i know this has been discussed many, many times but fundamentally major league baseball doesn't want to open that can of worms where a scorer is left to choose who gets credit for a win or a loss when these things are still so vital in salary discussions. pitchers will be griping nonstop, the noise will be ridiculous

the situation now is unsatisfying but the instances of where a pitcher gets truly screwed are limited enough from a cost/benefit perspective making a change to the current approach isn't justified. and by cost/benefit i am speaking in terms of a few pitchers getting some stray wins versus the chronic griping and complaining from pitchers who feel they were shortchanged in some manner.

and if you touch how folks get assigned wins don't you have to assign how losses get assigned? because if a team is losing 9-2 and ties it up and then in the 14th a reliever gives up a run isn't the guy who got the team in a 9-2 hole really responsible for the loss? folks may think that is a crackpot argument but pitchers will bring this up if baseball goes down the path of allowing official scorers latitude

and assigning wins would put a lot of stress on official scorers. this is way beyond errors in the infield stuff.

don't like the current approach. but boy, the alternative looks to be a sticky wicket

The simple alternative would be to not assign wins and losses to individuals. Honestly, I don't think it would ever have been done if baseball started out using 4-5 pitchers per contest. It's a relic of the complete game era.

Well Harvs, I propose a simple change (or 2) would restore some balance to the W-L record of SPs.

First, altho the rules make for silly assignments of "W"s and "L"s to relievers, I don't think anyone cares. Who tracks a bulpen guys' W-L record anyway? So I put that problem into the category of annoyance, not critical.

The real problem is starting pitchers, who often get no decisions based on either leaving in a tie game, or leavng when ahead/behind and the bullpen alows a lead change that negates their efforts. Or, the SP who only goes 4 IP and is banned from getting a decision.

Easily solution #1: change the min IP rqmt from 5 to 4. The notion that a SP has to go 5 IPs to get a W is a vestige of a very different game. Today you could have an SP go 4.1 and then five relievers combine for the rest of the game; why insist on giving someon eelse the W if the team wins? Yeah, Bob Gibson et al would whine, but it's the right thing to do in a time when it's tough to even win 20 games when you only get a max of 34 starts per year.

A little harder solution #2: Change the rules so that if an SP (who goes at least 4 or 5 IP) leaves the game with the team ahead (plus any runners he leaves on don't score to lose the lead) and the team eventually wins, the SP gets the win. Conversely, if the team is losing when he leaves and the team loses, the SP gets the loss; maybe make an exception if a reliever gives up more runs than the SP.

I still enjoy the fun little thought experiments, like who will be the next guy to get to 300 wins, even while knowing that the stat is overrated. I'm against the idea of doing away with it completely.

What would records look like if we assigned wins and losses based on Win Probability Added?

Easily solution #1: change the min IP rqmt from 5 to 4. The notion that a SP has to go 5 IPs to get a W is a vestige of a very different game.

I always assumed it was at 5 IP because it would be an official game at that point.

One thing that you sometimes see calculated is the team's W-L in the games that a pitcher started (this is, naturally, easily available on B-Ref). It's of mild interest when compared to his traditionally-calculated W-L record, or to those of his teammates. The problem is that when you think about it, it's not all that interesting. It's still going to track the W-L of the team overall for the most part, with better pitchers doing somewhat better and worse doing worse. And there are lots better ways of finding out who pitched better or worse.

The problem with assigning wins and losses isn't that they are calculated wrong it's that people take them seriously. Wins and Losses are correctly calculated, they just don't have much use. If I'm looking for something to give me a game to game look at a pitcher I like to look at Quality Start percentage. It too has its flaws but it's much more reflective of what the pitcher accomplished.

The simple alternative would be to not assign wins and losses to individuals. Honestly, I don't think it would ever have been done if baseball started out using 4-5 pitchers per contest. It's a relic of the complete game era.

No, it never made any sense. Pitchers don't win or lose games, TEAMS do. The entire idea of giving a pitcher a "win" is batty.

arom

true. and maybe the game snaps back in the other direction. (boy i hope so)

changing rules like this should always be given careful deliberation so that they still hold even as the game transforms over time.

andere

very minor quibble in that at some point others around and very close to the game did think it made sense especially given that pitcher win totals quickly became a talking point in comparing pitchers.

other stats have come and gone and not received anywhere near this type of attention.

i understand what you are saying. just trying to provide a different perspective

and to be clear i am not 'in love' with pitcher wins. but nor do i consider them 'useless' as long as the reader understands the context

I still enjoy the fun little thought experiments, like who will be the next guy to get to 300 wins, even while knowing that the stat is overrated. I'm against the idea of doing away with it completely.

