The Angels Are Still Clutch

The Angels are 36-32, in second place in the AL West, and two games behind first place Texas. They are in that position despite having a below average offense (-10 wRAA), a below average defense (-19.3 UZR), and a below average pitching staff (4.33 FIP). How are they winning games? The Angel way – clutchness.

If it seems like we write about this every year, well, we do. I mentioned the Angels penchant for clutch hitting two years ago, showing that they were consistently among the league leaders in our metric that shows the gap in wins added that comes from hitting well in high leverage situations. Matt Klaassen wrote about it earlier this spring.

And here we are today, talking about it again, because once again the Angels are clutching their way to victories. They have +1.34 clutch wins from their hitting and +2.48 clutch wins from their pitchers, totaling just under four wins added by coming through when it counts. Not surprisingly, they lead the league in clutch wins added.

It isn’t surprising because they do this every single year. Last year, they added seven wins to their total through clutch performance. In 2008, they had fifteen clutch wins. As noted in the first linked post above, this is basically an annual trend. The Angels are consistently among the league leaders in clutch wins, and have been for the better part of the last decade.

There are all kinds of explanations for why they are able do to this year in and year out. Generally, Mike Scioscia gets the credit for getting the most out of his team. The Angels minor league development staff is lauded as teaching their players how to play the game the right way. But the explanations are never very specific, nor are the players who put up the crazy clutch seasons consistent.

In 2008, the monster clutch Angels were Maicer Izturis (+1.46), Mark Teixeira (+1.21), and Howie Kendrick (+1.15) on offense, and Jason Bulger (+1.20), John Lackey (+1.11), and Joe Saunders (+1.03) on the mound. Last year, it was Gary Matthews Jr (+2.11, seriously) , Jeff Mathis (+1.05), and Chone Figgins (+0.95) carrying the clutch load. This year, it’s Kendry Morales (+0.91), Hideki Matsui (+0.77), and Bobby Abreu (+0.57) doing it with the bats, while Ervin Santana (+0.96), Jason Bulger (+0.55), and Fernando Rodney (+0.45) have done it on the mound.

Seriously, what do those guys have in common, besides being Angels at the time? Teixeira was a rental player who came up through the Rangers organization. It’s hard to argue that the Angels had much influence on the development of Matthews, Matsui, Abreu, or Rodney. If it’s good coaching, why did Santana have a negative clutch rating before this season, when he’s a home grown kid who has been on the roster for five years?

I’m not asking these questions to insinuate that the Angels have nothing to do with these performances – I’m genuinely curious what they could possibly be doing to extract such performances from wildly different players, but do so almost every single year? At this point, the odds of it just being luck are pretty slim, so it seems reasonable to suspect that the Angels are doing something right. But none of the theories advanced so far seem to have any kind of evidence to support them, and there does not seem to be any discernable trail we can follow that will lead us to the answer.

Until someone figures out just what the Angels are doing, all we can really do is sit and stare in amazement. Right now, there’s no explanation. The Angels are a phenomenon.




Print This Post



Dave is a co-founder of USSMariner.com and contributes to the Wall Street Journal.


58 Responses to “The Angels Are Still Clutch”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Stupid question probably, but do they play in a stadium that increases clutchness?

    Vote -1 Vote +1

    • Dave Cameron says:

      I don’t even know how that would work, but I instinctively want to reject any hypothesis that could give credit to the Rally Monkey.

      Vote -1 Vote +1

      • DonCoburleone says:

        I watch a ton of Angel games and I can’t figure it out either. My best guess is that it has something to do with the team’s emphasis on contact at the plate and aggressiveness on the basepaths. But I don’t know how you could measure that to show it works in the clutch… The other thing is potentially bullpen management? Scioscia always seems to bring in the right guy late in close games. But again, no idea how you would quantify that to show how it correlates to “clutchness”. Could just be one of those “intangible” things where Scioscia just knows his players better than other managers know their players (and I know “intangible” is a dirty word on this site, but it really is hard to ignore when every MLB player past or present uses the term and believes in it).

        Vote -1 Vote +1

      • Torgen says:

        I wondered about it being something to do with aggressiveness, on the basis that you don’t steal bases unless it’s close, which means that base stealing runs have a disproportionate effect on wins, but it turned out that the Angels were terrible at catching runners stealing and got caught stealing all the time, which should have made them unclutch if the hypothesis was true.

        Vote -1 Vote +1

    • Keanu's Favourite Line says:

      …that’s… a fantastic idea.

      What’s the home/road splits for their clutchness?

