The Yankees unclutch hitting

Series Preview: Seattle Mariners
Scouting The Trade Market: Edwin Jackson

It seems odd to read complaints about an offense that had scored the second most runs per game in baseball. Yet as the deadline approaches and the Yankees make plans to shore up their roster, there have been an increasing number of calls for the Yankees to add a bat. Some of that might be the temporary frustration of an A-Rod-less lineup, but it does extend a bit beyond that. Something feels off about this offense. They put plenty of men on base, for sure — their .342 team OBP ranks second in the AL and is 20 points better than league average. But it seems as though they hit a degree worse once those men are on base.

We can look to fairly standard stats to debunk that idea. With men in scoring position the Yankees are hitting .261, which ranks fifth in the AL. Even better, they’re slugging .439 in those situations (.178 ISO). The added power helps ensure that runs score. Even when you add in situations when they have a man on first, they’re right around the same area. That is, while they might not be a super team with runners in scoring position, they are in no way deficient. When you put that many men on base in the first place, you’re going to score plenty of runs by being just above average with men on base.

Still, there’s some context lacking from the runners in scoring position stats. For instance, on Friday night the Yankees hit extremely well with runners in scoring position. But there came a point where all those extra runs were superfluous. That is, they were putting up numbers with men on base, but not in what would be considered clutch situations. The leverage in those situations was low. Thankfully, we have some numbers we can examine to help us mete out the higher pressure situations.

At FanGraphs there are leader boards and splits for almost everything. One thing I started checking this weekend, when researching an article on the Reds for ESPN, is how teams are performing in high-leverage situations. As it turns out, the Yankees are pretty poor in this regard. They rank 10th in the AL in both batting average (.244) and wOBA (.301) in high pressure situations. The Rays and Red Sox, by comparison, are their superiors, with .338 and .335 wOBAs in those same situations. In fact, the AL East takes the top three spots; the Blue Jays rank third with a .330 wOBA when it counts the most.

We can take this notion of high-leverage performance a step further, still, when we include WPA-based stats. There are two in particular I’ve grown to enjoy on a team level: WPA/LI and Clutch. WPA/LI basically strips leverage out of a team’s performance and gives you a context-neutral rating of a team’s offensive output — that is, how they hit, period, regardless of a situation’s leverage. The Yankees here rank third with a 3.54 WPA/LI, though they trail the Red Sox and Rangers by a considerable amount. But they’re about an equal degree ahead of the No. 4 team, the Tigers, while nine of the 14 AL teams have negative scores.

(On an individual level, Curtis Granderson leads the way, while Jorge Posada has been the worst on the team.)

The Clutch stat is similar in nature, but instead of comparing a team’s production to its context-neutral state, it compares overall performance to performance in high level situations. Here’s where the Yankees are at their worst. Their -3.65 Clutch score is dead last in the AL. This is in part because they hit so well overall; the Red Sox and Rangers are the second and third worst teams in terms of clutch. But when combined with their poor overall performance in clutch situations, it helps illustrate why there is a perception that the Yankees have underperformed this year. They have scored plenty of runs, but they’ve failed when a hit would have made a huge difference.

(Russell Martin actually has the best Clutch score on the team, followed by, gulp, Francisco Cervelli. Mark Teixeira is the trailer at -1.41, and A-Rod is pretty bad at -1.19. Brett Gardner, Derek Jeter, and Robinson Cano round out the bottom five.)

In a way, a similar issue afflicted the Yankees last year. In 2010 they had a -2.50 Clutch score, fourth worst in the AL, while their WPA/LI was second best. It’s a bit different this time, though, as Alex Rodriguez led the way in Clutch last year. That is to say that Clutch score is not skill-based. Players will often regress to the -1 to 1 range if given enough time — even Captain Clutch himself has a 0.78 career Clutch score. Obviously high leverage situations have different effects on different players, but they normally don’t perform out of line with normal expectations. The problem is that the season has particular start and end dates, giving only a finite time for adequate regression.

