By In Baseball

Winning and Losing, Take 3

Another post on this idea of team wins and losses? Well, you know, in the old days this would have been one massive 6,000-word post. But I’m getting older, so it comes out in segments now.

* * *

In 1963, Sandy Koufax officially won 25 games, which was obviously very good. But if you could team-wins like I think you should, it was actually an all-time year — the Dodgers won 34 of the 40 games Koufax started that season. That’s THIRTY-FOUR wins for Koufax. That’s tied with Whitey Ford in 1961 for most wins in a season since the end of World War II.

So why was Koufax’s record 25-5 instead of 34-6? Glad you asked. Here, one by one, is how he pitched in the nine games his team won but he did not get the credit.

— 6 2/3 innings shutout ball, won game 2-1

— 4 2/3 innings, 6 runs allowed, team came back against Cubs 11-8

— 4/23 innings, 5 runs allowed, team came back against Milwaukee 6-5

— 5 1/3 innings, 3 runs alowed, won game 5-4

— 9 1/3 innings, 1 run, won game in extra innings

— 1/3 inning, four runs, team came back and won 7-5

— 12 innings, 1 run, team won 2-1 in 12 innings.

— 8 2/3 innings, 1 run, team won 2-1

— 7 2/3 innings, 3 runs, team won 5-3

And here’s the one game the team lost and he did not the L.

— 7 innings, 1 run, team lost 3-2 by giving up two runs in the ninth.

Yes, that’s a bit of a mishmash. What can we learn from this? Well, let’s talk a minute about DESERVE, something I generally don’t like to talk about when talking statistics.

Koufax clearly did not DESERVE a win that game when he lasted only a third of an inning.

He probably not really DESERVE the win the two times he did not last five innings and gave up six and five runs respectively.

Heck, if you want to push it a little bit, you can say he did not DESERVE to win the time he gave up three runs in 5 1/3 innings.

So, that’s four wins where you can argue that the team win stat isn’t really fair and that Koufax is taking a win that would be better suited for someone else. Looking specifically at the 1/3 of an inning game, Dick Calmus pitched five innings and allowed just one run. By almost any measure, you would have to see he DESERVES the win more.

Conversely, though, how can you justify not giving Koufax the win when he went 12 innings and gave up one run? You know who got the win in that game? Larry Sherry — for two scoreless innings. How can you justify not giving Koufax the win when he went 9 1/3 innings and gave up one run? Or the time he went 8 2/3 innings and gave up one run? Or the time he threw 6 2/3 shutout innings? You telling me someone else DESERVES the win more those days? No chance. That’s four wins certainly deserved, and I would say that he  deserves the 7 2/3-inning start where gave up three runs.

We come back to the basic fact: People supposedly love simplicity. Well it doesn’t get any simpler than this: When Sandy Koufax started games, his team went 34-6. They did not go 25-5. And this is my point, the one I’m not sure I”m getting across: I’m not coming up with some new way of counting things here. Koufax’s record should be 34-6 because HIS RECORD REALLY WAS 34-6.

Let’s do this same breakdown for Whitey Ford in 1961. Remember, that was the great Yankees offense of Mantle and Maris; that team scored a boatload of runs. Ford’s official record was 25-4. But the team actually went 34-5 when he pitched.

Here is how he pitched in the nine games where the Yankees won but he got a no-decision:

— 6 1/3 innings, 10 runs (6 earned), Yankees came back to win 13-11.

— 8 innings, 2 runs, Yankees won 5-4.

— 4 2/3 innings, 7 runs, Yankees won 11-8.

— 8 innings 5 runs, Yankees won 6-5.

— 10 innings, 6 runs, Yankees won 7-6

— 4 innings, 6 runs, Yankees won 8-6

— 4 2/3 innings, 0 runs, Yankees won 1-0

— 2 innings, 5 runs, Yankees won 7-6

— 6 innings, 0 runs, Yankees won 3-1

And in the one loss:

— 7 innings, 4 runs, Yankees lost 5-4

OK, as a whole, Ford did not pitch nearly as well in his no-decisions as Koufax — that’s good for our purposes. Let’s look at DESERVE again. By our usual thinking he probably only DESERVED a win in two of his nine no-decisions. He gave up at least six funs in four starts, five runs in two others, if we’re playing Judge Dredd he probably did not DESERVE the wins in those games,

But is that really true? The Yankees had a really, really good offense that year. If you have an offense that scores 11 runs, how good do you have to be to DESERVE a win? Once again, by messing around with these special rules, we are not counting things, but we are instead making judgment calls. Here is how many innings the official “winning pitcher” threw in each of those games:

— 2 innings
— 1 inning
— 2 innings
— 1 inning
— 3 innings
— 3 innings
— 1 inning
— 6 innings
— 2 innings

Now, you tell me — how many of those pitchers DESERVED the win? Again, the simple fact is that when Whitey Ford took the mound, the 1961 Yankees almost never lost — they won 34 of 39 games. That, to me, is the only underlying truth here.

