Sale 6th in baseball in WAR

[southsider2k5]

It would be interesting to see if someone could come up with a way to isolate factors such as defense and run support while factoring in their own ERA and other pitcher numbers, to get some sort of an expected win total to see which pitchers "out-perform" their expected win totals.

[StrangeSox]

Because it's still the most important stat in the game.

 

Not for evaluating the effectiveness of a pitcher

[witesoxfan]

It would be interesting to see if someone could come up with a way to isolate factors such as defense and run support while factoring in their own ERA and other pitcher numbers, to get some sort of an expected win total to see which pitchers "out-perform" their expected win totals.

 

FIP, xFIP, fWAR

[chw42]

It would be interesting to see if someone could come up with a way to isolate factors such as defense and run support while factoring in their own ERA and other pitcher numbers, to get some sort of an expected win total to see which pitchers "out-perform" their expected win totals.

 

I think WAR pretty much tries to do just that.

[Eminor3rd]

It would be interesting to see if someone could come up with a way to isolate factors such as defense and run support while factoring in their own ERA and other pitcher numbers, to get some sort of an expected win total to see which pitchers "out-perform" their expected win totals.

 

As they've said above, this is essentially what FIP (Fielding Independent ERA) is. FIP is a stat scaled to ERA that ONLY factors in things that a pitcher controls completely, which are Ks, BBs, and HRs. The idea is that the rest of what factors into ERA have at least something to do with defense. And so one way that people commonly identify pitchers that are candidates to improve or get worse is by looking for large discrepancies between FIP and ERA. If a guy's FIP is lower than his ERA, he has been getting bad results from some combination of defense and batted ball placement luck, and he'll likely see better results if he continues doing what hes's doing, and vice versa.

 

This works out well most of the time, but there are some notable exceptions, typically whenever pitchers show abnormal consistency in homerun rates, either in an ability to suppress them or to give them up. Matt Cain is sort of the posterchild for a guy that always had a better ERA than FIP suggested he should. Zack Greinke is a guy who always had a worse ERA than his FIP suggested.

[southsider2k5]

As they've said above, this is essentially what FIP (Fielding Independent ERA) is. FIP is a stat scaled to ERA that ONLY factors in things that a pitcher controls completely, which are Ks, BBs, and HRs. The idea is that the rest of what factors into ERA have at least something to do with defense. And so one way that people commonly identify pitchers that are candidates to improve or get worse is by looking for large discrepancies between FIP and ERA. If a guy's FIP is lower than his ERA, he has been getting bad results from some combination of defense and batted ball placement luck, and he'll likely see better results if he continues doing what hes's doing, and vice versa.

 

This works out well most of the time, but there are some notable exceptions, typically whenever pitchers show abnormal consistency in homerun rates, either in an ability to suppress them or to give them up. Matt Cain is sort of the posterchild for a guy that always had a better ERA than FIP suggested he should. Zack Greinke is a guy who always had a worse ERA than his FIP suggested.

 

None of that actually translates to some sort of actual win total that can be compared to what a pitcher actually has.

[farmteam]

It's going to tell me that he was on a team with a good offense.

That's probably most of it. But there's something to be said for doing enough to win the game, regardless of everything else that goes on.

 

Like I said originally, it has far less value than other stats. I'm just saying that it is not 100% meaningless, as some have suggested.

[IlliniKrush]

That's probably most of it. But there's something to be said for doing enough to win the game, regardless of everything else that goes on.

 

Like I said originally, it has far less value than other stats. I'm just saying that it is not 100% meaningless, as some have suggested.

This is nonsense. A pitcher can't "do enough" when the team is on offense. He can pitch a shutout and still not get a W, or even a L if a run was given up due to an error. A pitcher also doesn't try to work with any cushion he has. If they score 6, he doesn't take the foot off the gas and give up 5 so he "does enough." He still tries to throw a shutout.

 

A pitcher can go 7 IP, 1 ER and lose. A pitcher can go 5 IP, 6 ER and win. That right there tells me it's a pretty meaningless statistic.