There is a chance still for the Yankees offense to perform like the juggernaut it is on paper. Their numbers in high leverage situations stink now, but it’s not as though the team collectively wilts under pressure. In some cases, such as Robinson Cano and Mark Teixeira, it might be a matter of approach. With others it could be as simple as dumb luck. The perception that they need a bat is understandable, and at DH they certainly could use some help. But what they need more than anything is for their current lineup to hit in high-leverage situations as they’re capable of normally. If that aspect of the game falls into place, the Yankees will have little to worry about in the final two months.

email
Series Preview: Seattle Mariners
Scouting The Trade Market: Edwin Jackson
  • first time lawng time

    DH spot has been pretty weak. I love Jorge, but I’d like to see Montero in there.

    • Sayid J.

      The more Montero hits, the more he forces the issue. Some of the writers and others have been critical of the “his OPS is too low” commentary, but it’s not just about OPS alone. However, his OPS is indicative of the fact that he is not playing up to his potential. The last few days it seems he has began to hit the ball with more authority and extra base hits. Everyday Montero has a good day at the plate, he is one step closer to getting the call. I would imagine that if he continues hitting well and survives the trade deadline, we will see Montero soon enough.

    • David, Jr.

      It is a no brainer. Watch for it after the deadline if he doesn’t go, which I hope he doesn’t unless it is for a major starter in a headline making blockbuster.

    • Brooklyn Ed

      8/1. Montero will be here, if he’s NOT traded then.

      • Cuso

        I’d bet your kids’ eyesight that we won’t see Montero until September

  • Rich in NJ

    That’s why I can’t wait for the trading deadline to pass. The best move is likely no move.

  • JobaWockeeZ

    I wonder how the 2009 team fared in this regard. And in before someone says we should have signed Matsui and Damon.

    • Rich in NJ

      Although I get your point, they made it a priority to sign Damon, offering more over two years ($14m) than he ending up making in terms of base salaries ($13.25m), so I’m not sure that fans can legitimately assign fault to the Yankees.

      • JobaWockeeZ

        I have no problem with Damon being signed it’s all those jokes making fun of the people who really wanted them back.

  • infernoscurse

    eduardo nuñez is clutch

    • Monteroisdinero

      On second base, 1 out, Jeets up. “Better steal 3rd and be aggressive when Jeets hits his grounder after I get there”

      /clutch

  • ItsATarp

    We hit too many line drives at fielders instead of away from them. This team is not like 2009 where they could score and hit on will…We’re not even going to match the win total of 2009 and are going to be swept that the Mariners who blah blah blah blah

    • David, Jr.

      I will take line drives. They aren’t going to go at fielders forever.

  • Bronx Byte

    Most moves can be done within the organization.

    • http://www.riveraveblues.com Mike Axisa

      Yep. Maybe one of these days they’ll actually get around to making them.

      • Crime Dog

        What are you talking about? They did just bring Steve “The Franchise” Garrison up.

  • mike

    this is a case where a new statistical analysis proves a point obvious to the fans – the Yanks do not seem to take advantage of scoring opportunities, and their line-up (either due to prolonged slumps or poor approach) is inconsistant and has too many dead spots.

    while i believe most players will play to their historical norms, i think the Yanks need to make a move to get a bat – this complaint has been common in the past (Tino, Brosious, overall lineup with no spark)and it seems that every year the Yanks realize/recognize it and make a move ( Fielder, Mike Stanley, Justice, Hill, Canseco, Nady, Berkman etc) to pump up the offense when required.

    I can’t expect this year to be any different

  • toad

    I think this complaint is more a result of frustration with specific games and situations than of careful analysis. The Yankees score 39.1% of their base runners; the Red Sox 38.9%, the Rays 35.3%. Nor is it fair to use the Friday night game to claim that they waste runs. As I’ve commented in other threads, the statistics suggest that the Yankees are very consistent in their run production per game – one of the most consistent teams in MLB and much more consistent than the Red Sox.

    I appreciate that you are using more complex measures, but remember that things like WPA and LI – derived from WPA – have a lot of error in them. They sound good, but the complexity and necessary underlying asumptions make them sloppy.

    • JobaWockeeZ

      It’s funny that you choose WPA/LI to criticize when it’s helping your overall point.

      • toad

        I guess so, but just because the number that comes up supports my argument doesn’t mean I have to think it’s a great stat.

        I do wonder about the value of another stat that tells us how a team hits regardless of context. Unless WPA/LI is largely uncorrelated with OBP, OPS, etc. it doesn’t add much, does it?