Concur. It's a flawed stat, but not a useless one. If pitcher wins/losses didn't exist the first thing we'd have to do is invent it.

One thing that's interesting, regardless how usage has changed, it's my understanding that decisions: IP has remained fairly consistent through the years. Whatever flaws exist with the stat (and, obviously, they are multiple), those weaknesses aren't specific to one era over another.

I don't see why relievers should be assigned wins at all. I also don't see why every game must have a winning pitcher. Not every game has a "save."

I don't want a change. Very generally, the longer and more effectively a player pitches in a game, the more wins he will receive. Usually, the pitchers with the highest win totals are the best starting pitchers in the game. Is it perfect? No. Should wins be viewed more as a team stat than as an indivual stat than it is now? Yes. But usually, pitchers who have pitched well receive a larger share of wins. It is a good enough measure of pitcher performance in the context of team performance.

What would records look like if we assigned wins and losses based on Win Probability Added?

This

What would records look like if we assigned wins and losses based on Win Probability Added?

Totally and absolutely useless, probably even more useless than current W/L records.

#5 Take a look at Support Neutral W/L (or whatever they've changed the name to)

Here's an illustration of the silliness of the 4 inning rule:

Mariners lead the A's 7-2 after 4 innings. Omar Olivares has pitched decently, has thrown 68 pitches, and just has to complete 1 more inning to qulaify for the win. Mariners pull him and let Randy Johnson pitch 2 innings. The starter who went 4 can't get a win, but the reliever who came into a game that was pretty much already over can get one. But then, who cares about Omar Olivares? Randy Johnson, 20 game winner. That's worth celebrating.

A few years ago in an APBA league I exploited that loophole to get a 30 win season out of a middle reliever.

The heyday of the usefulness of W/L record, it seems to me, was in the deadball era: lots of CG, low offense, pitchers way ahead of hitters. Take the 1908 Giants (whom I chose pretty much at whim):



There were probably a fair number of teams in that era where the major indicators for the top starters all correlated with W/L record. (Not that W% was ever free from illusion.) But with more shared work in the rotation, larger bullpens, fewer IP per start, fewer CG, the correlation dwindles (in part because you get smaller and smaller samples of run support for each starter). Now it's at a point where a .500 pitcher might well be significantly better than a .622 pitcher for the same team.

EDIT: And even then, of course, the weak Doc Crandall's .500 for a good team like the Giants is not helpfully comparable to, say, .429 for a good pitcher (Vive Lindaman) for the weak Boston Braves, obviously.

#18 Disagree partially. I think SNWL is the best measure of starting pitching value. It's not precisely WPA, more like WPA assuming league average offensive and bullpen support. (Mind you, those adjustments may address your specific complaints)

the biggest problem with the stat has been its name

#18 Disagree partially. I think SNWL is the best measure of starting pitching value. It's not precisely WPA, more like WPA assuming league average offensive and bullpen support. (Mind you, those adjustments may address your specific complaints)

I doubt it, unless somewhere mixed into those adjustments they remove Leverage Index. Even then, WPA/LI is pretty useless too, it's just not totally and completely useless like WPA.

the biggest problem with the stat has been its name

I'd say the biggest problem, as is often the case, is the people using it. Though the name is a close second.

How often do scorers use discretion in crediting wins? Jerry Blevins got a win for the A's on Friday by pitching a scoreless bottom of the 9th in the A's 14-9 win. The A's closer, Ryan Cook, gave up 3 hits and the lead in the bottom of the 8th before the A's stormed back with 6 runs in the top of the 9th.

The irony is that if Cook had pitched even worse and given up 2 more runs (making it 14-11 for the bottom of the 9th), Blevins would have been credited with a save and Cook likely would have gotten the win.

If pitcher wins/losses didn't exist the first thing we'd have to do is invent it.

Why?

I'd say the biggest problem, as is often the case, is the people using it. Though the name is a close second.

I was thinking that the name has steered the usage

I was thinking that the name has steered the usage

Well, people misuse BA and RBIs (and the advanced metrics, for that matter), so I think it would have happened regardless.

If pitcher wins/losses didn't exist the first thing we'd have to do is invent it.

If we were looking at modern baseball, with all the use of relievers we have today and an alphabet soup of stats to choose from, I don't think too many people would feel a void not having pitcher wins. Somebody might invent it, but I don't think it would get any more attention than catcher won lost record.

Here's an illustration of the silliness of the 4 inning rule:

The extreme silliness was a few weeks ago when Ned Yost left Bruce Chen to complete the fifth, even though he had looked shaky all game, and admitted after the game that he wanted Chen to qualify for the victory. Chen blew it and the Royals lost. But we're the stat nerds.

