Minnesota Twins News & Rumors Forum
  • Year in Review: Twins Hitters In The Clutch

    Josh Willingham was one of the most valuable hitters in the majors.Oddly, a lot of sabrmetric geeks I know donít like the Win Probability Added (WPA) statistic. I donít want to speak for them as to why, but the comment I hear that drives me the craziest is something like ďAll you have to do is see that Erik Komatsu was more valuable then Ben Revere to understand that it is worthless.Ē

    It drive me crazy because 15 years ago, I would hear the same question from baseball traditionalists when Iíd suggest that an on-base machine like Bobby Abreu was more valuable than a guy with 20 more RBI. And I would say ďYes, thatís exactly what that means.Ē And I felt confident because:

    1. wins for a team correlates closely with run differential and
    2. the runs a team scores mirrors closely their On-base Plus Slugging (OPS)
    3. and a teamís OPS built is on their playersí OPS and
    4. Bobby Abreu has a crazy good OPS.

    And they would say, ďThatís nice that you have all those correlations and stuff, but Abreu only had 79 RBI last year!Ē They might agree with the method, but couldnít accept the results. To me, thatís just being closed minded.

    Similarly, I believe in the method of computing WPA. Hereís how it works:

    1. Analyzing dozens of years of baseball, you compute every game situation and how often a team in that situation won or lost the game. So, for instance, a home team that has runners on the corners and one out and is down by a run may have won games 55% of the time.
    2. Give the batter and the pitcher credit for how much they change those probabilities. So if the batter bounces into a double play, and the percent chance drops to 30%, then the batter loses .25 points and the pitcher gains .25 points.
    3. Do this for every play of every game throughout the year.


    Itís not perfect Ė it doesnít take into account fielding. If also isnít especially predictive. And a player who plays a lot has plenty of opportunities for negative scores as well as positive scores.

    However, it also is not dependent on other players; the player who is on third isnít affected by the batter who grounds into the double play. And it rewards players who make big hits Ė hits that change the course of the game. Finally, if you look at the players with the highest and lowest WPA at the end of the game, it is almost never a surprise. It lines up with who you, as a fan, thought the heroes and goats of the game were.

    In fact, I have rarely heard anyone criticize the method. Itís fairly simple to understand and, though it means handling a lot of data, the logic is straight-forward and elegant. But the resultsÖthatís a different story.

    And that will be the case when you see the Twins WPA this year. Today weíll start with the hitters and get to the pitchers next time:



    I suspect few people have trouble with the top two names on the list. Willingham not only had an enormous positive impact in games amongst Twins, his is one of the highest in the majors. It is the 2nd highest, right now, in the American League, sandwiched between Mike Trout and Miguel Cabrera. (Actually, Cabrera is fourth. Edwin Encarnacion is in 3rd, .05 points above Cabrera.

    And fifth in the AL belongs to the next name on that list Ė Joe Mauer. Heís currently above Prince Fielder but a few percentage points. For all the talk about how ďclutchĒ Mauer might not be and how many double-plays he grounds into, Mauer has had an enormous positive impact on the Twins this year. Statistically, itís not debatable.

    On the other hand, I suspect some folks are going to have trouble accepting that Ryan Doumit and Ben Revere have, offensively at least, cost the Twins several wins. Statistically, both have been fairly strong, but overall, theyíve had a lot more negative impacts on games than positive impacts so far this year. Because of that, they rank lower than subs that arenít even with the team any more like Ö. well, Erik Komatsu.

    That doesnít mean the statistic is worthless. It just means Revere (and Doumit) didnít have the offensive impact that we would have like to have seen. Of this, Iím confident.
    This article was originally published in blog: Year in Review: Twins Hitters WPA started by John Bonnes
    Comments 15 Comments
    1. Curt's Avatar
      Curt -
      Thanks for posting this. Though I have seen this referenced occasionally I've never delved into it. Doumit is most surprising to me (aside from Thomas which is SSS). Not that he is negative but that he is worse than Span, Revere and Carroll. I would have expected him to be in that group.

