Why the Win is a Bad Baseball Statistic – Example 2,346,717

I’ve mentioned before, specifically in the context of pointing out why I think that over their respective careers, Mike Mussina was a better pitcher than Tom Glavine (see here), that wins is a poor measure of how good a pitcher really is.  So many outside factors go into which pitcher gets the win, that it isn’t really about being a “winner.”

Take, for instance, today’s Yankee game, and this line:

E Ramirez (W, 3-0), 1 IP, 2 H, 4 ER, 2 BB, 2 K, 1 HR

The context: Edwar Ramirez (yes, the same one that I annointed as the next coming of Mariano Rivera after his first MLB inning) was brought in for the 8th inning to protect an 8-5 lead.  He proceeded to blow it, and left the game after the top of the 8th down 9-8.  The Yankees came back, scored half a dozen runs in the bottom of the frame for the winning margin of 14-8.  Because Edwar was the pitcher of record when the Yankees scored the go-ahead runs, he was given the “win.”  Not because he did anything to help the Yankees “win” or was the “winner” but because of circumstance. 

But hey, his record is 3-0, he must be having a great season.

That’s why “wins” are a stupid stat, and just not a good measure of how good a pitcher really is.

Advertisements

17 responses to “Why the Win is a Bad Baseball Statistic – Example 2,346,717

  1. My friend asked a good question last week: How much would you have gotten if you’d bet on the AL league leaders in Wins as of August 1st would be Lee, Mussina, and Saunders?

  2. I don’t think “wins” are a stupid stat – you just have to take into account what it really means. I like the win because I like the formality of having a pitcher of record, but obviously one shouldn’t judge a pitcher just by his win/loss record.

  3. Adam – They’re a poor measurement of how good a pitcher someone is, certainly. What does it really mean?

  4. Adam, I don’t have a problem with the “win” as it stands by itself, I have a problem with the way it’s used.

    For instance, if you listened to WFAN a few weeks ago, they were actually debating whether, at 7-6 with an ERA of 2.76 or something, Johan Santana was having a good season. As if the games he pitched when he gave up one run and got 0 in supports, or the several games his bullpen blew for him were indicative of how he pitched. I’ll take any pitcher with an ERA of 2.76 and 5-6 record over a pitcher with a 12-1 record and an ERA of 3.95 any day (obviously, on a going forward basis).

    As for needing a pitcher of record, I kind of understand, but then you end up with situations like the one in the Yankee game record, where Ramirez ends up winning the game, but really didn’t do anything to advance the cause of the Yankee win. (Quite the opposite, in fact.)

  5. Okay, so that’s one of baseball quirks.

    Obviously, when evaluating a player (I’ll qualify this statement by adding: “in the short term”), you need to go by more than just wins.

    That said, over the span of a career, the lucky wins and unlucky losses should even out. And if a pitcher gets too many no-decisions because he played for a team with a bad bullpen, then that’s partly his own doing – had he gone deeper into games, the bullpen would have less of a chance to blow them. Yes, Johan would probably be a 12-game winner right now (or maybe more) if not for the bullpen. But the bullpen was only necessary because he couldn’t go deeper, so it is, in fact, a reflection of how good a chance he gives his team to win.

  6. Also – there is obviously a difference in the meaningfulness of the “win” when applied to a starting pitcher vs. a relief pitcher.

  7. over the span of a career, the lucky wins and unlucky losses should even out

    Not necessarily. Take, for instance, a comparison between Tom Glavine and Mike Mussina. For the bulk of their careers, Mussina and Glavine played for teams that were vastly different (Glavine for the dynasty Braves and Mussina for the so-so Orioles). Glavine has many more wins for his career, but wasn’t the better pitcher (when you adjust the other numbers for ballpark and league influence).

    Aside from just bullpen, there’s defense and offense that factors into the pitchers “win” and it simply isn’t reflective of ability, as much as, say, ERA or WHIP.

    See this post at FJM for a little more on Johan Santana this season. Notably:

    Here are the scores of the games the Mets have lost with Santana pitching, and the # of earned runs Santana gave up while in the game:

    3-2 (2)
    3-2 (3)
    5-2 (1)
    6-1 (4)
    5-4 (0)
    2-1 (1)
    4-2 (3)
    5-3 (4)
    3-1 (1)

    So, in the nine Santana-involved losses the Mets have suffered, they, the Mets, have scored a total of 18 runs. 2 runs a game. Their offense averages 2 runs a game, in those losses.