 

It's 99% meaningless, if you don't like 100%.

[Harry Chappas]

Wouldn't a team of all 0 WAR players hypothetically go 81-81. Everyone is completely average?

 

A WAR of 0 means you are average not that you suck.

 

The Median for this stat, not the average is 0. Thus if every player had a WAR of 0, they would be average in every aspect of the game.

 

Throw a player of 10 WAR on this team and now they are at 86 wins. This was the early 2000's Red Sox with Pedro Martinez.

[chw42]

Wouldn't a team of all 0 WAR players hypothetically go 81-81. Everyone is completely average?

 

A WAR of 0 means you are average not that you suck.

 

The Median for this stat, not the average is 0. Thus if every player had a WAR of 0, they would be average in every aspect of the game.

 

Throw a player of 10 WAR on this team and now they are at 86 wins. This was the early 2000's Red Sox with Pedro Martinez.

 

A replacement player is not an average player.

[chw42]

None of that actually translates to some sort of actual win total that can be compared to what a pitcher actually has.

 

FIP is used to calculate WAR, which is a win total.

 

The calculation of pitching WAR is actually quite complicated and involves scaling the pitcher's FIP to his run environment.

Edited August 20, 2013 by chw42

[cabiness42]

Wouldn't a team of all 0 WAR players hypothetically go 81-81. Everyone is completely average?

 

A WAR of 0 means you are average not that you suck.

 

The Median for this stat, not the average is 0. Thus if every player had a WAR of 0, they would be average in every aspect of the game.

 

Throw a player of 10 WAR on this team and now they are at 86 wins. This was the early 2000's Red Sox with Pedro Martinez.

 

No, the 'R' in WAR stands for 'Replacement'. A replacement-value player is a player you could pick up off waivers at just about anytime. Somebody like Casper Wells or Tyler Greene, or most of the players on the Astros. A team full of replacement-value players would have a cumulative WAR of 0.0 and would win about 30% of their games. A roster with 25 guys all with a 1.3 WAR would win about 81 games, so I guess in that respect a 1.3 WAR player would be average.

Edited August 20, 2013 by HickoryHuskers

[southsider2k5]

FIP is used to calculate WAR, which is a win total.

 

The calculation of pitching WAR is actually quite complicated and involves scaling the pitcher's FIP to his run environment.

 

The highest pitcher total for WAR is 7.0 to Clayton Kershaw. So that stat is telling me that Kershaw should have 7 wins and no pitcher should have more than that at the around the 25 games started mark? The best pitcher in baseball should have won about 1 in 3.5 of his starts this year?

[witesoxfan]

The highest pitcher total for WAR is 7.0 to Clayton Kershaw. So that stat is telling me that Kershaw should have 7 wins and no pitcher should have more than that at the around the 25 games started mark? The best pitcher in baseball should have won about 1 in 3.5 of his starts this year?

 

Not quite. The Dodgers are 72-52. Theoretically speaking, if you replaced Kershaw with a replacement level player, they'd be 65-59, which is obviously an incredible disparity.

 

I think you're looking for an equivalent to W-L record. About all you can figure is that, if you assume a guy to win half and lose half (because we're dealing with a hypothetical league of replacement level players) with 5 ND's through 25 starts, every pitcher is 10-10. In this instance, Kershaw would be - again theoretically - 17-3.

 

If you want to consider it in that context, that's about the best comparison that can be made.

Edited August 20, 2013 by witesoxfan

[LittleHurt05]

So if a roster of 25 "replacement players" would be expected to win about 40 games, can one assume that each "replacement player" is worth about 1.6 wins to their team every full season?

[Eminor3rd]

Wouldn't a team of all 0 WAR players hypothetically go 81-81. Everyone is completely average?

 

A WAR of 0 means you are average not that you suck.

 

The Median for this stat, not the average is 0. Thus if every player had a WAR of 0, they would be average in every aspect of the game.