      Clearly this is a function of opportunity so a rate might be in order. Here are the values pro-rated to 600 plate appearances:

      Player PA WPA WPA/600 PA
      Josh Willingham 615 5.04 4.92
      Joe Mauer 606 3.88 3.84
      Clete Thomas 29 0.05 1.03
      Justin Morneau 553 0.87 0.94
      Darin Mastroianni 186 0.11 0.35
      Chris Parmelee 181 0.09 0.30
      Matt Carson 48 -0.05 -0.63
      Trevor Plouffe 439 -0.48 -0.66
      Denard Span 537 -1.04 -1.16
      Jamey Carroll 521 -1.01 -1.16
      Ben Revere 518 -1.11 -1.29
      Erik Komatsu 37 -0.12 -1.95
      Alexi Casilla 313 -1.05 -2.01
      Ryan Doumit 508 -2.13 -2.52
      Tsuyoshi Nishioka 14 -0.06 -2.57
      Danny Valencia 132 -0.60 -2.73
      Pedro Florimon 125 -0.75 -3.60
      Brian Dozier 340 -2.16 -3.81
      Drew Butera 114 -0.74 -3.89
      Eduardo Escobar 39 -0.44 -6.77
      Chris Herrmann 10 -0.13 -7.80
      Luke Hughes 11 -0.22 -12.00
      Sean Burroughs 18 -0.59 -19.67
    1. snepp's Avatar
      snepp -
      I like WPA, it's a fun tool, and nice to look at when you don't feel like digging into the depths of a player's situational numbers.
    1. Brock Beauchamp's Avatar
      Brock Beauchamp -
      Doumit is quite the surprise there, as is Span. The rest, not so much.
    1. 70charger's Avatar
      70charger -
      On the other hand, I suspect some folks are going to have trouble accepting that Ryan Doumit and Ben Revere have, offensively at least, cost the Twins several wins. Statistically, both have been fairly strong, but overall, they’ve had a lot more negative impacts on games than positive impacts so far this year. Because of that, they rank lower than subs that aren’t even with the team any more like …. well, Erik Komatsu.
      Eyes lie and data don't. If people can't accept that this is what happened, then they're creating a narrative that doesn't fit the data.

      What I think people probably have a hard time understanding is why "good" baseball players end up lower than "bad" ones. Ultimately, it isn't a function of whether they're good or bad, it's a function of how they performed in certain situations. It's an ex post measurement.

      But ex ante, I'm absolutely running Ryan Doumit out there as a pinch hitter before Erik Komatsu. After all, past performance is no indicator of future results.
    1. 70charger's Avatar
      70charger -
      One final thing, apropos of nothing here. Since this is a measurement of how well the hitters have done in the clutch, should the Mauer haters really be complaining about Joe taking a ninth inning walk when the guy who comes up next is Willingham?
    1. Seth Stohs's Avatar
      Seth Stohs -
      It likely means that Revere or Span are the guy on 3B or 1B in those situations where Mauer or Willingham came through. Doumit may have come up after Willingham already came up with the big hit.

      It's a fun little state, but there are about 25 stats I would use first.
    1. ashburyjohn's Avatar
      ashburyjohn -
      Am I right that WPA is defined in a way such that 0.0 is league-average? And it's not in any way positionally adjusted? Then it's not a surprise if both Span and Revere show up with slightly negative numbers per PA - their offense is good for defensively gifted players (as they indeed are), but they would be quite disappointing at bat (and wasted on defense) if stationed at first base.

      As Seth points out, what's probably missing from WPA is an "opportunity" adjustment, if for instance a decent hitter like Doumit comes to bat with fewer opportunities for large-leverage results due to someone ahead of him in the lineup doing his own job well.

      I took a look at Fangraphs, and there they say: "Cumulatively, season-long WPA is not predictive, making it an ineffective number for projections of a player’s talent. ... A fun way to think of WPA is as a storytelling statistic." I guess I can go with their judgement on that.
    1. Willihammer's Avatar
      Willihammer -
      Speaking of stories, here are a few from Posnanski. The 10 clutchest performances of all time, as defined by WPA

      http://joeposnanski.blogspot.com/201...d-stories.html
    1. Jeff P's Avatar
      Jeff P -
      Nice article, it was interesting to read. A few thoughts/questions:

      1) Does it take into account stolen bases?