      Vote -1 Vote +1

  2. Chris says:

    You guys obviously haven’t seen the movie “Angels in the Outfield”.

    +16 Vote -1 Vote +1

  3. Aaron Murray says:

    Could just be luck. 30 teams in the bigs what are the odds than no team consistently scores highly in the “clutch” category?

    Vote -1 Vote +1

    • DavidCEisen says:

      Most likely.

      Vote -1 Vote +1

    • DavidCEisen says:

      Seriously though, this is a joke right? The statement: “At this point, the odds of it just being luck are pretty slim, so it seems reasonable to suspect that the Angels are doing something right,” is a terrible example of statistical analysis.

      The odds of one particular baseball team putting up somewhat consistently high clutch ratings by chance is slim. The odds of a baseball team putting up somewhat consistently high clutch ratings by chance is not slim. You’re trying to create a narrative where one probably doesn’t exist.

      Vote -1 Vote +1

      • Aaron Murray says:

        I agree. I think that we’re forgetting that there are thirty teams out there so the fact that this has been done is simply not that unlikely, even assuming random Clutchiness. It’s kind of like saying that the odds of getting struck by lightning are really small for any individual so the fact that Joe Blow was struck last week must mean that he did something to attract the lightning.

        Vote -1 Vote +1

  4. scatterbrian says:

    Offensively, I think it’s a combination of hitters with high contact rates, and the Angels employment of the contact play. Most Angels run on contact, and are often sent home from third, and they force teams to make plays. I have no idea if they do this more than other teams, but it definitely seems to work for them.

    Vote -1 Vote +1

  5. MJ says:

    rally. monkey.

    Vote -1 Vote +1

  6. TheJoeFrom1993 says:

    The Angels are not very transparent. I have searched for sometime now for something tangible via quotes from the manager or the front office about how, what, or why they do what they do. Or even quotes which at least give a deeper glimpse into organization philosophy. I’m not sure the Angels have ever disclosed if they employ a sabermetric analyst. (They may or may not. I don’t know.).
    Seems that the Angels are kind of hush-hush.

    Vote -1 Vote +1

  7. t ball says:

    Three possibilities, probably sharing a part:
    1. good roster depth. They have a roster with players able to play well. Some of them do so at the right time each year, some of them don’t, but they construct a roster with depth.
    2. A manager who gets the most out of his players.
    3. Luck. Yes, it is possible that they’ve been lucky combined with the other two factors. Just because the chances are slim does not mean there is no chance it’s luck.

    Vote -1 Vote +1

  8. frank says:

    What is the error bar on the clutch stats? And is this significant?

    Looking at the clutch stat, the Astros lead all of MLB this year, led MLB last year and were 2nd in 2008.

    I guess Astros + clutch does not fit the same perception as Angels+clutch…but shouldn’t this article be about the Astros? Or perhaps trying to mix and match winning with this clutch stat is sketchy….

    Vote -1 Vote +1

    • Dave Cameron says:

      You’re only looking at hitting. You need to add hitting and pitching together to get total clutch value.

      Vote -1 Vote +1

    • CJ says:

      One thing the Astros and Angels have in common: emphasis on contact hitting players. Maybe the high contact rates has something to do with the clutch hitting.

      Vote -1 Vote +1

      • Jason B says:

        Do the Astros have an emphasis on contact hitters? I mean, it appears they have an emphasis on *no* hitting from SS.

        And C.

        And 3B.

        And…

        Vote -1 Vote +1

    • vivaelpujols says:

      “error bars on clutch”

      What does that even mean? It’s a stat that’s self defined – it’s not an estimate like UZR. Do we need error bars on batting average?

      Vote -1 Vote +1

      • Torjazz says:

        I think it would mean something like error bars on true talent level for batting average. If you were making a projection, you’d want to get a sense of how tight your estimate could be. Projecting their ‘clutchiness’ going forward would depend on such error bars. Dave is evidently convinced that their error bars don’t include zero.

        Vote -1 Vote +1

  9. notdissertating says:

    I’m not sure focusing on the players who do or do not show as clutch Angels will be very informative. There could be something going on at the team level that makes all its players clutch, but because clutchiness is so noisy at the player level (I’m guessing) you won’t see anything there. Rodney could be clutch this year and appear to be useless in high-leverage situations next year due to bad luck, but still have that thing that is Angels clutch working in his favor.