    This is somehow reflective of Santana’s abilities? I don’t think so.

  8. That’s not a complete list losses with Santana missing. It’s missing the last two blown saves.

  9. Sorry, as of the date of the post…July 21st.

    Add in

    8-6 (2)
    5-4 (1)

    changes the average to 2.55.

    Point is still made, I think.

    And as to your “he should stay in” argument, the choice isn’t always (or ever, really) his, there are factors involved (like long term health concerns leading to pitch counts) that he doesn’t control (the team does). And in the 7/22 game against Philly, he went 8 innings and handed the game to his closer, who is supposed to be elite. You really can’t blame him.

  10. What about his wins? What is his ERA in the games he wins? (In other words, while he seems to have suffered many tough-luck losses where his team didn’t hit for him and the bullpen blew a lead, does he also get lucky in his wins? I.e., does he give up many runs but get a lot of offensive run-support? To take it a bit further, does he leave with men on base, and, in his wins, does the bullpen bail him out?)

    That would be a good measure of whether his unlucky losses are balanced off by lucky wins.

    Also, part of my “he should stay in” argument is also about how you can compare pitchers today not just with contemporaries but also historically. Johan leaves games after 100 pitches, usually not more than 7 innings (and often less). So when he ends his career with fewer wins than Bob Feller, it’ll be because Feller started 484 games in his career and completed 279 of them. (Feller also had nearly 100 relief appearances in his career.)

  11. (So, to an extent, the “win” is also a mark of endurance.)

  12. Adam, you comment about endurance would make more sense if the majority of pitchers went close to the distance. But they don’t. I don’t have the stats to prove this bc I’m too lazy but I would bet that aside from Sabathia and Halladay, most starters typically do not go the better part of 7.

    So you’re really just looking at how lucky a pitcher is with how much run support he gets (in this year, it’s been very little for Santana) and how the bullpen holds up after he leaves the game after about 6 innings and 100 pitches.

    Better example: Before tonight, Halladay and Burnett had the same number of wins with 13. Same shitty offense backing them up and basically seeing the same opponents. Yet, aside from striking out more batters, Burnett has been worse in pretty much every other meaningful statistic. I don’t think that anyone would suggest that Burnett is having close to the kind of year as Halladay (or even a good year for AJ’s own standards).

    And why would you say that over the course of a career it should normalize? There are scores of pitchers that pitched their most effective years for crappy teams.

  13. As an aside, what bugs me about the whole win/loss thingy isn’t so much as how it’s misleading.

    It’s how it’s used.

    Because the typical beat writer cites wins (or win/loss record) as the penultimate mark of a pitcher’s success, the typical idiot fan believes things like “he’s a .500 pitcher”, “he just knows how to win” and “he doesn’t know how to keep his team in the game” and completely ignores more meaningful stats.

    The other day someone was trying to convince me how awesome Ted Lilly is this year because of he’s 11-6. I spent about 30 seconds trying to explain to him why he’s wrong, why he’s been very average as a whole this year and is benefitting from the best offense in the NL, didn’t go anywhere, and then just stopped. The casual fan will always revert to what he’s read in the sports section his entire life.

    Kind of like calling a pitch a “breaking ball”.

  14. I’m also too lazy to check, but I’d be interested to see how many runs Johan has given up in games the Mets have won with him on the mound.

  15. Johan’s ERA in different situations:

    No decision for Johan, Mets Loss – 1.53
    No decision for Johan – 2.60
    Mets Loss – 2.64
    Johan Win – 2.70
    Mets Win – 3.07
    Johan Loss – 3.35
    No decision for Johan, Mets Win – 4.59

    Unless I copied and pasted wrong (I spent two minutes in Excel on this), this would sort of support Noam’s thesis. But the sample size for some of the categories is pretty small.

  16. Yeah, I mean it’s clear that the “win” has flaws. I don’t disagree. One thing you will notice from these ERAs is that while Johan has been cheated out a few wins (1.53 ERA in games with a no-decision that the Mets lost), he was also bailed out and spared a few losses (4.59 ERA in games with a no-decision that the Mets won). And, interestingly, his ERA is a 0.43 runs higher in games the Mets win than games the Mets lose.

  17. The 4.59 ERA stat is based on the smallest sample size among all the different situations. It’s based on three starts (15 2/3 innings), one of which was a 4 IP, 5 ER start in Cincy that the Mets won 10-8.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s