One of my favorite ever seasons was Ron Guidry’s 1978 season. He famously went 25-3. But really … that was a 30-win season, one of only eight 30-win seasons since World War II. Guidry went 30-5 in his starts.

The five times the Yankees won but Guidry did not officially get the win he pitched like so:

— 6 2/3 innings, 3 runs, Yankees won 4-3

— 6 1/3 innings, 2 runs (0 earned), Yankees won 3-2

— 6 innings, 4 runs, Yankees won 6-4

— 9 innings, 5 runs, Yankees won 7-6

— 6 2/3 innings, 3 runs (2 earned), Yankees won 4-3

You telling me he didn’t deserve to win all five of those starts? Sparky Lyle and Goose Gossage vulture those five wins by throwing a couple of innings.Forget that. The guy went 30-5, not 25-3.

Let’s do one more, then I’ll list off all of the 30-game winners since 1945. In 1992, Pete Harnisch went 9-10 with a 3.70 ERA — a thoroughly uninteresting year if you look at his won-loss record. But Harnisch was actually a 20-game winner that year. His 11 winning no-decisions went like so:

— 6 1/3 shutout innings in a 3-1 win.

— 7 shutout innings in a 2-1 win.

— 6 innings, 1 run in a 4-1 win.

— 7 innings, 1 run in a 2-1 win

— 7 innings, 3 runs (2 earned) in a 4-3 win

— 7 shutout innings in a 2-0 win

— 5 innings, 1 run in a 5-4 win

— 2/3 of an inning, 0 runs, in a 5-4 win

— 5 innings, 6 runs in a 14-9 win

— 6 1/3 innings, 4 runs in a 7-6 win

— 4 2/3 innings, 3 runs in a 6-5 win

He pitched very well in 7 or 8 of these games and only pitched very poorly twice. Instead, those 11 wins were spread out to:

— Xavier Hernandez (1 scoreless)

— Hernandez (1 scoreless)

— Hernandez (1 2/3 scoreless)

— Joe Boever (2 scoreless)

— Al Osuna (1 scoreless)

— Rob Murphy (1 scoreless)

— Boever (2/3 inning scoreless)

— Hernandez (2 scoreless)

— Willie Blair (1 inning, 1 run)

— Doug Jones (1 1/3 scoreless)

— Osuna (1 scoreless)

You tell me how those guys deserve the win more than Harnisch.

OK, here are your 30-game winners since World War II ended:

Whitey Ford, 1961
Won-loss record: 25-4
Actual record: 34-5

Sandy Koufax, 1963
Won-loss record: 25-5
Actual record: 34-6

Denny McLain, 1968
Won-loss record: 31-6
Actual record: 33-8

Mike Cuellar, 1970
Won-loss record: 24-8
Actual record: 31-9

Don Drysdale, 1962
Won-loss record: 25-9
Actual record: 31-10

Ron Guidry, 1978
Won-loss record: 25-3
Actual record: 30-5

Mudcat Grant, 1965*
Won-loss record: 21-7
Actual record: 30-9

Sandy Koufax, 1965
Won-loss record: 26-8
Actual record: 30-11

*In case you are curious, Mudcat’s nine winning no-decisions are pretty typical. He threw quality starts in five of them, and pitched lousy in four of them. Grant, like Koufax in 1963, had a game where he lasted just 1/3 of an inning. Someone suggested changing the rule so that a pitcher — starter or reliever — has to throw at least one inning in order to get a win. Discuss.

72 Responses to Winning and Losing, Take 3

  1. John M. says:

    How about whichever pitcher records the most outs in a game the team wins, gets the win.

    • Anon says:

      To me, Joe’s version of wins ARE an important stat — just not if you want to evaluate the quality of the pitcher. Rather, the “true” won-loss record is more of a “Fan Appreciation Statistic” (FAS).

      Pitchers with high FAS are pitchers who, when you turn on the TV you say “Yes! This guy is pitching.” Why? Because he’s pitched 14 times this season and you’ve won 10 of them. It’s probably because he’s a good pitcher, but it could be because he’s gotten lucky, or you just have a really good team. But for whatever reason, when you’ve seen this guy warming up before the game, happiness has often followed.

      The best FAS for hitters would probably be something like the (much more complicated) net WPA.

    • NevadaMark says:

      I like it John M. I like it.

    • tangotiger says:

      I like that.

  2. ibrosey says:

    Tough to add exclusions and keep a straight face. The couple of time a pitcher may not ‘deserve’ the win are still far less than the number of times pitchers are currently vultured and screwed.

  3. I’m not convinced. In terms of trying to state how the starting pitcher contributed to the teams’ winning, it’s about as good as counting QB wins in football. QB wins is fairly meaningless because, though the QB (as with the Pitcher) does play a larger role in winning/losing, there are too many other factors that roll up into a win or loss. Some part of the “win” has to do with the opposing pitcher (or QB) and how they performed. There is also how the defense played and key hits. Koufax plays a bigger role than Ford did because of the way the rest of the team was constructed. The Dodgers were a lighter hitting team playing in an extreme pitchers park, while the Yankees were bombers playing in an extreme hitters park. Ford had more room for error than Koufax, but of course, Ford’s home park was less forgiving. So, park effects also play into it.

