Which WAR should I fight?

Déjà Bru · 04-21-2018, 12:13 PM

Well, not fight, anymore. More like, accept. WAR is here and has become so popular that I feel the need to wrap my head around it and actually use it! (Folks who know me, don't gasp in astonishment. Even dinosaurs plod ahead a few steps from time to time.

)

I am NOT going to ask OOTPD to divulge its calculations of WAR. Not only is it a well-guarded secret like the formula of Coca-Cola, but it would probably take pages of glassy-eyed technical explanations such as this one at Baseball Reference.

Instead, I would ask a simple question which I would hope someone in the know would answer. Please see this WAR Comparison Chart (which, by the way, I fully expect OOTPB to be another column therein someday).

If you, Mr. Person-In-The-Know, had to select which of the five alternative definitions listed there, which one would most closely approximate the definition of WAR used in OOTPB? Here is the top of the chart so that you can get an idea of what I am talking about:

rudel.dietrich · 04-21-2018, 12:24 PM

Whatever your choice just don't make it a two front WAR. Those rarely end well.

Déjà Bru · 04-21-2018, 12:31 PM

Quote:

Originally Posted by rudel.dietrich

Whatever your choice just don't make it a two front WAR. Those rarely end well.

Yes, and let's please keep this a civil WAR. No debates about the propriety of WAR anymore (such as I would have waded into on the negative side years ago). I'm just asking in which direction I should go to perhaps get an approximation of how WAR is waged in OOTPB. End of puns.

Airdrop01 · 04-21-2018, 12:40 PM

WAR...what is it good for? Absolutely nothin!

Couldn’t resist.

But, although WARDS (WAR Derangement Syndrome) is treatable, I implore you: Don’t volunteer to be infected.

Shoeless' Socks · 04-21-2018, 01:00 PM

fWAR, IMO.

Having said that, I believe ALL of the WARs overrate defensive abilities in their calculations. A player like Kevin Kiermayer is just not nearly as valuable as his WAR compares to other players. He's just not.

Déjà Bru · 04-21-2018, 01:11 PM

Quote:

Originally Posted by Shoeless' Socks

fWAR, IMO.

Shoeless, not to put you on the spot, but do you have the chops to make that call? I am not going to ask you to elaborate, though. Just do you have the necessary knowledge for an informed opinion?

Shoeless' Socks · 04-21-2018, 01:30 PM

Quote:

Originally Posted by Déjà Bru

Shoeless, not to put you on the spot, but do you have the chops to make that call? I am not going to ask you to elaborate, though. Just do you have the necessary knowledge for an informed opinion?

Am I a sabermetrician. No. Do I think Kevin Kiermayer belongs in the same discussion as other 6+ WAR players, HELL no.

ThePretender · 04-21-2018, 02:56 PM

fWAR is significantly better than rWAR because of the defensive numbers they use. For pitching I prefer FIP based vs runs allowed.

Also, Shoeless, Kiermaier is exactly as valuable as WAR suggests. Your issue is that you don't believe that Kiermaier is as good defensively as he is. He's unique, and in a class of his own, as the majority of defenders aren't as good as he is.

For the record... Kiermaier has never been a 6 WAR player. He might have been on pace for 6 WAR in one season, but I wouldn't call him a 6 WAR player. For his career he's averaged 4.4 WAR per 550 PA. So he's not a 6 WAR player. I'm not sure where you got that number from, either.

NoOne · 04-21-2018, 04:38 PM

anything that tries to weight hings and put them in relative terms is a good stat.

there's no magic bullet though.. war is not 100% comprehensive nor omniscient (any type). incorporate it into how you look at things, even let it change how you look at things a bit... but never depend on it... never say to your self this "5" war is better than this "4" war..

Matt Arnold · 04-21-2018, 04:51 PM

We're closest to fWAR, but obviously we don't use the exact same formulas. Basically batters you add some value of offense plus some value for defense. For pitching fip based WAR and runs allowed WAR have their uses - That's why we added them both in this year. Fip is more useful for true talent, rWAR more useful when looking back at what happened and for award voting.

drhay53 · 04-21-2018, 04:52 PM

rWAR is now a separate column for pitcher stats. So I'm guessing the plain 'WAR' is fWAR.