  • In the Shadow of Scott Brosius

    A fine article, missing only a convenient definition of what the statisticians count as “clutch:” RISP w. 2 outs? RISP w. 2 outs when team is tied or trailing by <2 runs, and tied or winning by 1 or less? ABS from the 7th inning on in those situations? All extra inning ABS regardless of RISP? As fans, w.o. advance stats, we all think we know what counts as clutch hitting, and I'm glad these stats make clear what I've often felt about Cervelli's contributions to the team. For instance, in Sat.'s loss to Oakland, Girardi pinch-hit Posada for Laird in a clutch situation, and I half-thought to myself "I'd so much rather see Cervelli hit here; he's more likely to get it done."

    • http://www.riveraveblues.com Joe Pawlikowski

      It’s not that he’s more likely; it’s that he HAS produced in those spots. Again, there is no predictive value in clutch stats, at least that anyone has yet found.

      I like the Clutch stat, because it doesn’t relate to any specific situation. It’s any at-bat where the Leverage Index (LI) is above 2.00. That could be in the fifth, down two with the bases loaded and two outs. It could be in the ninth with a man on, down by one. So it doesn’t take one category wholesale. It’s all about specific instances where the pressure is highest.

      • StanfordBen

        Glad to hear you say that, Joe. It wasn’t obvious from your post whether you thought any of this stuff was meaningful or predictive (which it isn’t really), rather than just summarizing what’s actually happened (which it does).

      • toad

        If there is no predictive value, why calculate them?

        A stat either measures something or it doesn’t. If it has no predictive value then all it is is a random draw from a distribution.

        • http://www.riveraveblues.com Joe Pawlikowski

          If we eliminated stats that had no predictive value we’d have very few stats indeed.

          Statistics are a record of what happened on the field of play. This is a record of how the Yankees have, to this point, performed in high leverage situations. As I say in the last two paragraphs, this doesn’t mean that it will continue, and that the biggest addition the Yankees can make is translating their overall offensive success to these situations.

          Again, this whole post is about the perception that the offense is in some way lacking.

          • toad

            If we eliminated stats that had no predictive value we’d have very few stats indeed.

            I don’t think so. I do understand that you recognize that past values of so-called clutch stats tell us nothing about future clutch performance. But are most stats really not predictive?

            I thought that one of the major contributions of sabermetrics was persuading people that performance predicts performance. Not so?

            Commenters here seem to believe that. Everyone cites all sorts of stats when discussing trades, whether to bring up a minor leaguer, whether to change the lineup, etc. If stats by and large are not predictive, then why are they relevant to these discussions? If things like FIP and WAR and WHATNOT don’t tell us anything about likely future performance, why bring them into the discussion?

            • http://www.riveraveblues.com Joe Pawlikowski

              A lot of these stats try to isolate a skill, which can be more predictive than something like, say, batting average, which fluctuates greatly from year to year. That’s FIP, essentially. It tries to isolate certain skills a pitcher possesses.

              Many of the stats are aimed to accurately assess created value. wOBA is an example of that. It basically improves on BA/OBP/SLG by assigning a real-world value to each event and tallying it for players. There’s not a ton of predictive value in that, but it’s great for assessing how a player fared. And yes, with a large enough sample you can get a good grasp of what to expect from a player in the future.

              So yes, some of these things can tell us something about future performances. But baseball is a game with so many moving parts that we have a hard time predicting anything. We can project — that is, set up a certain set of expectations. But predicting? I think John Sterling has the definitive line on that one.

              • toad

                Sure.

                FIP tries to isolate certain skills. But if 2010 FIP isn’t generally predictive of 2011 FIP then it fails at that task. Of course, for any individual there are any number of reasons it might not work out – injuries, etc. But on the whole, if this year’s FIP isn’t pretty well-correlated with next year’s (across pitchers) then it does not isolate anything.

                I’m not claiming it needs to be perfect or anywhere close, just that there needs to be a reasonable correlation. The whole point is to estimate what he’s likely to do next year. Otherwise, you might as well look at how a guy hit against pitchers who were born in California or something in deciding whether to trade for him.

  • BklynJT

    They should come up with a metric to classify unclutch pitching!!!

    What stats would we look at?

    -strand rate
    -2-out hits
    -???

    It can help bridge the gap between FIP and ERA. If you give up a bajillion hits that arent HRs (even if they are off the top of the wall), strike out much more than you walk, your FIP would go down, even if you gave up 5+ runs in the process… but maybe your unclutch pitching stat would be bad.

