Yes, wins really are overrated!
Posted by Andy on June 16, 2010
Want to see a prime example of how most folks don't realize that Wins is one of the most overrated stats for pitchers?
Check out this article from ESPN.com by Enrique Rojas. It's about Pedro Martinez's stats from last year for the Phillies. Here's a quote:
"Martinez, 38, finished the 2009 season with a 5-1 record and a 3.63 ERA in nine starts for Philadelphia. The three-time Cy Young Award winner did not have the same success in the playoffs; he was 0-2 with a 3.71 ERA, including two starts against the New York Yankees in the World Series."
So, let's see. Pedro's ERA went up 0.08 runs against two of the best teams in baseball. But he didn't have the same success? I guess that's based on his record going from 5-1 to 0-2, even though his record has little to do with his performance.
Let's see. In Game 2 of the NLCS against the Dodgers he pitched 7 innings, gave up no runs, and got a no-decision. Yeah, just terrible.
In his first World Series start against the Yankees he gave up 3 runs in 6 innings, along with 2 walks and 8 strikeouts. Yeah, those quality starts are brutal. Pedro got the loss in that game.
In Game 6 he gave up 4 runs on just 3 hits (including a Hideki Matsui homer) and 2 walks. He got the loss again.
So he throws 7 shutout innings, another quality start, pitches to a cumulative ERA essentially identical to what he did in the regular season and...?
I realize I'm making a big deal out of just one phrase from Rojas' article. My point is just that it would seem that he picked that phrase based on wins alone without looking at the numbers in a little more detail, and that going by wins alone is a mistake.
June 17th, 2010 at 11:44 am
Another thing to consider is this: If the better pitcher is the one with the better W/L record, then by definition it should follow that the better pitching team is the one with the better W/L record. Yet, I don't think that anybody would argue that the 1999 Indians were a great pitching team, or that they were a better pitching team than the 2003 A's, or that they compared with any of the Braves teams from that era, teams with similar records. When we evaluate these teams, we see the 1999 Indians, correctly, as a great hitting team with enough pitching to win, and the 2003 A's as a great pitching team with enough hitting to win. People have no difficulty looking at a team like the A's of the last few years as a pretty good pitching team that hasn't been successful due to a lack of hitting, or that the Rangers of the A-Rod years could really hit, but lacked quality pitching.
Should we consider the pitching equal for all 92-70 teams, regardless of the hitting, or is it better to accept that, like with teams, an average pitcher on a team with great hitting can have the same record as a great pitcher on a team with average hitting without being deemed equally successful at his job?
June 17th, 2010 at 11:47 am
Yup, there are neutralized stats. There has been some debate (paging Johnny Twisto) about the accuracy of the calculation that B-R uses, but it's not a bad start.
On any pitcher's page, click on the 'more stats' next to his main pitching stats, then scroll down until you see the 'neutralized stats' section.
For example, here is a link to Andy Pettitte's neutralized stats.
June 17th, 2010 at 11:47 am
The best pitchers give up the least runs, wins do not correlate well with least runs given up, therefore wins are not a good way to rank the best pitchers.
June 17th, 2010 at 12:11 pm
@89
I've always believed that if a starter goes 6 complete innings, regardless of what happens after he leaves, it's HIS game, and thusly should be saddled with the win or the loss. That way we don't have to have this debate because than we can all just agree that wins don't tell us much.
Seriously though...
Wins, in a vacuume, are a nice quick and dirty stat. Baseball is not played in a vacuume and we should be assigning percentage of wins to all players that got into a given game (isn't this WAR?). We can rebuild the win. We (well, you guys) have the technology.
June 17th, 2010 at 12:16 pm
This is WAR all right.
June 17th, 2010 at 12:18 pm
This is one long semantics argument. Angel (Side A) is arguing that the word "performance" is defined as the objective W/L/ND. Most everyone else (Side B) is defining the word "performance" as a subjective statistical analysis. Meanwhile Side A is arguing that the stats don't indicate good performance and Side B is arguing that a win doesn't indicate good performance. So everyone is right because they're analyzing the other side's arguments using their own base definition. I declare this thread a ND.
(for the record, I'm in the Side B camp - I define performance as accumulated stats)
June 17th, 2010 at 12:46 pm
WilsonC, an obvious but excellent point. I'd love to hear the studio talking heads fumble around with that one.
June 17th, 2010 at 1:17 pm
Okay, one more thing...