My hunch is that for batters it is a combination of RC + ZR + BSR (baserunning runs, which as far as I can tell from analyzing the output from OOTP compared with player ratings is based solely on stolen bases).

I haven't actually attempted to decompose the OOTP WAR estimates, but I use full-league season stats to estimate the effect of each player rating on a pseudo-war that I estimate myself. For pitchers, I use FIP- as my sole metric. For batters, I use ZR/10, bsr/10 (although in practice, I simply make baserunning ratings a flat percentage of the maximum-weight hitting rating; which is becoming power over contact in my league, due to homer-happy modern stats) and wRC+.

For wRC+, I estimate the effect of each batter rating on wRC+, then I rescale the wRC+ to be on a rough WAR scale from -2 to 6 WAR. Basically every season the worst batter above the median number of PA gets scaled to -20 runs in batting, and the best player gets scaled to 60 runs. This is just for the purpose of getting weights for each player rating.

So in the end, every player is evaluated from their ratings based on their predicted total WAR. Generally, the most important defensive positions get ~20% of their weight from defense.

Anyway, that got a bit rambly and off topic, sorry.

Déjà Bru · 04-21-2018, 06:03 PM

Thanks everybody, particularly Matt who functioned as the "Person-In-The-Know" that I sought for an "insider" answer. That is all I was looking for; a point in the right direction to get a grasp on the WAR definition closest to OOTPB's WAR.

So now I will try to wrap my head around fWAR. Yeah, right.

But at least I will have some idea of what's involved. There's one thing I now realize about WAR. If it is indeed scientifically, reliably, and consistently arrived at BY A SOURCE WHICH YOU TRUST AND WHICH YOU USE EXCLUSIVELY FOR THE PURPOSE (really important considerations, hence the emphasis), then when your source says Kiermaier is a 4.4 WAR player and this other guy is a 6 WAR player, then you can relate to that, rely on that, and use that with confidence.

I trust OOTPB to manage its WAR calculations scientifically, reliably, and consistently. Suits me.*

*EDIT: I should clarify this. It suits me but I still would not be walking around talking about WAR and quoting WAR figures in a vacuum. Basically, it's hard to use a figure that is so arcane, so complicated, so abstract, and so variable according to the source quoted. But if WAR is being calculated reliably and applied to players equitably within one source (like OOTPB), then I can see its value in that context. So with all due respect, an argument about whether Kiermaier is a 4.4 or 6 WAR player according to different sources is not very fruitful in my opinion. I am interested only in how Kiermaier's WAR is rated within OOTPB in relation to all other players in the game.

drhay53 · 04-21-2018, 08:12 PM

It should be noted that WAR is really still a measure of what happened
and not a measure of intrinsic skill. I.e., WAR still contains some elements of chance.

Even the best bayesian model I have for wRC+ in OOTP as a function of player ratings has a residual scatter of about ~9 runs. Some of that is scouting accuracy, and some of it is "baseball". But one way to look at it is that in 68% (1-sigma) of OOTP seasons, a player's WAR at the plate will be +- 1 from his 'true' skill. So if you have a league average (2 WAR) player, he's equally likely to have a 1 WAR or 3 WAR season (what I mean is that there's a 68% chance his WAR will be between 1 and 3). And it's not all that unlikely for him to even have a 0 WAR or 4 WAR season (2-sigma from true skill; i.e. there's a 95% chance his WAR will be between 0 and 4, with a true skill of 2. Ballpark numbers here.)

Like everything in baseball, WAR is a better representation of a player's true skill when you have several seasons of data informing your decision.

Claybor · 04-21-2018, 09:20 PM

Personally I can evaluate a player just fine based on his actual baseball stats and accomplishments rather then needing to make studies to create stats that are extremely subjective.WAR makes a pretty good comparative tool, but I don't see it as the number one most important number in the game as it seems to have become. To each their own.

drhay53 · 04-21-2018, 11:17 PM

WAR and other advanced stats are simply an attempt to take alot of different stats and determine what weights to combine them with to predict wins. There's really nothing all that complex about it, conceptually.

jaa36 · 04-21-2018, 11:18 PM

Quote:

Originally Posted by ThePretender

fWAR is significantly better than rWAR because of the defensive numbers they use. For pitching I prefer FIP based vs runs allowed.