    In another thought… is there another “fielder independent pitching” stat that takes into account batted ball data? I mean really, a ball hit where no fielder can reasonably be expected to get to (not just HRs) should count against the pitcher.

  • StanfordBen

    I’m surprised to read this on RAB, because most sabr people agree that clutchness either doesn’t exist or the effect is extremely minimal.

    Also, this post looks at a bunch of different stats, some of which make the Yanks look unclutch, some of which don’t (hitting with RISP). If you look at enough different stats, you are bound to find a couple that support some hypothesis due to randomness. If the clutch-related stats all supported the same hypothesis, I guess you might be able to make an argument, though you’d still have the overwhelming evidence of the history of baseball showing that clutchness doesn’t exist.

    • http://www.riveraveblues.com Joe Pawlikowski

      This is where the argument gets misunderstood. No one’s arguing that there is no such thing as clutch. The argument is that it is extremely difficult to pick out clutch as a skill. Players’ performances in clutch situations vary so wildly, because they come in small and irregular samples, that it’s tough to separate the signal from the noise.

      To your second paragraph, I think you misunderstood the point. I used the RISP stats as a setup to show why they’re not telling the whole story. You can raise your average with RISP when you’re up big and you’re down big, so it’s not a great measure of clutch performance. Numbers in high-leverage situations are far more telling, and they don’t look good for the Yankees this year. So it’s not all stats being in agreement. Rather, it’s that certain stats are more telling than others.

      • StanfordBen

        First of all, I sort of take back what I said (or what I thought) because of your clarification above that clutch stats have no predictive value. That was the main issue I had with your post.

        Second, there are plenty of people who argue that there is no such thing as clutch. For example, the seminal study by David Grabiner (http://remarque.org/~grabiner/fullclutch.txt):
        “The basic conclusion didn’t change; I still don’t have evidence that clutch hitters exist, and if they do, they cannot be very important…
        The correlation between past and current clutch performance is .01, with a standard deviation of .07. In other words, there isn’t a significant ability in clutch hitting.”
        This is averaged over a very large set of players, so it’s not an issue of small sample size (even though the individual sample sizes are small). At the very least, the consensus is that if clutch hitting exists, its effect is minimal compared to what most people think. So the argument isn’t just that it’s hard to estimate and pick out because of small sample size and other factors, it’s that clutchness is either barely an effect or not an effect at all.

        Third, as for the RISP stats, I understand that you were saying that RISP doesn’t tell the whole story. And I also agree that the other stats are “more telling” in the sense that if I were to define clutchness I’d probably go with those stats rather than overall RISP. I was just saying that if you look at enough clutch-related stats, you’re bound to find a couple that will support your conclusion, so people shouldn’t read too much into the fact that the stats say that the Yanks are unclutch.

        The bottom line is that if we agree that these clutch stats have no predictive value, then the take-away from your post (which I think is a good take-away) is that you can respond to Yankees observers as follows:
        – If they say, “Here we go again, the Yanks suck with RISP!” you can respond, “Actually, they’re hitting just fine with RISP, it’s just that you remember particularly bad nights that are simply due to random fluctuations.”
        – If they say, “The Yanks are unclutch in high leverage situations” you can respond, “It’s true that they’ve been unclutch in those situations so far this season, but they’re just as likely to be clutch or unclutch going forward as if they’d been outrageously clutch so far this season, so don’t worry about it.”

        • http://www.riveraveblues.com Joe Pawlikowski

          Agree with everything here. The second point is saying the same thing a different way. Grabiner is saying the same thing: we cannot define clutch as a repeatable skill. As I said in the post, even in my favorite measure of a player’s clutch performance, Derek Jeter is right within the normal range.

  • David K.

    I agree with the gist of this article, that is, that the Yankees hit poorly in “clutch” situations where the game is on the line and they need a hit or two. This has been going on all season, especially for Teixeira. I would not expect him to be able to turn it around. As a left hand hitter, he is simply a dead pull all or nothing hitter. We see Tampa Bay and other teams now employ the shift against him, and it is killing him. There is a reason he hits worse against Tampa and Boston than against anyone else. Unlike most good major league hitters, Teixeira only has one type of swing from the left side and he cannot make the adjustment.