 

Throw a player of 10 WAR on this team and now they are at 86 wins. This was the early 2000's Red Sox with Pedro Martinez.

 

No, this is a common misconception. 0 WAR is replacement level, not average. An average player usually comes in around 2 WAR

 

EDIT: Didn't see this was already said.

Edited August 20, 2013 by Eminor3rd

[chw42]

So if a roster of 25 "replacement players" would be expected to win about 40 games, can one assume that each "replacement player" is worth about 1.6 wins to their team every full season?

 

Pretty much.

[Eminor3rd]

None of that actually translates to some sort of actual win total that can be compared to what a pitcher actually has.

 

As chw said, fWAR is a FIP based metric, and is thus essentially this. You'd have to add more "wins" based on what a replacement level pitcher would actually produce if you wanted to compare it directly to an actual W-L record. I'm guessing that no one has done this yet because no one thinks it's particularly useful to think about W-L records, though I could be wrong. WAR by itself serves the purpose of giving you a number to compare pitchers, but it removes the context of the performance of the rest of the team that exists in W-L.

[southsider2k5]

Not quite. The Dodgers are 72-52. Theoretically speaking, if you replaced Kershaw with a replacement level player, they'd be 65-59, which is obviously an incredible disparity.

 

I think you're looking for an equivalent to W-L record. About all you can figure is that, if you assume a guy to win half and lose half (because we're dealing with a hypothetical league of replacement level players) with 5 ND's through 25 starts, every pitcher is 10-10. In this instance, Kershaw would be - again theoretically - 17-3.

 

If you want to consider it in that context, that's about the best comparison that can be made.

 

Something that equates to where Kershaw should be. The idea being that if you factor in all of Kershaw's numbers, and then factor anything that isn't the pitchers fault (If he played behind really good offensive support, he should have more wins, if he played with a really bad defense behind him, he should have less wins). Once you combine all of those numbers you get a number that given his numbers, and what the team has done behind him, Kershaw should have say 15 wins (pulling a number of out air for arguments sake). In reality he has 12, so he has underperformed where he should be by 3 wins. Then you could also chart the guys who are overperforming their expected win totals for "clutchness" and guys who are underperforming their expected win totals for "chokers".

[Eminor3rd]

Something that equates to where Kershaw should be. The idea being that if you factor in all of Kershaw's numbers, and then factor anything that isn't the pitchers fault (If he played behind really good offensive support, he should have more wins, if he played with a really bad defense behind him, he should have less wins). Once you combine all of those numbers you get a number that given his numbers, and what the team has done behind him, Kershaw should have say 15 wins (pulling a number of out air for arguments sake). In reality he has 12, so he has underperformed where he should be by 3 wins. Then you could also chart the guys who are overperforming their expected win totals for "clutchness" and guys who are underperforming their expected win totals for "chokers".

 

But you can basically do this with FIP and ERA, and then you don't have to deal with run support at all. In both scenarios, you'd need to look further to find out how much of the disparity can be attributed to BABIP, good/bad defense, "choking," competition level, etc. If you turn it into expected w/L, you're just adding more noise to it by including more variables unrelated to pitching performance.

[southsider2k5]

But you can basically do this with FIP and ERA, and then you don't have to deal with run support at all. In both scenarios, you'd need to look further to find out how much of the disparity can be attributed to BABIP, good/bad defense, "choking," competition level, etc. If you turn it into expected w/L, you're just adding more noise to it by including more variables unrelated to pitching performance.

 

You are missing my point where. I want an expected win loss stat that actually does realize variables unrelated to pitching performance to see exactly how clutch a pitcher is. The idea is that if you took two pitchers with theoretically identical stats, defensive support, and offensive support over the course of a season, there would probably be a guy that won more often than the other one did, when in theory their win totals should be identical. It would actually give you a measuring stick for the "clutch" ideal. Some pitchers pitch over errors, while others go to s*** on the drop of a dime. Some pitchers pitch over poor run support while others seem to lose by one run no matter what their offense puts up.