Also, Shoeless, Kiermaier is exactly as valuable as WAR suggests. Your issue is that you don't believe that Kiermaier is as good defensively as he is. He's unique, and in a class of his own, as the majority of defenders aren't as good as he is.

For the record... Kiermaier has never been a 6 WAR player. He might have been on pace for 6 WAR in one season, but I wouldn't call him a 6 WAR player. For his career he's averaged 4.4 WAR per 550 PA. So he's not a 6 WAR player. I'm not sure where you got that number from, either.

Interestingly, Fangraphs changed some of their assessments of players' defensive performances from 2012-2016 recently based on changes in the way Ultimate Zone Rating (UZR) was calculated: https://www.fangraphs.com/blogs/inst...nd-uzr-update/

This had the effect of smoothing down some of the more "extreme" positive or negative UZRs. Kiermaier's "performance" in particular was downgraded by about one win in both 2014 and 2015. As of now, Kiermaier is listed with 21.3 WAR on Baseball Reference, and 14.1 WAR on Fangraphs in his career. The difference is essentially entirely made up by which defensive metric you use: UZR or Defensive Runs Saved (DRS).

I think it's fair to say there is plenty of uncertainty about how much an individual player contributes in an individual season on defense. We will probably never have a consensus about what the "best" way to rate defense is. That's OK. There are arguments about the best framework for the offensive inputs of WAR and pitching WAR as well, and even whether the end result should be expressed in wins at all: http://joeposnanski.com/more-on-war/

That's OK too. There's value in all the different formulations of WAR. Having several ways to describe a player's overall performance is a good thing.

endgame · 04-21-2018, 11:27 PM

These exchanges always seed education. I welcome them, particularly between individuals who not only seem well-versed in the area, but who also can demonstrate that knowledge with civility and not immediately draw lines and build fences. Commendable. As for me, the first thing I learn is always how much I have to learn.

drhay53 · 04-21-2018, 11:38 PM

Quote:

Originally Posted by jaa36

Interestingly, Fangraphs changed some of their assessments of players' defensive performances from 2012-2016 recently based on changes in the way Ultimate Zone Rating (UZR) was calculated: https://www.fangraphs.com/blogs/inst...nd-uzr-update/

This had the effect of smoothing down some of the more "extreme" positive or negative UZRs. Kiermaier's "performance" in particular was downgraded by about one win in both 2014 and 2015. As of now, Kiermaier is listed with 21.3 WAR on Baseball Reference, and 14.1 WAR on Fangraphs in his career. The difference is essentially entirely made up by which defensive metric you use: UZR or Defensive Runs Saved (DRS).

I think it's fair to say there is plenty of uncertainty about how much an individual player contributes in an individual season on defense. We will probably never have a consensus about what the "best" way to rate defense is. That's OK. There are arguments about the best framework for the offensive inputs of WAR and pitching WAR as well, and even whether the end result should be expressed in wins at all: http://joeposnanski.com/more-on-war/

That's OK too. There's value in all the different formulations of WAR. Having several ways to describe a player's overall performance is a good thing.

IMO, the effects of positioning will never be fully disentangled from scouting and management decisions; however, with the current ability to track player's movement, and likely improvements in the future, things like route efficiency, reaction time, actual range, etc. will be pretty measurable. So I think our ability to isolate a single player's defensive ability will improve significantly over the coming decade or two.

Shoeless' Socks · 04-21-2018, 11:49 PM

IMO at the major league level, my theory is that defense only matters at the marginal extremes. Unless a defender is elite, or atrocious, it doesn't matter much.

Nothing will ever convince me otherwise. The lower you go, the more defense matters. Coach pitch little league it is vital. At the ML level, it matters very little, IMO.

I do not believe that there is a player alive that can save as many runs with his glove that he can create with his bat. Not given 500+ ABs.

Déjà Bru · 04-24-2018, 09:37 AM

This, from Baseball Reference, is key - to me, that is. For others used to dealing with WAR for years now, it's probably basic but to me this was quite illuminating. I am assuming that the concept is the same in OOTPB even though I know and accept that OOTPB's calculations vary somewhat from Baseball Reference and fWAR.