    Bottom line: I get that a guy that threw 2/3 of an inning doesn’t “deserve” the win. But there are just too many other factors involved in winning the game. I’m not sure why Joe is harping on this one, besides of course, just to have some fun & get some interesting words on a page. But ultimately, it’s replacing one bad stat with another.

    • Ed says:

      I agree that neither are great stats… but Joe’s version seems like a “less bad” version of the stat. People like using win-loss records for pitchers and it’s not going away any time soon, so why not replace the current silly stat with one that’s not quite as bad and nowhere near as convoluted?

    • shoelesskc says:

      It’s all arbitrary. Assigning a W or L to a pitcher is obviously not a solid truth as to whether it is deserved or not. But there is a simplicity to it that adds worth. Same as goalies in hockey are assigned wins and losses. It’s a team win, and they certainly didn’t score any goals in the win (cough.. Hextal) but they are arguably the main contributor to the decision of a game, same as a pitcher in baseball. No one is going to assign wins and losses to the second baseman.
      However, totaling wins and losses provides any obvious fact for teams in determining worth. When they start pitcher A instead of pitcher B they notice they tend to win more games. Whether its fully deserved or not seems to be readily trumped by the simplicity of who has more W’s than L’s. That’s an easier stat to analyze.
      If baseball stats history had started with assigning wins and losses to starting pitchers based on team wins when they start instead of the current 5+ innings with a lead win, we would probably have the same argument that teams wins don’t reflect a pitcher’s worth or true value.
      I apologize for the rambling.

    • dshorwich says:

      bellweather22 –

      Yankee Stadium in 1961 was not an “extreme hitters park” at all – in fact, it was a pitcher-friendly park. Per bb-ref its multi-year park factor was batting: 95 pitching: 92 (one-year batting: 97 pitching: 94). Not as extreme as Dodger Stadium, but still leaning well towards the pitcher side.

    • wogggs says:

      Exactly. Joe’s solution really isn’t any better than what we do now, it is just differently bad and differently less instructive.

      For example, Scott Kazmir of the A’s pitched 3 scoreless (and hitless) innings yesterday before leaving with an injury. Ultimately, the A’s lost 3-2 because Dan Otero gave up a 3 run homer to former A Yoenis Cespedes. Under Joe’s construction, Kazmir gets the loss. How does that make sense? Kazmir had nothing to do with the loss, indeed, he did the most a pitcher can do.

      On the other side of the coin, the Tigers had to start a reliever because the scheduled starter went on the bereavement list. He also pitched 3 innings of hitless ball before being relieved by a guy up from AAA who couldn’t get to the park on time from Toledo to start the game. Under Joe’s construction, the starter gets the win. Maybe that would make sense, as he contributed as much as any pitcher to the Tigers’ victory by not allowing the A’s to score during his time on the mound.

      Overall, I think I would leave things the way they are, if the only two choices are the current system and Joes’ solution.

      • Noah says:

        …which is why I think using WPA, while certainly not perfect, is at least a modest improvement. It certainly doesn’t tell the full story, but it definitely tells part of it in a quantifiable, easily-looked-up way. If I see Pitcher A get a WPA-Win, I know for a fact that by one metric, Pitcher A was most responsible for his team’s victory.

  4. PhilM says:

    My technique for addressing this issue is to use annual ERA+ figures and negative binomial distribution to allocate a pitcher’s wins and losses as though they were pitching for an average team. That doesn’t tackle the question of assigning the DECISION, which is think is more what Joe is pointing out here. So I end up with a more “pragmatic” win-loss record that has the same number of decisions: thus Walter Johnson is 476-220 instead of 417-279.

    • David says:

      A fairly easy way to do this is with Pythagorean W-L. Take ERA+ and square it in the numerator; the denominator is ERA+ squared plus 100 squared (the average runs scored by and average team). That gives a winning percentage, with you multiply by (IP/9), which gives a pretty damn good estimate for number of decisions. I get a “fair” W-L of 449-208.

      • PhilM says:

        You and I are both coming up with about WinPct = .683 for the Big Train, so I think we’re saying the same thing. But I guess my point is that I am interested in finding something to evaluate the actual pitcher’s ability/performance by improving the win-loss allocation, while JoPos is talking about eliminating win-loss entirely as a measure of the pitcher. I’m trying to neutralize the team, while Joe is highlighting only team performance.

    • flcounselor says:

      That works fine for you and for many of the readers of this blog.mbut for the average fan, let alone the casual fan, it is totally inaccessible.

      There are already enough sophisticated formulas for stat heads. So with other stats it would be nice to keep the average fan in mind.

  5. njwv says:

    Has anyone suggested anything like giving the win to the pitcher who pitched the most innings?
    Not sure how to go about assigning blame for losses though.

    • VTmike says:

      “Assigning blame” is exactly what Joe is trying to get away from. His whole point is to record things that are true. Did the team win? Then it is a win. If they lost, then that pitcher started, and lost. That’s all. “Deserve” is nebulous nonsense that has no place in recording statistics. Joe went down this exact same path a few years ago, trying to remove errors from batting average & OBP. Stop making up false histories about what could have happened or you think should have happened. Record the truth.