Honestly, I don't think it would ever have been done if baseball started out using 4-5 pitchers per contest. It's a relic of the complete game era.

Of course, in the very early days, when teams had just one pitcher, W-L didn't make any sense then either. The 1878 Boston Red Stockings went 41-19, so I can't imagine it had any significance to anyone that Tommy Bond went 40-19.

Which brings up the question: When did people start using pitchers' W-L records? Probably sometime after 1880.

Which brings up the question: When did people start using pitchers' W-L records? Probably sometime after 1880.

I'm sure. I agree with Dag that pitcher wins/losses would have emerged in some form because pitchers don't go everyday, and therefore someone would have wanted to see how the team performed with different pitchers on the mound. I think they were somewhat inevitable.

Probably sometime after 1880

By 1884, as Ed Achorn shows in his excellent book 59 in '84, Hoss Radbourn's W/L record was a big deal in the media. With the advent of full-overhand pitching, the one-man rotation was on the way out, so when Radbourn took on the role of one-man rotation that year people were suitably impressed (particularly since the team won so much).

Assigning team wins to individual players is a terrible idea for any sport.

Football does it with quarterbacks.
Baseball does it with pitchers.
Hockey does it with goalies.

Basketball doesn't do it at all.
Soccer doesn't seem to do it, either.

The difference seems to be that the first three sports have a position that (according to people) is more important than any other position on the field/ice. So the majority of the success/blame rests on their shoulders.

Those positions are also unique compared to others on the field/ice, and often have very specific tasks that aren't assigned to any other position. The only other positions like them that I would think might deserve "win/loss" credit for being "unique" would be baseball catcher, soccer goalie, and football kicker.

The interesting part for me is how the quarterback and the pitcher are only on the field for half the time. They get credit/blame if the other half (defense and offense, respectively) do/don't perform well.
At least the hockey goalie and soccer goalie are always "in play", even though they don't have (or, very little) impact on the offense.

I too liked the SNWL stats. (Or like. I don't look at them anymore because b-r doesn't have them.)

If pitcher wins/losses didn't exist the first thing we'd have to do is invent it.

Why?

1) While pitchers aren't the only one who have an impact on who wins or loses the game, the starting pitcher has the most impact on if his team wins/loses. He has FAR more impact. Even now he'll face 30-35 batters while a hitter will only come to the plate 4 times and have a few chances in the field. To put it another way, it sure as hell ain't random happenstance that guys like Lefty Grove and Pedro Martinez have among the best winning percentages of all-time.

2) Yes there's SNWL and all that but it's nice to have a nice, crisp, clean easy to understand counting stat. Not that SNWL is bad, it's great- but it's nice to have something basic and easy to process. (Besides, if we didn't have W-L, we sure wouldn't have SNWL).

If pitchers have to be credited with wins and losses, I don't understand why the starting pitcher isn't simply crediting with a win if the team wins or a loss if the team loses. Any win is already dependent on the support of the offense and defense, so what does it change to make it also dependent on the pitching of the relief corps? The fact that there is such a thing as a no decision makes less sense than crediting pitchers with wins and losses in the first place.

very minor quibble in that at some point others around and very close to the game did think it made sense especially given that pitcher win totals quickly became a talking point in comparing pitchers.

I suppose in some ways this comes back to AROM's point about it coming from a time when the complete game was normal, and that has not been true for a long time. But even then, it starts with a bad premise: that because the pitchers probably had more of an impact on the game result than any other positions (and that is probably over-estimated), he deserves credit for the win.

and to be clear i am not 'in love' with pitcher wins. but nor do i consider them 'useless' as long as the reader understands the context

I appreciate the POV here, but I am still convinced that the only reason pitcher "wins" is a useful framework for understanding is that it's what people are used to.

A lot of stats attempt to translate an individual player's actual contribution in terms of run-scoring into contribution in terms of wins: Win Shares, SNWL, WAR. This makes sense; ultimately, a fan would like to know how a player has contributed to what matters. These concepts do not simply look at the scoreboard and then point a finger at a pitcher who happened to be there and met a few arbitrary criteria.

A player's job is to score runs for his team with the bat and his legs, and to keep the other team from scoring with his arm and glove. It's really best to keep your primary focus on these things, and not to attempt to back estimate an individual's contribution from the team's result.

If pitchers have to be credited with wins and losses, I don't understand why the starting pitcher isn't simply crediting with a win if the team wins or a loss if the team loses. Any win is already dependent on the support of the offense and defense, so what does it change to make it also dependent on the pitching of the relief corps? The fact that there is such a thing as a no decision makes less sense than crediting pitchers with wins and losses in the first place.