04-21-2018, 12:13 PM	#1
Déjà Bru Hall Of Famer Join Date: Apr 2009 Location: Long Island Posts: 10,991	Which WAR should I fight? Well, not fight, anymore. More like, accept. WAR is here and has become so popular that I feel the need to wrap my head around it and actually use it! (Folks who know me, don't gasp in astonishment. Even dinosaurs plod ahead a few steps from time to time. ) I am NOT going to ask OOTPD to divulge its calculations of WAR. Not only is it a well-guarded secret like the formula of Coca-Cola, but it would probably take pages of glassy-eyed technical explanations such as this one at Baseball Reference. Instead, I would ask a simple question which I would hope someone in the know would answer. Please see this WAR Comparison Chart (which, by the way, I fully expect OOTPB to be another column therein someday). If you, Mr. Person-In-The-Know, had to select which of the five alternative definitions listed there, which one would most closely approximate the definition of WAR used in OOTPB? Here is the top of the chart so that you can get an idea of what I am talking about: Attached Images __________________ - Bru

04-21-2018, 12:40 PM	#4
Airdrop01 All Star Starter Join Date: May 2009 Location: Lenexa, KS / Wilson, WY Posts: 1,354	WAR...what is it good for? Absolutely nothin! Couldn’t resist. But, although WARDS (WAR Derangement Syndrome) is treatable, I implore you: Don’t volunteer to be infected. __________________

04-21-2018, 06:03 PM	#12
Déjà Bru Hall Of Famer Join Date: Apr 2009 Location: Long Island Posts: 10,991	Thanks everybody, particularly Matt who functioned as the "Person-In-The-Know" that I sought for an "insider" answer. That is all I was looking for; a point in the right direction to get a grasp on the WAR definition closest to OOTPB's WAR. So now I will try to wrap my head around fWAR. Yeah, right. But at least I will have some idea of what's involved. There's one thing I now realize about WAR. If it is indeed scientifically, reliably, and consistently arrived at BY A SOURCE WHICH YOU TRUST AND WHICH YOU USE EXCLUSIVELY FOR THE PURPOSE (really important considerations, hence the emphasis), then when your source says Kiermaier is a 4.4 WAR player and this other guy is a 6 WAR player, then you can relate to that, rely on that, and use that with confidence. I trust OOTPB to manage its WAR calculations scientifically, reliably, and consistently. Suits me.* *EDIT: I should clarify this. It suits me but I still would not be walking around talking about WAR and quoting WAR figures in a vacuum. Basically, it's hard to use a figure that is so arcane, so complicated, so abstract, and so variable according to the source quoted. But if WAR is being calculated reliably and applied to players equitably within one source (like OOTPB), then I can see its value in that context. So with all due respect, an argument about whether Kiermaier is a 4.4 or 6 WAR player according to different sources is not very fruitful in my opinion. I am interested only in how Kiermaier's WAR is rated within OOTPB in relation to all other players in the game. __________________ - Bru Last edited by Déjà Bru; 04-21-2018 at 06:25 PM.

04-21-2018, 08:12 PM	#13
drhay53 Major Leagues Join Date: Aug 2016 Posts: 355	It should be noted that WAR is really still a measure of what happened and not a measure of intrinsic skill. I.e., WAR still contains some elements of chance. Even the best bayesian model I have for wRC+ in OOTP as a function of player ratings has a residual scatter of about ~9 runs. Some of that is scouting accuracy, and some of it is "baseball". But one way to look at it is that in 68% (1-sigma) of OOTP seasons, a player's WAR at the plate will be +- 1 from his 'true' skill. So if you have a league average (2 WAR) player, he's equally likely to have a 1 WAR or 3 WAR season (what I mean is that there's a 68% chance his WAR will be between 1 and 3). And it's not all that unlikely for him to even have a 0 WAR or 4 WAR season (2-sigma from true skill; i.e. there's a 95% chance his WAR will be between 0 and 4, with a true skill of 2. Ballpark numbers here.) Like everything in baseball, WAR is a better representation of a player's true skill when you have several seasons of data informing your decision. Last edited by drhay53; 04-21-2018 at 08:14 PM.