      • jroth95 says:

        And yet Joe doesn’t complain about DIPS, which is 100% about “what should have happened.” It kills me that Joe fails to see that the spirit behind ERA and FIP are identical, even though they are based on very different premises.

      • njwv says:

        The thing though is that whatever stat you use is making a value judgement in some way. It’s not a truth or not thing but rather a question of what story you want the statistic to tell. Joe’s W/L tells the story that starting pitchers deserve the credit/blame for whether or not the team wins. Other suggestions on here tell the story that the pitcher who pitches the most innings in a game or gives up the most runs in a game deserves credit for the outcome.

        Those are all legitimate points of view. But the debate about stats is really a debate about what story you’re telling with them and whether or not those are the stories we’re interested in.

  6. Ben Parker says:

    This is not my own thinking, exactly, but part of the appeal of the pitcher W might be precisely that it is something vanishing, something that people used to rack up in great numbers, because they were “true men” or giants of yore, and now that we are namby-pambies, and don’t throw complete games any more, the cost is that wins decline. (I should say, I am focusing on the decline of the 20-win season. If you follow Joe’s rule, you will be artificially inflating wins BACK UP TO the Cy Young and Warren Spahn numbers.) Again, not my own thinking exactly, but I can see someone liking the W because the Greats of The Past really racked them up, and because our pitchers are more coddled, the same statistic punishes them a little. And Joe’s redefinition, or the idea of quality start, etc. etc. kind of “gives back” that penalty.

    • Karyn says:

      I’m going to nitpick your use of the word ‘artificially’. This is all artificial, including the original definition of pitcher wins. Why 5 innings instead of 4 and 2/3? Because someone made the rule.

  7. I don’t think there is a good way to assign pitcher wins. I agree Joe’s way is “less bad” than the current practice, but ignoring a pitcher’s W-L record is probably the best answer. As much as I hate “quality starts” it is probably a better stat than W-L record. I personally like ERA+ and average innings pitched per start.

  8. Steve says:

    Perhaps it’s best for those who think similarly to Joe to do what football does. Phrase it as ‘Football City is X-Y when Joe Wyoming is the starting quarterback’.

    An MLB broadcaster could say both ‘Sandy Mound has a 7-3 record in 2015’, and ‘The Steel Boots are 10-4 in 2015 when Sandy Mound is the starting pitcher’

    If I’m a Steel Boots fan, I want to know the 10-4 record. Then again, that’s just me.


  9. rucksack says:

    A simple alternate proposal:

    The win goes to the pitcher on the winning team whose (innings – runs) is the highest.
    The loss goes to the pitcher on the losing team whose (runs – innings) is the highest.

    Apologies if someone thought of it first (or if it’s a bad idea (or both))

    • wogggs says:

      Not a bad idea. How do you handle ties?

      • jroth95 says:

        Extra innings.

      • rucksack says:

        I thought what the other responder thought at first. There are no ties in baseball…
        Innings pitched breaks the tie – win goes to most innings, loss to fewest. This attempts to reward the starter (and punish the firestarter).
        Remaining ties go to the staff or team, aren’t rewarded to an individual, or could be split, I suppose, like sacks are sometimes in the NFL. I imagine these unbreakable ties would be uncommon and usually fall to relievers where we don’t care as much.

  10. The real point Joe is making here is very nearly the same one which has been made about the relationship between run support and pitcher wins. Pitchers who get good run support gets wins, including some in which they don’t pitch well. The beef against pitcher wins which sabermetricians have posited (and have pretty much won by now) is that they all too often reflect run support, which is something pitchers have nothing to do with.

    A lot of what we’re seeing in many of these Posnanski wins (Poswins?) is run support that came LATE instead of early. Koufax pitches 11 innings, leaves with the score 1-1, Regan vultures the win when the Dodgers score in the bottom of the 12th. (I believe that the term “vulture” as a verb was actually invented by Koufax himself to describe exactly this situation — he described Regan as sitting out in the bullpen like a vulture, waiting for him to die so he could come in and get the win.) Koufax got enough run support to win, but he got it so late that he was no longer around to get it according to the traditional rules.

    So to the extent that pitcher wins actually do matter (and regardless of whether we’re discussing wins or Poswins, pitcher wins are so ingrained in the thinking about the game that we’re at least a century away from getting rid of them even if we want to), Poswins would seem to be every bit as legitimate a way of figuring them as the traditional one, and the number of glaringly unfair Poswins is almost certainly less than the number of such traditional wins.

  11. Paul says:

    Though I think most of us agree that Joe’s methodology is better and gives more meaningful information to fans at large, I wonder though if anything COULD ever change. Though this may not be the stat to push to make century long changes, what would it actually take to change the definition of a traditional “baseball card” stat? Based on baseball’s historical fanatical ties to traditional stats, I just fear no altered stat definition will ever see the light of day. Yes, OBP is on the screen now and WHIP sometimes, but could PosWins ever crack into a column of the newspaper, or on a telecast, or in a broadcast from this generation. So I’m on board with the idea, but I pose another question: what would it take in 2015 to alter day to day use of Wins, RBIs, etc. let alone have MLB recognize a change? Or are we all doomed to be limited to reading about wOBA, WAR, etc. only on fine sites like these or FanGraphs, B-Ref, etc as an “advanced” stat?