      2) I wonder how much SSS impacts this calculation, even for players with 600 PA. In other words, do the results of the 10-20 PA (or whatever the number is) in the 8th-9th innings with runners on in a close game make a huge impact on the calculation? Not saying they do, just curious.

      3) Intuitively this is a good indicator of how clutch the hitter is, but I don't think that is a complete overlap with how good the hitter is overall so personally I still like the main BA/OBP/SLG stats better along with stats like WAR which I know encompases defense and by position offense.

      Jeff
    1. Willihammer's Avatar
      Willihammer -
      Willingham's top 3 clutchest games.

      3. Apr 20 @ TBR. Top 7, 2 out, bases loaded, 1-2 count. Joel Peralta pitching. Twins down 2-4. Willingham doubles to clear the bases. His only hit of the day. + WPA .42

      2. June 24, @ Cin. Top 7, 0 out, Mauer on 1st. Score 1-1. Willingham doubles Mauer to 3rd. +WPA .17 on the play. Then in the ninth, against Chapman, Twins down 2-3, Span on 2nd, 3-1 count. Willingham hits a home run. +WPA .59 for the play and +.71 on the day.

      1. May 29 vs. Oak. Bottom of 9th, 2 out, Carrol on 3rd, Mauer on 1st, Twins down 0-2. Willingham homers off a 1-0 Fuentes pitch to win the game. +.91 WPA on the play, and .91 for the day.
    1. snepp's Avatar
      snepp -
      Quote Originally Posted by Jeff P View Post
      1) Does it take into account stolen bases?

      2) I wonder how much SSS impacts this calculation, even for players with 600 PA. In other words, do the results of the 10-20 PA (or whatever the number is) in the 8th-9th innings with runners on in a close game make a huge impact on the calculation? Not saying they do, just curious.

      3) Intuitively this is a good indicator of how clutch the hitter is, but I don't think that is a complete overlap with how good the hitter is overall so personally I still like the main BA/OBP/SLG stats better along with stats like WAR which I know encompases defense and by position offense.
      1. No, it's strictly the results of the batter/pitcher matchup.

      2. Yes, small samples can potentially make a huge difference.

      3. It's not intended for any kind of talent evaluation. Per its creator, "WPA is not a way to evaluate the talent of a player."
    1. John Bonnes's Avatar
      John Bonnes -
      Quote Originally Posted by Seth Stohs View Post
      It likely means that Revere or Span are the guy on 3B or 1B in those situations where Mauer or Willingham came through. Doumit may have come up after Willingham already came up with the big hit.

      It's a fun little state, but there are about 25 stats I would use first.
      Seth, that's not true. Revere and Span get credit for getting themselves to 3B.
    1. CDog's Avatar
      CDog -
      WPA/LI (also on fangraphs) tries to get at what people are looking for with an attempt to "rate" the stat by opportunity.
    1. Oxtung's Avatar
      Oxtung -
      Quote Originally Posted by John Bonnes View Post
      Quote Originally Posted by Seth Stohs View Post
      It likely means that Revere or Span are the guy on 3B or 1B in those situations where Mauer or Willingham came through. Doumit may have come up after Willingham already came up with the big hit.

      It's a fun little state, but there are about 25 stats I would use first.
      Seth, that's not true. Revere and Span get credit for getting themselves to 3B.
      I tried to look up the formula or even a good explanation of this stat but couldn't find one. Anyone have a link?
    1. lightfoot789's Avatar
      lightfoot789 -
      How about a couple of Twins Farm Hands- DJ Hicks for hitting a Grand Slam to win the Appy Championship on 3-2 count in bottom of 12th inning or Adam Walker for tying the game on a 3 run HR with 2 outs in bottom of 9th inning of the Final game of championship series. Win or Go Home!!! That's CLUTCH
      How Bout them Elizabethton Twins
©2014 TwinsCentric, LLC. All Rights Reserved.
Interested in advertising with Twins Daily? Click here.