All the analysis I see about average run support per game and the like are missing the point. The players do not play in a simulator. They know the score.
Example: Pitcher A takes a 4-0 lead into the 9th gives up 4 runs in the inning and the team eventually loses 5-4. He had control of the ball after it left his hand, he had the win in his grasp. Pitcher B allows 5 runs in the first, then pitches 7 shutout innings and his team comes back in the 9th to win 6-5. Pitcher B pitched better than pitcher A even if all the stats are identical. The run support is irrelevant. Pitcher B deserved the win, pitcher A deserved the loss (but gets a no-decision) because he blew his game. I understand that this website is all about statistical analysis, I guess I'm the Cito Gaston.
And a question? If a pitcher has a career record of 125-25, could you tell whether or not he was a good pitcher? Okay, another pitcher has a career ERA of 3.00. Can you tell me how he was in his era without analyzing the average ERA for his era, his run support...etc?
June 17th, 2010 at 2:03 pm
@108
Nice try, but neither is a definitive yes or no.
That's like saying:
I have an even temperment, does that make me a nice guy?
I have alot of patience, does that make me a nice guy?
Before you answer...
There have been plenty of documented seriel killers that are both patient and maintain an even temperment.
June 17th, 2010 at 2:28 pm
WINS IS THE ONLY THING THAT MATTERS, COLLECTIVELY. WHATELSE?
June 17th, 2010 at 2:36 pm
Angel, do you think Jack Morris was a guy who "knew how to win"?
Because people that have actually broken down his career start by start have found no evidence of that. He didn't pitch any better in tight games or big situations, and he had no special ability to preserve leads once he had them. He won a lot of games because he was a good pitcher for teams that scored a lot of runs. That's it. I'm just preemptively throwing Morris out there, because he's the guy who is inevitably identified as someone with a special "winning" skill, and I think you mentioned him earlier. But he didn't have some special ability to only give up runs when it didn't matter, artificially inflating his ERA. One World Series game ain't proof of anything.
Do pitchers adjust their approach based on the situation? Of course. But pitchers cannot see the future, don't know how many runs their team is going to score, and don't let up when they have a big lead. You're not going to find pitchers who consistently shut teams down when they have to, and allow runs when it doesn't matter. For the most part, a pitcher's W-L record will be dependent on his runs allowed and his run support. There might be a few guys who over a long career out-perform or under-perform those expectations. There is some evidence Bert Blyleven won fewer games than should be expected. But the effects are minimal. If it is a "skill," like clutch hitting, it is exceedingly difficult to identify and the results vary greatly from year to year.
Can you identify a pitcher under age 30 who you believe wins more games than one would expect from his ERA and run support? And do you expect he will continue to do so over the next five years?
June 17th, 2010 at 2:56 pm
@108: "Pitcher B deserved the win, pitcher A deserved the loss (but gets a no-decision) because he blew his game"
Playing devil's advocate for a minute, your premise thus far has been that a pitcher cannot deserve to win a game and then subsequently lose it because by definition a loss is a failure on his part. The opposite then must be true because if one pitcher deserves the win and gets the win the other pitcher invariably deserved to lose and got the loss. Here, you claim the possibility of deserving to lose and ending up with a ND. In this case he had a L performance but ended up with a ND, therefore his W/L record isn't a fair representation of his performance.
I know that's a small little thing to pick at, but it leads me to my next question: we know a pitcher can deserve to win and deserve to lose (whatever your definition of 'deserve' may be), but can a pitcher deserve to get a ND?
June 17th, 2010 at 3:07 pm
@108
In that particular example, I'd take Pitcher A over Pitcher B easily.
The stats don't tell the whole story in either case. For Pitcher B, he struggled early and then pitched great for 7 innings, and deserves credit for settling down after the first inning, but that doesn't erase the fact that he put his team in a position where they needed to battle back from a massive deficit to win. No matter how well he pitches after the first inning, the only way he doesn't lose that game is if his offense does a superb job, which in this example they do. Suppose instead of someone hitting a ninth inning HR to give his team the lead, the opposing center fielder makes a great catch to rob him for the third out and he loses. Pitcher B's out of the game icing his shoulder, yet somehow, whether or not the opposing defense makes that play dictates whether the pitcher deserves a win or a loss?