    • flcounselor says:

      Actually, things can and do change.

      This year the Orlando Sentinel is devoting a half page once a week to listing the leaders in a number of advanced stat categories.

  12. EnzoHernandez11 says:

    If we were inventing baseball in 2015, and we decided we absolutely needed a statistic called “pitcher wins”, I would say, sure, let’s do it the way football and hockey do. But since we already have such a stat and it’s been with us for over one hundred years, the real question should be, “Does this new formulation add sufficient value to reverse a century’s worth of understanding of what it means for a pitcher to win?”

    Based on the analysis above, I’m inclined to say that the answer is “no”. It looks like “starting in games won” is every bit as misleading and context-dependent as the current definition of pitcher’s wins. At the very least, the value added doesn’t seem sufficient to make the change. We might argue for getting rid of pitcher wins altogether (good luck with that), but I don’t think replacing an old murky stat with a new almost equally murky stat moves us forward in any meaningful way.

    We could contrast this with the success the sabermetric community has had in changing the emphasis on evaluating batters from using batting average to using on-base percentage. This was a case where the “new” stat (not really new, just widely ignored) greatly enhanced our understanding of a hitter’s contributions vis-a-vis the old stat. Or even WHIP, which hasn’t replaced ERA, but has gained its own prominence.

    I think baseball analysts have done enough to dilute the meaning of the almighty win (e.g., the CYA no longer automatically goes to the guy with the most W’s), that there’s no need to advocate for a replacement. (Though it would be fun to watch the anti-sabermetrics types, the ones who say that wins and losses are the only way to judge pitchers, go to war against a measure that most simply and cleanly measures wins and losses.)

  13. Noah says:

    No thoughts on WPA? Give the win to the pitcher with the best WPA in the game and the loss to the worst. That would help for pitchers like Ford who gave up a bunch of runs, but had an all time offense. His WPA would be pretty low, because most of the 50% would go to the hitters. It would also help relievers who enter the game in high leverage situations. Meanwhile, going back to hitters adding WPA in high scoring affairs, maybe we should give the win to a position player every once in a while. Thoughts?

    • invitro says:

      GWRBI FTW :).

    • EnzoHernandez11 says:

      One time, just for fun, I tried to replicate Bill James’ win shares using WPA. Since each Team win is worth three win shares, I simply took each team win and gave win shares to the three players on the winning team with the highest net WPA for the game. In some ways, the results tracked James remarkably well, but there were some big distortions. First, closers were geatly over-valued vis-a-vis James. Also, the system underrated (obviously) good defensive players and those who walked a lot. Anyway, I think using WPA to determine W’s and L’s would, as invitro says, be a bit like using GWRBI to rate hitters. Your best closers would end up being your 20 game winners.

      • Noah says:

        I don’t know that I agree with that. If a closer comes in with a 2 or 3 run lead, the team already has at least 95% chance of winning, so it’s minimal. On the other hand, if he comes in with a 1-run lead, the bases loaded, and nobody out, and he manages to close things out, he sure as hell will earn his 75% or whatever it is. Thus, if a closer has a lot of tough, high leverage saves, I don’t see why he shouldn’t be awarded with plenty of WPA. That, after all, is what a closer should be.

        • EnzoHernandez11 says:

          Fair point. I should have mentioned that I was looking at Padres teams from the 70s and 80s, a time when closers often went more than one inning. Still, if I recall correctly, one year, the closer (either Fingers or Gossage) ended up with over 30 “win shares”, which seems excessive.

  14. While I see some of Joe’s logic, I wonder. I would guess that as you go further back in history, using Joe’s system would have less of an impact, at least on the better pitchers, as pitchers tended to complete more games and thus get credit under both systems for the win. Let’s say that assumption is true and we change all pitcher wins to wins under Joe’s system. Let’s say that Cy Young gains only a few wins while a more modern pitcher gains a lot. However, there is no question that that more modern pitcher did less than Young did to earn those wins, since he only pitched 7 innings (for argument’s sake) per game.
    My thinking is that small sample sizes are nice, but over time, the better pitchers tend to rise to the top and get their wins. And if that is not enough, there is a whole raft of newer statistics that can balance out the difference.

    • nevyn49 says:

      Even if we used the pitcher win Joe is talking about, he would not be using it to compare and evaluate players. As you say, there are better stats for that.

      Wins are a team stat. Assigning them to a pitcher is simply trivia. Joe is merely arguing that his way is trivia that at least makes a little sense.

  15. sanford943 says:

    Hard to believe that Spahn did not have a 30 win season. And now after looking at couple of seasons it is no wonder. He completed so many games that he was pretty much involved in the decision one way or another.