[farmteam]

This is nonsense. A pitcher can't "do enough" when the team is on offense. He can pitch a shutout and still not get a W, or even a L if a run was given up due to an error. A pitcher also doesn't try to work with any cushion he has. If they score 6, he doesn't take the foot off the gas and give up 5 so he "does enough." He still tries to throw a shutout.

 

A pitcher can go 7 IP, 1 ER and lose. A pitcher can go 5 IP, 6 ER and win. That right there tells me it's a pretty meaningless statistic.

 

It's 99% meaningless, if you don't like 100%.

1st bolded: Everyone tries to throw a shutout. Some are better at holding onto leads than others, whether that means giving up 4 ER in 6 IP or 2 ER in 7IP.

 

2nd bolded: That's why I said over the course of a few years, not one year, and definitely not one game as you indicated.

 

Would I rather have the guy who consistently 2ER in 7IP instead of the guy with 4ER in 6IP? Well duh. But if the latter guy, over the course of a few seasons, consistently gets wins throwing like that, I'm going to give him a bit of credit for at the end of the day getting the W.

[StrangeSox]

1st bolded: Everyone tries to throw a shutout. Some are better at holding onto leads than others, whether that means giving up 4 ER in 6 IP or 2 ER in 7IP.

 

2nd bolded: That's why I said over the course of a few years, not one year, and definitely not one game as you indicated.

 

Would I rather have the guy who consistently 2ER in 7IP instead of the guy with 4ER in 6IP? Well duh. But if the latter guy, over the course of a few seasons, consistently gets wins throwing like that, I'm going to give him a bit of credit for at the end of the day getting the W.

 

Why? The offensive production of his team got him the "w", not his pitching.

[chw42]

1st bolded: Everyone tries to throw a shutout. Some are better at holding onto leads than others, whether that means giving up 4 ER in 6 IP or 2 ER in 7IP.

 

2nd bolded: That's why I said over the course of a few years, not one year, and definitely not one game as you indicated.

 

Would I rather have the guy who consistently 2ER in 7IP instead of the guy with 4ER in 6IP? Well duh. But if the latter guy, over the course of a few seasons, consistently gets wins throwing like that, I'm going to give him a bit of credit for at the end of the day getting the W.

 

That still doesn't make any sense. I mean if he was winning those games where he drove in 5 runs during every start, then sure. But that would mean he's a better hitter than pitcher. But the point remains. Run support should not decide a pitcher's value.

 

The entire point of advanced metrics is to take away dependent variables like run support and focus on the independent variables, i.e., things the pitcher can control. If you were to switch the teams your two example players pitched on, giving them opposite run support (where the guy who throws 7 IP and gives up 2 ER gets good run support and the guy who gives up 4 ER in 6 IP gets crappy run support), would you give the latter guy the benefit of the doubt when he loses 15 games?

 

I have an idea of what you are trying to say. That being able to hold a lead and pitch to the situation is valuable. But it's certainly not valuable enough to determine a pitcher's value when something like that is also not totally up to the pitcher.

Edited August 20, 2013 by chw42

[Eminor3rd]

You are missing my point where. I want an expected win loss stat that actually does realize variables unrelated to pitching performance to see exactly how clutch a pitcher is. The idea is that if you took two pitchers with theoretically identical stats, defensive support, and offensive support over the course of a season, there would probably be a guy that won more often than the other one did, when in theory their win totals should be identical. It would actually give you a measuring stick for the "clutch" ideal. Some pitchers pitch over errors, while others go to s*** on the drop of a dime. Some pitchers pitch over poor run support while others seem to lose by one run no matter what their offense puts up.

 

But you can just look at how pitchers perform with runners on or in high leverage situations. Why create a proxy for something when you can just look at the actual numbers?

 

I'm still not sure I get what you want. Is it RA9-WAR? That's a relatively new FanGraphs stat where it calculates WAR based on ERA instead of FIP: http://www.fangraphs.com/blogs/introducing...ndent-pitching/