I’ve been thinking about clutch lately. A couple weeks ago, friend of Purple Row Drew Creasman wrote about Nolan Arenado as a clutch player. Using that as a springboard, I thought it would be fun to go back a decade in Rockies history to look at the most clutch seasons the Rockies have had.
It’s first necessary to establish the parameters of terms. When it comes to clutch, there are both definitive and uncertain answers to clear questions. For instance, the answer to the question, “does clutch exist in baseball?” is a categorical “yes.” Clutch is both observable and quantifiable. Statistics such as Win Probability Added (WPA) and the aptly named Clutch, both found at FanGraphs, confirm this.
The question with less certainty is, “are some players clutch and others not?” If the first question gives us an answer about clutch in the past tense, this question attempts to determine whether or not “clutch” is part of a baseball player’s DNA, thus informing his future tense. Part of the uncertainty surrounding this question is the baseline used. Is a good hitter who hits well in high leverage situations clutch, or does he have to overperform his normal level of performance to be considered clutch? I lean to the latter, but let’s start with the former.
It’s obvious that when a batter is up at the plate in the bottom of the ninth with his team down by one run, a runner on second, and two outs, he’s in a clutch moment. This situation can also be measured. Based on run expectancy given the base/out state and a historical view of these situations, the visiting team has a 72.6 percent chance of winning. If the batter gets a hit and scores the runner to tie the game, the home team then has a 57.4 percent chance of winning. In other words, that hit increased the home team’s chance of winning by a full 30 percent. That’s coming through in the clutch.
Given the same situation in the first inning, however, the visiting team has a win expectancy of just 54.9 percent, and a game-tying base hit results in a 11.3 percent reversal and gives the home team a 56.4 percent win probability. It’s still a meaningful situation, but the stakes aren’t as high because of context. (You can play with the win probability situations here.)
Throughout the course of the season, players accrue their WPA based on how much, in aggregate, they move the needle in both directions. For instance, if the same batter was up at the plate in both of the situations above in the same game, those two plays would give him 0.41 WPA (convert the 30 and 11 percent increases in win expectancy to decimals of .30 and .11 to get 0.41 WPA). Poor performance in these situations cause batters to lose WPA (just like the pitchers facing the batter in these scenarios would receive -0.41 WPA), and typically, season totals range from 6.00 on the high end to -3.00 on the low side. Doing better in high leverage situations, such as the first will give the player a higher WPA.
The table below has the 10 best WPA seasons from Rockies hitters over the past 10 seasons, 2007-2016:
WPA Leaders, 2007-2016
The first thing that stands out here is that these are all great hitters during great seasons. That makes sense. It shows that these players hit very well in high leverage situations and contributed a lot to Rockies’ wins during each respective season. The most recent season here is Nolan Arenado’s 2016. If it felt like Arenado got a lot of hits in high leverage situations, that’s because he did.
The chart only suggests that these were clutch seasons, however, if the baseline is neutral. If each player serves as his own baseline and the clutch-ness of the season is derived from how well he overperformed his typical performance, we need a different metric. That metric is Clutch.
In addition to a player’s WPA, Clutch considers how often a player found himself in high leverage situations (this isn’t the same for every player, as guys who hit at the top of the lineup tend to be in high leverage situations more often than those at the bottom), as well as the player’s context neutral contributions. The end product is a measurement of performance in high leverage situations using the player himself as the baseline. Put differently, Nolan Arenado in 2016, who hit .294, wouldn’t get Clutch credit for hitting .294 in high leverage situations.
A high WPA is a condition for a good Clutch score, but they don’t always overlap. Only two of the top 10 WPA seasons of the past 10 years are also among the top 10 Clutch seasons (Clutch scores tend to land between 2.0 and -2.0):
Clutch leaders, 2007-2016
This list is still full of great hitters during great seasons, but because of that, it’s noticeable that the list also includes players like Yorvit Torrealba and Scott Podsednik. The WPA list gave us great hitters performing well in high leverage situations; this one offers hitters outperforming themselves in those same scenarios. That’s how the Podsednik’s of the world end up here.
There are two players here worth dwelling on a little bit more. The first one is Arenado. He showed up on the WPA leaderboard for his 2016 season, but his 2016 Clutch score was a modest 0.27. He was a very good hitter in 2014 as well. In a shortened season, he hit .287/.328/.500 with 18 home runs but did better than that in high leverage situations. His 1.43 Clutch score is the sixth best since 2007. Interestingly, Arenado’s Clutch score in his breakout 2015 season was on the wrong side of the ledger: -0.92. It was a phenomenal year, but he didn’t excel in high leverage situations.
The other compelling Clutch leader is Tulowitzki. During Tulo’s time in Colorado, he had the reputation of doing poorly in high-leverage situations. The bird’s eye view of a statistic like Clutch allows us to move beyond anecdote. In his rookie season, Tulo put up one of the best Clutch season’s the Rockies have seen in the past decade.
However, for every leaderboard, there’s an un-leaderboard. Here are the 10 worst Clutch scores of the past 10 years:
Un-Clutch leaders, 2007-2016
Well then, maybe the reputation was somewhat earned. While Tulo excelled in high leverage situations during his rookie campaign, he did the opposite of that in at least three other seasons. But there’s something of a lie in these figures as well. Take Tulowitzki in 2014: The man hit .340/.432/.603. It’s pretty damn hard to outperform yourself when you’re hitting like Miguel Cabrera. Finally, it’s worth stating that this is not a list of “bad” seasons (Gerardo Parra’s 2016 very much excepted). This is one measure drawn from contextual situations—nobody would trade 2009 Tulo for 2008 Podsednik.
There are things that neither WPA nor Clutch account for. For instance, they both measure games and compile throughout the season. They don’t account for the higher leverage, for example, of a September game in a playoff race. They also don’t consider opposing pitcher or opponent. Late innings often provide higher leverage, and higher leverage also set up matchups with tough relievers. That sort of granular look is a better fit for single-game analysis.
There’s still one final question about Clutch: is it repeatable? To get a sense, let’s see what happened the year after the 10 highest and lowest Clutch seasons:
Next season Clutch
|Player||Season||Clutch||Next season Clutch|
|Player||Season||Clutch||Next season Clutch|
Because these are “top 10” looks, we shouldn’t really expect much repetition. Fowler put together two consecutive seasons of Clutch play, and Tulo had consecutive un-Clutch ones. Otherwise, however, there’s a not unexpected move toward 0. This suggests that while we can expect good hitters to continue to be good year-to-year, consistent outperformance of one’s own baseline in high leverage situations will likely fluctuate.
Baseball that has happened is and always will be more compelling than baseball that could happen. But the former informs the latter, so these aren’t two truly distinct things. Some of the fun gets sucked out of it when baseball’s past isn’t taken as repeatable prologue, but is instead used to moderate excitement.
For instance, if a player put up four consecutive Clutch seasons over 1.00, I’d probably remain skeptical that he could repeat it, and rightly so. But if he did repeat for four more seasons until he retired, I’d finally conclude that, yeah, dude maybe had clutch in his DNA. The joy of it would come after the fact, and I might feel that I lost something by focusing too much on what I think will happen.
But, then again, I might also lose something if I took those first four Clutch seasons and anointed a new deity of the high leverage situation. Perhaps instead of four more seasons of Clutch over 1.00, this player hung around 0 and even found himself in the negative for the remainder of his career. If I spent those years trying to insert a WPA-shaped peg into a Clutch-shaped hole, I might lose something there, too.
Perhaps the real lesson is to embrace the unexpected present and take in clutch situations as they arise. That’s what makes baseball so fun.