’s, with a similarly affected record.
And those weren’t the only teams that underperformed significantly. Add in the Astros, and you’re looking at the three unluckiest teams since BP began tracking third-order wins in 2003. Add in the Reds, and you have four of the seven unluckiest teams ever measured by BaseRuns, with the data going back to 2002. Granted, 12 to 13 years isn’t a huge period to draw from, but in that time, there’s never been anything close to a cluster of underperformance this dramatic.
But that doesn’t mean this underperformance isn’t worth a closer look. Underperforming is often treated as the simple equivalent of bad luck, a weird quirk of chance that’s bound to plague a few teams each season. And to some extent, that’s true (again, the nature of random variation in sequencing). But, as Jeff Sullivan noted at FanGraphs last season, bad luck and unsustainably bad outcomes are not necessarily interchangeable ideas. A team that hits unexpectedly badly in high-leverage situations, for instance, shouldn’t be expected to continue hitting that badly. But that doesn’t mean that the best label for the team’s past performance in those situations is a one-size-fits-all “bad luck.”
To that end, it’s worth noting that there are different ways to underperform, and some can be explained a little bit beyond the catch-all of chance. For example, a bad bullpen can’t come close to explaining the whole story, but it can certainly tell a piece of it. And while poor clutch hitting performance is often a culprit, it’s not a requirement. This cluster of 2015 teams underperformed not just significantly, but distinctly.
Let’s take a look: There’s nothing to suggest that this one-year cluster is due to anything other than random variation. Underperforming to this degree indicates outcomes that are supposed to be unsustainable and difficult to explain at the team level, let alone the league level—there’s no reason to think that there was something about 2015 that made it extra unlucky (though there were more Friday the 13ths than in a typical calendar year, so maybe there’s that to consider). This sort of underperformance can largely be attributed to run sequencing, and just as there isn’t evidence to suggest that good sequencing is a repeatable skill, there isn’t anything to suggest that bad sequencing is necessarily a systematic flaw.
Oakland Athletics
-12.0 BaseRuns differential, -11.9 third-order wins differential
The A’s received the most attention for their underperformance this season, largely because of their spectacularly bad results in one-run games. Early in the season, they were on a near-record pace for one-run infamy—they didn’t quite keep it up in the second half, but their 35 one-run losses were far and away the most in the majors. From there, a natural assumption would be that the A’s were bad clutch performers. But at the plate, that wasn’t true. They actually hit slightly better under high-leverage circumstances than low-leverage ones, and they were one of the season’s better hitting teams under pressure—their .327 wOBA in high-leverage situations was good for fourth in baseball. Their team Clutch score was only slightly negative and put them solidly in the majors’ top half.
The main culprit, then, was the bullpen. And it was bad, by almost every measure. A particularly telling one is reliever WPA, where the A’s were dead last, with their pen costing them nearly eight runs above average. A bad bullpen does not an underperforming team make, but it can certainly play a factor, and it was a big one here.
Houston Astros
-10.9 BaseRuns differential, -11.8 third-order wins differential
Unlike Oakland, Houston had a fairly decent bullpen. But also unlike Oakland, the Astros were pretty terrible at the plate when the pressure was on. While they were one of the best-hitting teams in baseball in low-leverage situations (a wOBA of .341, second in the majors), they were one of the worst in high-leverage ones (.287 wOBA, ranked 26th). This contributed to a team Clutch hitting score that was second-worst in baseball, which in turn contributed to their significant underperformance.
Cincinnati Reds
-9.6 BaseRuns differential, -8.3 third-order wins differential
The Reds were pretty bad at most things in 2015, and performing under pressure was one of them. In a sense, their underperformance can be seen as a combination of Oakland’s flaws and Houston’s—a weak bullpen coupled with poor high-leverage hitting doesn’t give you a pretty result. But where the Astros were bad at the plate under pressure, the Reds were downright atrocious. Their team Clutch score of -8.28 wasn’t just the worst in baseball, it was the worst in the stat’s entire history, which dates back to 1974.
The natural question, then, is how a team that was this historically terrible in the clutch could manage not to top these standings. The answer is that the Reds didn’t give themselves too many opportunities to perform in the clutch at all. In other words, they were ridiculously and spectacularly bad in clutch situations, but they didn’t get into too many of those (shown by the fact that their team WPA, like their clutch score, was awful, but their WPA/LI wasn’t too far below average). This eased the effect of their poor performance under pressure a bit, but it certainly didn’t negate it, and they were left with a team that underperformed by a large margin.
Cleveland Indians
-8.6 BaseRuns differential, -12.3 third-order wins differential
The Indians are perhaps the most mystifying of 2015’s underperformers. The bullpen wasn’t the culprit—in fact, it was fairly solid. It wasn’t clutch performance, either—the team was by no means spectacular in high-leverage situations, but not too bad. Their team Clutch score was only slightly below average (18th in the majors), and they had just a small gap between their high-leverage and low-leverage wOBA (.015 worse when the pressure was on). A cursory look at some factors that indicate different forms of luck also comes up empty—their BABIP was solidly average, their record in one-run games wasn’t far from .500. At a glance, there’s nothing to suggest why they underperformed so severely.
The Indians, more so than any of these teams, simply appear to be a victim of a cruel variation in sequencing. But man, what a victim they were. One way to look at this is cluster luck, measured at The Power Rank—which shows how far off a team’s sequencing is from the expected—and as you might guess, Cleveland’s cluster luck was bad, the second-worst in baseball.
But what’s interesting here is not that it was bad, but how it was bad. Most underperforming teams draw their poor luck from either offense or defense, but not both. This makes sense, as there’s no reason for random sequencing variation on one side to be connected to the other. But the Indians were an exception here, with sequencing that was remarkably poor on both sides, and some pretty awful luck as a result.
It’s worth reiterating that most of this—apart from bullpen strength—isn’t necessarily indicative of a team’s true talent or what to look for in the coming season. And while the differences underneath these teams’ underperformance are distinct, they all add up to various shades of the same unexpected unluckiness, even if we can try to explain it in different ways. Basically, looking at adjusted standings in midseason can be a logical way to predict what might be next for a team; looking at adjusted standings in midwinter can be a sort of weird masochistic exercise down a rabbit hole of would have, could have, should have. But it’s January—what else are we going to do?
References & Resources
- Jeff Sullivan, FanGraphs, “The Unluckiest* Year of the Millennium”
- Dave Cameron, FanGraphs, “The Orioles and Accepting Random Variation”
- Matt Trueblood, JABO, “The Unluckiest Team in Baseball”
- Dayn Perry, CBS Sports’ Eye on Baseball, “The A’s have been amazingly bad in one-run games”
- Ed Feng, The Power Rank, “Cluster Luck”