  16. Philip Christy says:

    How about dressing the mere fact that ASSIGNING A TEAM WIN TO A PITCHER MAKES NO SENSE, AS HE ONLY FIELDED ONE OF THE NINE POSITIONS, AND ONLY TOOK AT THE VERY MOST ONE-NINTHS OF THE PLATE APPEARANCES (of course most likely much fewer) FOR THE ENTIRE TEAM?!? If a team wins a ballgame, the TEAM won the damn ballgame, even if the pitcher helped account for 50% or more of the work. Assigning wins to a single guy in this team sport is just dumb to begin with.

    • Marc Schneider says:

      I don’t understand why people get so bent out of shape over pitcher wins. Yes, it’s not an accurate picture of how good a pitcher is. So what? But it’s fun. People act as if baseball is supposed to be some sort of job that requires strict adherence to correct analytical procedures. Pitcher wins are a part of the game’s history and that’s one of the things that I enjoy the most. I mean, it’s fun to talk about how to make pitcher wins more accurate, but then you come along an by SCREAMING IN CAPS you say how dare we talk about pitcher wins.

      That’s really my objection to sabermetrics dogma in general.

  17. the leader in wins-in-games-started vs wins is:
    name – year – w/l – nw/nl – difference in wins
    Mike Krukow 1987 5-6 18-10 13

    where w/l is the win/loss record and nw/nl is the “new” wins/loss record.

    other notable seasons,
    Dwight Gooden 1999 3-4 15-7 12
    Nolan Ryan 1980 11-10 22-13 11
    Pete Harnisch 1992 9-10 20-14 11
    Bert Blyleven 1979 12-5 23-14 11

  18. BIP says:

    Um, why not just use WPA?

    • J Hench says:

      I’m pretty sure Tango has thoroughly debunked WPA as an effective measure for pitchers. I don’t have the link handy, but there was a 100+ post series of threads on it at the Book blog a couple of years ago.

      Basically, the argument is that WPA treats a starting pitcher differently based on when in the game the team scored its runs. Two pitchers can pitch identically well (or poorly), but if pitcher A’s team scored 4 runs in the top of the first, then his WPA will be lower than someone who’s team scores it’s runs in the 7th. That’s the gist of it anyway.

      • J Hench says:

        I should have said for starting pitchers. I personally like it a lot for relievers, especially in conjunction with Leverage Index. (not that you should base anything off what I personally like or don’t).

      • dfj79 says:

        Right, but if we’re just talking WPA vs. W-L (whether the official definition or Joe’s), WPA is still the much better stat for evaluating starting pitchers, even for all its limitations.

  19. Pak says:

    I find this intriguing. I’m usually old fashioned, I like old school stats such as pitcher’s win’s and RBIs, but I think this stat mirrors the win as it was traditionally viewed. I looked up Cy Young after reading an earlier comment, and it looks like this stat could cost Cy some wins, certainly some decisions. The same would be true for Walter Johnson. I wonder what their career records would look like with this definition? Were they “deserving” of their relief wins and losses? It would change the story of Johnson’s World Series heroics, but not diminish them, just as Madison Bumgarner demonstrated last year.

  20. rabidtiger says:

    Why not use the Pythagorean formula based on the the number of runs scored by the pitcher’s team while the pitcher is in the game versus the number of runs scored by the opposing team–again while the pitcher is in the game–and create a win probability number calculated to three digits, like batting average, OPS, FIP? Over the course of a season, someone like Koufax would be rewarded for scoreless innings and a lucky pitcher with junk wins would be shown in his hollowness.

  21. bake mcbride says:

    like the overall point Joe but always crediting starters with wins and losses is just as dumb as the current system.
    and the comparison to goalies and QBs doesn’t work either—they at least PLAY the whole game, pitchers don’t.
    I still think Tango’s old formula works the best…. don’t recall it exactly but it was from last year and made a lot of sense.

    • Paul Zummo says:

      If I’m not mistaken, the win (or loss) goes to the starting qb regardless of how long he plays. There was a stat about Kirk Cousins never having won a game even though he had come into the second drive of the second game of this past season and led the Redskins to a win over the Titans, but RGIII technically got the win because he started. So the NFL uses the system that Joe is proposing here.

  22. Michael says:

    I didn’t go back into the stats, but it surprises me that Bob Gibson never had a 30 win season by these metrics. If memory serves he had a lot of complete game losses or no-decisions in his career. How many of his starts did the Cardinals win in 1968 and 1970?

  23. Jay says:

    I think what is needed is to break this down into theee parts

    Number of games started
    Number of quality starts
    Number of games the team won

    This, though a middling compromise, gets to the heart of the ‘deserve’ argument. Looking at the number of quality starts gives a base calculation of how well the pitcher put his team in position to win. The weird statistics like 2/3 inning pitched and your team wins. It also eliminates games where the starting pitcher gives up a boat load of runs, but the opposing pitcher gives up a boat load of runs plus one more.

    Is this a great compromise, no, but what it does is remove the gray area, once we start talking about facts, then we take the emotion of the argument and just talk about the numbers.