    Maybe instead, look at team level performance in high leverage situations – is it something as simple as not laying down SF bunt when swinging away is optimal (or vice versa)? Not sure what you’d look for with situational pitching, though…

    Vote -1 Vote +1

  10. Aaron Murray says:

    I’d still like to see some numbers suggesting that the odds against one team consistently scoring this highly on Clutch is more than 1/30, or say one in 50 to give a little wiggle room, if there is no reason for the Clutch. Until then I don’t see any reason to look for reasons.

    Vote -1 Vote +1

    • Aaron Murray says:

      “Less” than 1/30 I meant.

      Vote -1 Vote +1

    • marcello says:

      How could you think it wouldn’t be? That is honestly baffling to me.

      Vote -1 Vote +1

      • Aaron Murray says:

        Really? You have that good of an intuitive grasp on probabilities? Like, if I were to ask you, “If there are twenty-five people in a room what are the odds that at least two of them have the same birthday?” you would instantly say, “Well, a little over 50%, I’d say.”

        If so then kudos. I simply don’t have such an intuitive grasp on the numbers. Also, if the numbers presented below by Bisonaudit are correct then it is very likely that random fluctuation could explain such an extended streak of positive Clutch scores so I think that your intuition may be incorrect in this case.

        Vote -1 Vote +1

      • Aaron Murray says:

        Perhaps your misunderstanding stems from my unclear setup-what I meant is what are the odds that ANY one team out of the thirty has such a string of Clutchiness. Because looking back and saying what are the odds that the Angels have that streak is ignoring the fact that, had it been the Mariners who achieved the feat we would have been wondering what the Mariners have done to produce those Clutch numbers.

        Vote -1 Vote +1

  11. bisonaudit says:

    Assuming that clutchness is random and centered on zero the chances of one of the 30 teams recording a positive batting clutch score for 8 of the last 9 seasons (as the angles have done) are 1.

    Vote -1 Vote +1

    • bisonaudit says:

      The A’s did the same thing for the 9 year period ending in 2006.

      Vote -1 Vote +1

      • Aaron Murray says:

        Thanks for the help but I’m not understanding. The odds are 1 in what?

        Vote -1 Vote +1

      • Kevin S. says:

        Not one in something, but 1 (as in 100%, which is the same as saying 100/100, which is obviously one). Statisticians tend to use decimals to represent probability, rather than percentages, so the range is [0,1].

        Vote -1 Vote +1

      • bisonaudit says:

        1 in 1 Andy. More or less a stone cold lock.

        Vote -1 Vote +1

      • Aaron Murray says:

        Thanks for the response. That definitely answers my questions – it’s very likely that a consistent run of positive Clutch could be the result of random variation.

        It can’t really be 1, though, can it? I mean, in one year fifteen teams could be above and fifteen below, then switch the next year, then switch again, then switch again-no teams at that point can have 8 out of 9. Is it rounded to 1 from the high 90%’s? Not central to my point, I’m just curious about the numbers.

        Vote -1 Vote +1

      • Jason B says:

        I wondered the same thing Andy. Perhaps high (in the 90’s) but a statistical lock at 100%? I can’t fathom that. Unfathomable!

        (I like the word ‘fathom’ also.)

        Vote -1 Vote +1

    • Steve says:

      1 is much bigger than 0. It is unlikely, but not impossible. I agree with the general sentiment that there’s likely some larger force at work here, but it could just be randomness.

      Vote -1 Vote +1

      • bisonaudit says:

        My conviction is that it is random.

        The Phillies busted a 10 year cooler last season (3/100) assuming random distribution, and I don’t remember anyone disparaging them for not being clutch hitters. Longest streak good or bad with the sticks since expansion.

        Vote -1 Vote +1

  12. Franco says:

    Just spitballing, but maybe the GM targets players who have a rep around the league of cool under pressure? I don’t know if I ever bought clutchiness with hitters, but I’ve seen it in New York with pitchers time and time again. Nervous middle relievers who walk lead off men in close games consistently, etc. The press kills them after a couple bad outings and it gets worse.

    Vote -1 Vote +1

  13. DR says:

    Maybe players are instructed not to try hard and to perform poorly than usual in low-leverage situations. This would increase their apparent clutch scores.

    Vote -1 Vote +1

  14. bSpittle says:

    When there are no logical explanations, there’s always Cheating as an explanation.