Pitcher A's the opposite. He pitched 8 strong innings, only to fade in the ninth. Did he just fall apart in the last inning? If so, why did the manager leave him in long enough to give up four runs. Did the opposition string together a series of bloop hits on quality pitches? Sometimes you just need to credit the other hitters. Was it his team's defense that fell apart? Unlike with pitcher B, who may have just gotten off to a rough start, a situation like Pitcher A is almost certainly a case where the responsibility for the loss is divided. If he's still pitching well in the ninth but either his defense abandons him or the other team just beats him, he still pitched well enough to deserve a win. If he's not pitching well, the manager deserves at least part of the blame for a slow hook.
Take another example:
Pitcher A, who plays for a weak hitting team, is facing Pitcher B, who plays for the best hitting team in the league. Pitcher A gives up one run in the second, on a fly ball that the left fielder misplayed, and then mows down this offensive juggernaut for seven innings of two-hit ball. Pitcher B's a little shaky from the beginning, walking a couple batters before getting out of a jam in the first. In the third inning, he walks the pitcher with two outs, hits the leadoff batter, and fails to cover first on a play that the first baseman fields, loading the bases. He gets out of the inning when his RF leaps at the wall to make a spectacular catch. He walks another couple batters in the fourth, and three more in the fifth to load the bases with one out. He then gives up a hard liner that the second baseman stares on a dive, then doubles up a runner to end the inning. The walks drove up his pitch count, driving him from the game, but his bullpen closes out the last four innings of the game cleanly for the win.
In this case, Pitcher A, by far, out-pitched Pitcher B, yet gets the loss. Pitcher A faced the better offense and dominated them, whereas Pitcher B got himself into all kinds of trouble, and had to be bailed out by his defense and his bullpen. Which pitcher is more deserving of the win here? Should the answer change if Pitcher A's team hits a two run double in the bottom of the ninth, with both starting pitchers already hitting the showers?
June 17th, 2010 at 9:49 pm
@Angel
What can a pitcher do (not his team, just the pitcher) to earn a win? Let up minimal runs. Why even look at wins when there is an arbitrary statistic (ERA) to fairly and effectively measure this ability?
@Johnny Twisto
I completely agree, "clutch" is not a skill in baseball.
June 17th, 2010 at 9:53 pm
@Angel
You can easily and extremely accurately assess how good the 3.00 ERA pitcher is; what more can you ask for than a prediction of how many runs a pitcher will let up?
June 17th, 2010 at 10:06 pm
Angel. Are you serious with your thoughts on Cain? Now i'm a die hard Giants fan who's seen every Cain start. Dude your talking outta your butt here. One game he threw a 2 hitter AND LOST because his team DID NOT SCORE. He (like Zito the year before) got the LOWEST RUN SUPPORT IN ALL THE MAJORS LAST YEAR. I believe it was a measley 2.1 a game.
"If he gets beat, he was not the best pitcher on the field that day. I know I'm in the minority on this one"
yes you are because there's no logic to your thoughts. None what so ever...NONE. Here's a thought look at Cain's league batting average against.
it tells you all you need to know. they don't get much hits off of him. he's losing game 2-1...1-0.
really I don't like to put people down but really I have to ask...how old are you? because your logic was 3rd grade.
June 17th, 2010 at 10:19 pm
"Pitcher B deserved the win, pitcher A deserved the loss (but gets a no-decision) because he blew his game"
dude really how old are you? seriously. What if pitcher B's team DIDN'T score them runs? Dude come on. No better yet what if those runs scored were errors caused by the fielders? so on and so on.
I'm open for debate and all but there actually HAS TO BE ONE. If youre just here to start stuff then say so but really....
a pitcher depends on the offense to win him games. His wins and loses will depend solely on the amount of runs he limits a team to and the amount of runs his team can score.
there's no special trick or "little things" a guy does to win games. that just down right 1909 baseball talk.
but if you don't like stats I feel you in that. But its fans like you who read money ball and missed the whole point. Money isn't the face of sabermetrics. To me the A's use sabermetrics to DEVAULE a players ability so they won't have to PAY THEM. they say stolen bases aren't good...the A's say this not sabermetrics. Why? Because a guy like scott podesidkintwsnos (yeah i can't spell his name and i don't care to) will ask for 10 million for doing nothing more than stealing bases.
you and joe morgan missed the whole point.