    Remember poor Harvey Haddix threw 12 perfect innings and lost.

  24. heres some fun data – official wins, official wins earned as starter, team wins as starter (Poz wins) and the difference between official wins as starter and team wins as starter (Poz wins) – 1950-2012

  25. using the data I linked to above, I find the 3 biggest winners in this (at least 100 wins as starters) are

    name | record (official (in starts)) | record (Poz wins)
    Jim Slaton | 127-146 (0.465) | 180-179 (0.501)
    Ralph Terry |101-89 (0.532) | 143-110 (0.565)
    Mike Flanagan | 165-133 (0.554) | 236-168 (0.584)

    the biggest losers,
    name | record (official (in starts)) | record (Poz wins)
    Cliff Lee | 125-78 (0.616) | 158-122 (0.564)
    Roy Halladay | 196-98 (0.667) | 231-146 (0.613)
    Pedro Martinez | 208-97 (0.682) | 253-157 (0.617)

    that is, in games that Pedro started and he was given a non-decision, his teams were 45-60.

    the rest of the bottom-10 are all-start to hall-of-fame caliber:
    Dwight Gooden
    CC Sabathia
    Bob Forsch
    J. Weaver
    C. Carpenter
    J. Rijo

    this means something (worst records in non-decisions are generally high-caliber pitchers), but I’m not quite sure what.

  26. 300 game winners (accumulated since 1950) in Poz’s system,
    name | record
    Clemens_Roger | 433-274
    Maddux_Greg | 431-309
    Sutton_Don | 418-338
    Ryan_Nolan | 405-368
    Carlton_Steve | 404-305
    Glavine_Tom | 391-291
    John_Tommy | 387-313
    Niekro_Phil | 378-332
    Johnson_Randy | 371-232
    Seaver_Tom | 369-278
    Blyleven_Bert | 364-321
    Perry_Gaylord | 361-329
    Kaat_Jim | 347-278
    Moyer_Jamie | 343-295
    Jenkins_Fergie | 338-256
    Mussina_Mike | 326-210
    Palmer_Jim | 325-196
    Spahn_Warren | 311-199
    Tanana_Frank | 308-308
    Martinez_Dennis | 304-258
    Morris_Jack | 302-225
    Pettitte_Andy | 301-190

    • tangotiger says:

      Can you do the same, but give the W to the pitcher with the most outs on winning team, and the L to the pitcher with the most runs allowed on losing team.

  27. you know….we keep track of the wins the team gets when X pitcher starts. it’s in the team won loss record.

    what we need to do is not give a shit about individual awarding of team wins.

    the rest of the stats for pitching….particularly the newer stats…give us a better picture about which pitchers are most effective. we can even tell who’s pitching in the highest leverage situations.

    the guy who throws the most, most effective innings….is the best pitcher in a league, in a given year.

    we can already determine that…or come close.

    just ignore the wins, as far as individual ‘deserves’ go…

  28. jim says:

    pitcher wins is a misleading stat, but many stats are. the awareness that it’s misleading and needs context is victory enough. besides, as noted previously, the best pitchers tend to stack up the most wins.

    • Marc Schneider says:

      I agree with you. Pitcher wins, as you say, are misleading out of context, but that doesn’t mean it’s a totally worthless stat. If you win 300+ games, that must say something about the quality of your pitching. It doesn’t mean you were necessarily better than someone that won fewer games, but it means something. I don’t see any reason to abandon the stat although a lot of people seem infuriated by the very notion. As you say, the best pitchers tend to compile the most wins although there are exceptions, at least on a short-term basis, as when Felix Hernandez only won 13 games when he won the Cy Young or, on the other end, when Bob Welch won 27 for a great Oakland team. Lousy pitchers don’t win 200 games. And the pitcher is the single most important component of who wins or loses a game. The problem is when you have these clowns like Joe Morgan talking about a particular pitcher “knowing how to win” because he has more wins than another guy who is actually better.

      Perhaps if you were inventing baseball today, you would not include pitcher wins as a stat, but it’s not a blank slate and this is a game where history is important. It means something that Denny McLain won 31 games in 1968, at least in terms of the history of the game.

  29. jpg says:

    If the true goal is reward the pitcher who gave his team the best chance at winning, then I think wins and losses should be awarded with no regard to the final score. We should be awarding a win to the winner of the pitching matchup not the guy whose team happened to have won that day using Game Score.

    Pitcher A – 7 IP, 4 H, 1 BB, 0 R, 0 ER, 12 K
    Pitcher B – 4 IP, 11H, 2 BB, 5 R, 5 ER, 3 K

    Pitcher A’s bullpen mates completely melt down and his team loses 6-5.

    Under the current setup, some relief pitcher would get the win. Under Joe’s proposal, Pitcher B gets the win despite getting hammered because his team won. I don’t think either one makes much sense.

    Under my proposal Pitcher A gets the win for out pitching his oppent with a Game Score of 91. Pitcher B gets the loss with a game score of 18.

  30. 85kc says:

    I think Dean Chance got the win in an All Star game without throwing a pitch, Came in scored tied and picked a ruiner off for the third out. His team then scored and won.