    Vote -1 Vote +1

  15. Aaron Murray says:

    I’m still a bit confused about why the random variation explanation is not satisfactory for most people. We would all scoff at an announcer who proclaims that because a batter is 6 for 10 against a certain pitcher that he should be pitched around but isn’t this the same thing? Looking for causation when there is no evidence that there is any? Especially given the turnover of players in the time period in question. If the numbers presented by Bisonaudit are incorrect and such a string of positive Clutch is unlikely then, by all means, feel free to step in and present the correct analysis but if it’s true that the odds of at least one ML team putting up positive Clutch numbers in 8 of the last 9 years approaches 1 then what else are we looking for?

    Vote -1 Vote +1

    • Jason B says:

      I would generally agree, but would hasten to add that given that while it is very statistically likely that you would have at least 1 team in 30 with a high “clutch” performance eight times in nine years, that fact doesn’t necessaily preclude another (non-random) explanation. I’m just not sure we’ve heard a good one offered yet that is superior to a statistical anomaly, and has some basis in fact and appears to be somewhat ‘proveable’ and not just someone’s opinion or guesswork.

      Vote -1 Vote +1

      • DavidCEisen says:

        I’m sorry, but the simplest explanation is random variation. Building a narrative in this situation is bunk. Its possible that…. Yeah its possible that 100,000,000 things could be happening, or its possible that this is an expected outcome of random variation. Stories are nice, but often wrong.

        Vote -1 Vote +1

      • Aaron Murray says:

        Jason, it’s true that there might be some causal factor beyond random variation but if there’s no need to use such an explanation, and there’s no evidence besides the fact that there is a string of Clutchiness, then why spend so much effort looking for a reason? It’s like hearing that someone won the lottery twice. Yeah, it’s possible that he has friends in the lottery commission or can influence the numbers with his mind but without any evidence of those events there’s no reason not to conclude that most likely the guy is just very lucky. And with all of the lotteries out there and all of the people playing it’s not so surprising that every once in a while somebody wins twice. In a vacuum it seems like a really unlikely event but really the unlikely thing would be no one ever winning twice.

        I still think that a lot of people are falling into the same trap that so many analysts and commentators do when they talk about a batter owning a pitcher because he likes a certain kind of pitch or a pitcher pitching well against a certain team because…I don’t know, he likes their team colors? That kind of stuff is all over the airwaves, people failing to realize how much variation can be due to complete randomness. We don’t need reasons for everything to prove that we’re smart.

        Vote -1 Vote +1

      • Zu Long says:

        Actually, there is other evidence besides the string of clutchness. The Angels have consistently outperformed pre-season statistical modeling over the last 6 years. For convenience and because the data is readily available, I’ll use PECOTA to demonstrate this.

        2004: +10 games
        2005: +12 games
        2006: +8 games
        2007: +8 games
        2008: +12 games
        2009: +16 games

        For the record, no other team can claim to have been either under or over predicted by pecota in more than 4 consecutive seasons. Neither can they claim the same consistency. Indeed, even if we accept Rob Neyer’s assertion that the standard deviation for pre-season prediction engines is 6-7 games, we find that the Angels have fallen outside of that range all 6 years. Other prediction systems such as CHONE or Diamond Mind have a lesser track record, but have similar results over the past 3-4 seasons.

        This ability seems to correlate with the perception that there is something beyond “luck” occuring here.

        Vote -1 Vote +1

      • Wally says:

        Zu,

        That could just as easily prove that your model sucks, as it could prove that the Angels are good at outperforming their baseline talent. And PECOTA has had plenty of criticism in the saber world over the last, oh 4 years or so.

        Vote -1 Vote +1

      • Zu Long says:

        Wally-

        Again, the only reason I use PECOTA is that it has a longer track record. Again, CHONE and DIAMOND MIND have not faired any better, nor do I believe any others have significantly outperformed them. Your criticisms also don’t address why the Angels and ONLY the Angels would be able to significantly outperform their PECOTA projections by remarkably similar margins every single year when other teams, while occasionally trending one way or the other, experience significant variation.

        I suppose from the perspective of an Angels fan though, how the team consistently wins despite sour predictions and indicators of doom (whether you call it luck, clutch, or the Rally Monkey) matters less than the fact that it shows no sign of disappearing any time soon.

        Vote -1 Vote +1

      • Aaron Murray says:

        Zu, I think it’s probably true that the PECOTA data are more persuasive than the simple positive Clutch numbers. My understanding is that the odds of outperforming a given record by one standard deviation are about 1 in 6 so doing so in six consecutive seasons is a long shot. Still, there are lots of problems in projections but I’m a little bit more swayed but not sold.