June 17th, 2010 at 10:20 pm
The Win isn't as valuable a stat as it used to be because starting pitchers don't pitch deep into games very often anymore. Traditionally, a pitcher who got a win would usually pitch a complete game, so yes, getting a lot of wins was an accurate measure of how good a starting pitcher was; it indicated that he pitched better then the other guy for the whole game. Nowadays most managers pull even their best pitchers after 6 or 7 innings and then use specialized bullpens, often bringing in 4 or 5 relievers to pitch 3 innings. When 5 pitchers are pitching in a close game, it is basically meaningless who gets the win. While it is fair to say that today a pitchers win/loss record is an overrated indicator of his effectiveness, people should understand that this is due to changes in the way teams are managed, NOT a flaw in the original reasoning which led statisticians to value pitching records.
June 17th, 2010 at 10:28 pm
Malcolm, that's an excellent point. As starters' IP drops and more pitchers are used in the game, fair and meaningful distribution of wins and losses decreases.
June 17th, 2010 at 11:15 pm
Gabe-114, just to clarify, I'm not saying that "clutch" (however one defines it) is not a skill. I'm just saying that it's very hard to identify, and many players identified as "clutch" will not appear as "clutch" in subsequent seasons. As Bill James described in his "Underestimating the Fog" essay, the variance in clutch performances exceeds the effect of clutch performance, so it's hard to identify.
June 18th, 2010 at 11:19 am
If you want to judge pitchers, why not choose the stats which are the best predictors of future performance? W/L is a horrible predictor relative to ERA, and ERA is not very good compared to ERA+. DICE, despite eliminating a significant component of some pitcher's games (the ability to generate weak balls in play) does about as well as ERA+ because it is fielding independent. So it seems to me that DICE and ERA+ are pretty good numbers to look at, while Wins or Win/Loss, without a lot of extra analysis beyond the raw numbers, simply don't tell you enough to distinguish between a decent pitcher who got great average run support, and a should be HOFer that got terrible run support (blyleven). Or rather, sometimes they give the wrong answer.
I really don't get the 300 win fetish. On the one hand, writers talk about how we'll never see 300 again, which is bs, in my opinion. It's getting rarer and it may die out one day if the game keeps changing in the direction it is now, but it's still quite possible for a guy with a 20+ year career on mostly good teams to win 300. OTOH, lots of pitchers from the 50s and 60s are in the hall with well under 300 wins, back when it was supposedly so much easier to do. But suddenly if you start your career in the 70s or later, 280ish isn't enough? What? I wouldn't vote for TJ or JK personally, but there are lesser pitchers with fewer wins in the hall from previous eras when it was supposedly easier to accumulate wins.
So if it's so hard now to get 300, why don't we cut some guys who almost made it some slack? and if it isn't, then stop talking about how Glavine will be the last one ever.
June 18th, 2010 at 12:43 pm
Michael, you keep mentioning DICE, and I don't recognize that abbreviation. Is it another name for DIPS ERA, or FIP?
Also, we don't always care about predicting future performance, sometimes we want to evaluate the performance that already happened. Still, though, Wins don't necessarily describe that very well either.
June 19th, 2010 at 6:34 pm
The fatal flaw in simply assigning the team's win or loss to the pitcher - even if both pitchers pitched the complete game - is that each pitcher faces a completely separate set of batters. Run support IS relevant. Defensive support IS relevant. If the pitcher is relieved, needless to say, THAT's relevant.
ERA was introduced for a very good reason. Its was apparent even long ago that pitchers on better teams would rack up wins even if they weren't that good while better pitchers on poorer teams might not win that often. ERA would rectify that and then you could see who the better pitchers really were regardless of how good or bad the teams they were on. How that could escape someone at this stage of statistical baseball evolution is rather odd.
June 29th, 2010 at 3:04 pm
If wins are over-valued, then where does that put the save? Less valuable a barometer for pitcher effectiveness than the Game-Winning RBI was for hitters, I'd say.
I think wins are way more valuable a measure of pitcher quality than the save.
A reliever should have to face the tying/losing run to qualify for "saving" a game.
Perhaps starters should get a "game pitched to a lead after seven", or some such certification for their efforts.
Or, maybe, we could accept that it all evens out -- more or less -- over the course of a long career. The truly great pitchers will win their share of the time. And the "just-good-enough-to-not win" won't.
Any proof in the pudding there? After all, there are plenty of pitchers who still managed to win a truckload of games for lousy teams.
Look at Walter Johnson. The Senators managed just 10 winning seasons during his 21-year career. And most of the time, he was the one that put them over the top.
Cream rises to the top.