  31. justin234124 says:

    In terms of counting an individual’s wins versus team wins, when October rolls around, the team wins determine seeding. No life-or-death decisions are based on whether pitcher wins are counted or how they’re counted; they’ve just been around for so long they have come to be shorthand for whether a pitcher was decent or not. The Hall of Fame, bullpens, rotations, trades, postseason awards, and contracts are no longer based on wins (although Ruben Amaro might be clueless enough to do it still). Statistics better suited for determining the value of the pitcher guide those conversations.

    In “Baseball” Dan Okrent talks about a .300 Hitter being a label, not a statistic, and I tend to agree with him. To hit .300 for a career, even with it being a statistic with its own issues, means something. Chumps don’t hit .300. Lots of really great hitters don’t get there, but .300 over a lifetime is a big deal and a way to quickly say, “he hit .300 so I know he was at least decent.” 20-game winner. 300-game winner. 3,000 hits. 500 home runs. Those are insanely difficult milestones to reach, which is why they aren’t reached often, and remain excellent shorthand to utilize before diving into more-descriptive statistics and/or discussing the era in which the numbers were achieved. The casual fan, the attend-one-game-in-a-Presidential-Administration fan, knows those labels as do all the people doing great work coming up with different ways of seeing things to tell the more-descriptive story. Not many pitchers vulture their way to 20 wins in a season or 300 in a career. Hitting those numbers are just a way to thumbs up or thumbs down at first glance.

    Based on the existing system, Joe’s system, or some other way, I want to count wins and have a simple benchmark to help me determine, “he is good enough to talk about.” If pitcher wins as a stat is worthless to you, it’s cool, simply disregard, but a pitcher’s won-loss record will never disappear from his stat sheet. (And, yes, at different times in my life I said there would “never” be a night game and “never” be a video board at Wrigley so I understand the word “never” itself is as worthless as pitcher wins.)

  32. Marc Schneider says:

    When I’m talking baseball with a less intense fan, I sometimes get frustrated at the level of conversation, but then I realize it’s all about enjoying the game. If people want to talk about a guy being “clutch” or “knowing how to win” or being a great RBI man, so what? Same thing with pitcher wins. This isn’t about the science behind global warming.

  33. John Autin says:

    Thought I would like this idea. Wanted to like it. But when I ran the numbers, i hate it.

    Starting pitcher stats for this year, through Sept. 2:
    — All Team Wins: 2.49 ERA, 6.34 IP per start (1,990 games)
    — Individual Wins: 1.89 ERA, 6.69 IP per start (1,376 games)
    Subtracting the latter set from the former, we get:
    — Team Wins/No Decision: 4.09 ERA, 5.56 IP per start (614 games)

    Individual wins are hideously polluted. But any good captured by the team win concept is more than offset by this stark fact: Collective SP ERA in team wins/no decisions is virtually the same as the average in ALL starts, while averaging about one out less per game. Including all those games in a pool of “starter wins” would make the SP win represent a much worse average performance than it does now. How can that possibly help make the “win” more relevant?

  34. John Autin says:

    This debate tends to center on good pitchers, and guys who “deserved” a lot more wins than they logged. Joe’s examples dealt only with team wins/no decisions, sort of implying that their actual wins were earned. The dark side of the matter is those who don’t even deserve many of the wins they got, and sure as shootin’ don’t deserve the boost that Joe’s idea would give. So let’s talk about Drew Hutchison.

    Hutchison is having one of the luckiest seasons in baseball history. His record is 13-2, despite a 4.87 ERA and 5.52 RA/9. He has just 9 quality starts out of 25; hell, CC Sabathia’s done better. But his run support is at an epic 7.29 R/G, with 6 or more in 19 of 25 (2nd-highest percentage in B-R’s searchable database), and at least 3 runs in all 25 starts (which I believe is unique in the modern era).

    His actual 13-2 record is clearly and horribly out of line with his performance. It’s not as though he’s pitched in a tough environment; Rogers Centre has a neutral park factor in the last few years, and has played to the pitcher’s advantage this year. (Don’t ask me, I just read the numbers.) He hasn’t pitched in Coors. And he hasn’t faced the runaway best offensive team, his own.

    But I would rather Hutchison be gauged by 13-2 than by 18-7, Toronto’s record when he starts. That’s just one team win off the MLB lead. He’s tossed just 139 innings, which ranks #80.

    His 10 no-decisions have a 9.80 ERA, the 2nd-worst of anyone with at least 9 no-decisions in the searchable era, and certainly the worst if put in league context. And the 5 that Toronto won were even worse.

    If we count team wins, Hutchison would be credited for 5 games in which he stunk every time, totaling 29 ER in 19.2 IP, a 13.27 ERA. The Jays averaged 10 runs to win those 5 games. I’m certain that a teammate more deserved every time.

    At least with 13-2, no one will talk him up for the Cy Young Award. At 18-7, I worry. So, no, thanks. I don’t think the win can be saved without adding so many conditions that it loses the quality of simplicity, which is really all it has going.

Leave a Reply

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