I think Wins and Losses attempt (and fail) to only include things that happened in the game while the pitcher is still in the game, and if things happen afterward (or beforehand), the stats "recuse themselves" and thus the no-decision. I can understand the desire to differentiate between a SP leaving the game with his team winning and a SP leaving the game tied (but they later win). But the employment of this desire is so crappy that you may be right that it doesn't make any more sense than just giving them the result of the game.

I think it would be really nice if someone went back through the records and determined just how many times a starting pitcher had a win vultured by not going 5 innings in a game where his team never relinquished the lead. What's the actual frequency of this occurrence?

Looking at 2011, there were 15 games in which a pitcher went 4-4 2/3 IP and allowed two runs or fewer in a game that his team won. That doesn't guarantee that the pitcher had the lead when he was removed from the game; we'd have to look on a game-by-game basis to determine that.

There were five pitchers who threw 4-4 2/3 shutout innings in such a game. One of those--Ervin Santana--left the game in a scoreless tie.

15 games is roughly 0.3% of the total number played in 2011.

I appreciate the POV here, but I am still convinced that the only reason pitcher "wins" is a useful framework for understanding is that it's what people are used to.

I think wins provide some descriptive information about a pitcher's season. Consider 2 versions of Bob Welch's 1990 stat-line which are identical except that one does not include his Wins and Losses. Both would give you the same amount of information concerning how well he pitched that year, but the one with the 27-6 staring out at you would give you more overall context about what happened in those 35 games he started that year: You could surmise, given his low CG totals, that his bullpen was unbelievably effective at protecting his leads and that the team overall was probably excellent given that they won at least 27 of his starts. So there is some information there, and I think we shouldn't only be concerned with information that is relevant to the evaluation and projection of individual players.

You could surmise, given his low CG totals, that [Bob Welch 1990's] bullpen was unbelievably effective at protecting his leads and that the team overall was probably excellent given that they won at least 27 of his starts.

No one would "need to invent" an individual stat that relayed that information.

I wasn't trying to comment on the 'need to invent' discussion...

Who cares?

I think it would be really nice if someone went back through the records and determined just how many times a starting pitcher had a win vultured by not going 5 innings in a game where his team never relinquished the lead. What's the actual frequency of this occurrence?

Looking at 2011, there were 15 games in which a pitcher went 4-4 2/3 IP and allowed two runs or fewer in a game that his team won. That doesn't guarantee that the pitcher had the lead when he was removed from the game; we'd have to look on a game-by-game basis to determine that.

There were five pitchers who threw 4-4 2/3 shutout innings in such a game. One of those--Ervin Santana--left the game in a scoreless tie.

15 games is roughly 0.3% of the total number played in 2011.

This is exactly what I was wondering. Thanks! The 5-inning rule is silly but has so little effect that it's not worth worrying about.

#36 They still have them. The name's just been changed.

I think it would be really nice if someone went back through the records and determined just how many times a starting pitcher had a win vultured by not going 5 innings in a game where his team never relinquished the lead. What's the actual frequency of this occurrence?

Looking at 2011, there were 15 games in which a pitcher went 4-4 2/3 IP and allowed two runs or fewer in a game that his team won. That doesn't guarantee that the pitcher had the lead when he was removed from the game; we'd have to look on a game-by-game basis to determine that.

There were five pitchers who threw 4-4 2/3 shutout innings in such a game. One of those--Ervin Santana--left the game in a scoreless tie.

15 games is roughly 0.3% of the total number played in 2011.

In that game, Santana had more than a shutout, he had 7 strikeouts and a no-hitter. Josh Beckett also came out in before 5 innings were done, there was a long rain delay. Angels eventually won it in the 13th. I'll bet most of those cases with a pitcher removed while pitching well were rain delays, or elevated pitch counts. Vulturing wins this way, intentionally, is probably extremely rare, which is why I had to pull an example from 1997.

Here's an example of Alfredo Aceves vulturing a win from Sabathia. Another rain delay situation. In that case, C.C. deserved to not get the win. He held a 6-3 lead in the bottom 5th, 2 out, and was working so godawful slowly between pitches that it was hard to believe. Here he was, 1 out from an official game, rain pouring down. I don't have much respect for the idea of playing for personal stats, but here is a case where pitching quickly to get that inning in would be important for the team. They were likely to win it (which they did, by score of 14-3) if play resumed, but there was a possibility that the game would be called if it rained long enough, and finishing that inning would give the Yankees a W, while waiting it out like he did set up the possibility of a rainout and start from scratch.

AROM, that story reminds me of when I was at this game and was infuriated at Beckett for pitching slowly during the top of the 5th in a driving rain with a 9-0 lead. He actually called Varitek out to the mound (I think with 2 outs) during the inning when it was clear to everyone else that the end of the game was imminent.