        Also, you’re assuming what you’re trying to prove when you say that the Angels show no signs of losing their Clutchiness. Think of it this way; if I flipped a coin eight times in a row and got heads eight times I wouldn’t then say, “this coin shows no signs of losing its tendency to come up heads.”

        Vote -1 Vote +1

      • Jason B says:

        David & Aaron –

        You misunderstood what I was getting at. Random variation *is* the simplest argument, and no other satisfactory ones have really been offered yet; but just because we haven’t been able to put a finger on some other explanation doesn’t mean its not there.

        I wasn’t saying it *is* something else; just that you can’t throw up your hands, say “random!” and stop looking. I’m not suggesting that anyone construct an elaborate narrative so that we can call it something else, anything else, besides random variation.

        The simplest explanation is often the correct one (and I haven’t seen a better one offered to explain this trend yet), but that doesn’t make it infallible, nor does it mean the search for an alternate cause should stop.

        Vote -1 Vote +1

      • Aaron Murray says:

        I don’t think we’ve misunderstood you, Jason, we’ve both said that the fact that this could be random variation doesn’t prove that it IS random variation. I’m also not saying that people should not look for a reason, people are free to do whatever they want. But what I am saying is that, without more evidence suggesting that normal variation is unlikely to account for such a string of Clutchiness, looking hard for reasons for the Angels run is just as good a use of our time as looking for correlation between minor league jersey numbers and the third digit of players’ career batting averages, or number of people who signed a player’s year book in 10th grade and his salary in his first year of free agency. Without more evidence suggesting that normal variation is responsible for the Angel’s Clutchiness (and if there is such evidence I’m certainly willing to change my conclusions about the issue) we’re fooling ourselves if we think that the answer to the question is any more likely to be fruitful than looking at jersey numbers or year book signature or tea leaves or astrology.

        Vote -1 Vote +1

      • Wally says:

        Zu,

        “Your criticisms also don’t address why the Angels and ONLY the Angels would be able to significantly outperform their PECOTA projections by remarkably similar margins every single year when other teams, while occasionally trending one way or the other, experience significant variation.”

        Well, there could be something PECOTA underrates that Angels typically have.

        Vote -1 Vote +1

  16. ClosetM'sFan says:

    I’m feeling way better now about my preseason pick of the Angels to take the West…it looked grim there for a while. I hate Texas more anyway.

    Vote -1 Vote +1

  17. philkid3 says:

    This article (and Dave by name) was just mentioned on the A’s broadcast.

    All they said was “there’s a very interesting article by Dave Cameron at FanGraphs on the Angels perceived. . . clutchiness.”

    And then there were like two minutes of dead air and he never expanded on that.

    Vote -1 Vote +1

  18. Will Peterson says:

    KS-Test of variable ‘Clutch’ 2000-2010, Angels vs. Everyone Else gives a p value of 0.044 making me inclined to reject the null hypothesis that the distributions are from the same population. t-Test agrees, (p=.009).

    I also wouldn’t assume the mean of ‘clutch’ is zero unless you know something special about the underlying nature of the variable.. I get 95% confidence that the ‘everyone else’ mean is between -0.94 and -0.32. You *can* conclude it is random, if by random you mean normal (p=0.79).

    Vote -1 Vote +1

    • Wally says:

      I’m not sure if this is a statistically valid approach. What you’ve done is looked at your data, found an outlier by inspection, tested if its outlier, and found that it is. Remember there is nothing magical about P=.05. You can still achieve P<.05 from randomness.

      Now if you could find something else that accurately predicts clutchiness outliers, that would be great, but just finding them at all is trivial and doesn't prove it isn't luck.

      Vote -1 Vote +1

      • Will Peterson says:

        I agree that this could be random and that I proved very little aside from the fact that it is an outlier of a certain size. It’s just quantification.

        Vote -1 Vote +1

      • Wally says:

        Yeah, its nice to see that, but I don’t think your KS or T test is accounting for the alpha error inflation created by having so many chances to observe this outlier.

        Vote -1 Vote +1

  19. Aaron Murray says:

    Thanks Will but can you walk me through your comment? My layman math skills aren’t up to snuff and I don’t want to misunderstand.

    Vote -1 Vote +1

  20. Has anyone noticed that the Dodgers were consistently GOOD when Scioscia was there and have been very hit and miss after his departure. The intangible of a happy clubhouse is probably due to the leadership of their Manager and Coaches and character of the players the Angels keep. 2% rally monkey.

    Vote -1 Vote +1

  21. Frank says:

    There’s maybe one or two people on the board who mention the possibility of psychology as a factor. Not including the author. FAIL.

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>