04-21-2018, 09:20 PM	#14
Claybor All Star Starter Join Date: Apr 2002 Posts: 1,360 Infractions: 0/1 (1)	Personally I can evaluate a player just fine based on his actual baseball stats and accomplishments rather then needing to make studies to create stats that are extremely subjective.WAR makes a pretty good comparative tool, but I don't see it as the number one most important number in the game as it seems to have become. To each their own. Last edited by Claybor; 04-21-2018 at 09:23 PM.

04-21-2018, 12:24 PM	#2
rudel.dietrich Hall Of Famer Join Date: Jul 2004 Location: Zürich, Switzerland Posts: 8,608	Whatever your choice just don't make it a two front WAR. Those rarely end well.

04-21-2018, 01:00 PM	#5
Shoeless' Socks Minors (Double A) Join Date: Jul 2017 Posts: 162	fWAR, IMO. Having said that, I believe ALL of the WARs overrate defensive abilities in their calculations. A player like Kevin Kiermayer is just not nearly as valuable as his WAR compares to other players. He's just not.

04-21-2018, 02:56 PM	#8
ThePretender All Star Starter Join Date: Mar 2008 Posts: 1,321	fWAR is significantly better than rWAR because of the defensive numbers they use. For pitching I prefer FIP based vs runs allowed. Also, Shoeless, Kiermaier is exactly as valuable as WAR suggests. Your issue is that you don't believe that Kiermaier is as good defensively as he is. He's unique, and in a class of his own, as the majority of defenders aren't as good as he is. For the record... Kiermaier has never been a 6 WAR player. He might have been on pace for 6 WAR in one season, but I wouldn't call him a 6 WAR player. For his career he's averaged 4.4 WAR per 550 PA. So he's not a 6 WAR player. I'm not sure where you got that number from, either.

04-21-2018, 04:38 PM	#9
NoOne Hall Of Famer Join Date: Apr 2015 Posts: 7,228	anything that tries to weight hings and put them in relative terms is a good stat. there's no magic bullet though.. war is not 100% comprehensive nor omniscient (any type). incorporate it into how you look at things, even let it change how you look at things a bit... but never depend on it... never say to your self this "5" war is better than this "4" war..

04-21-2018, 04:51 PM	#10
Matt Arnold OOTP Developer Join Date: Jun 2009 Location: Here and there Posts: 15,538	We're closest to fWAR, but obviously we don't use the exact same formulas. Basically batters you add some value of offense plus some value for defense. For pitching fip based WAR and runs allowed WAR have their uses - That's why we added them both in this year. Fip is more useful for true talent, rWAR more useful when looking back at what happened and for award voting.

04-21-2018, 04:52 PM	#11
drhay53 Major Leagues Join Date: Aug 2016 Posts: 355	rWAR is now a separate column for pitcher stats. So I'm guessing the plain 'WAR' is fWAR. My hunch is that for batters it is a combination of RC + ZR + BSR (baserunning runs, which as far as I can tell from analyzing the output from OOTP compared with player ratings is based solely on stolen bases). I haven't actually attempted to decompose the OOTP WAR estimates, but I use full-league season stats to estimate the effect of each player rating on a pseudo-war that I estimate myself. For pitchers, I use FIP- as my sole metric. For batters, I use ZR/10, bsr/10 (although in practice, I simply make baserunning ratings a flat percentage of the maximum-weight hitting rating; which is becoming power over contact in my league, due to homer-happy modern stats) and wRC+. For wRC+, I estimate the effect of each batter rating on wRC+, then I rescale the wRC+ to be on a rough WAR scale from -2 to 6 WAR. Basically every season the worst batter above the median number of PA gets scaled to -20 runs in batting, and the best player gets scaled to 60 runs. This is just for the purpose of getting weights for each player rating. So in the end, every player is evaluated from their ratings based on their predicted total WAR. Generally, the most important defensive positions get ~20% of their weight from defense. Anyway, that got a bit rambly and off topic, sorry.

04-21-2018, 11:17 PM	#15
drhay53 Major Leagues Join Date: Aug 2016 Posts: 355	WAR and other advanced stats are simply an attempt to take alot of different stats and determine what weights to combine them with to predict wins. There's really nothing all that complex about it, conceptually.

04-21-2018, 11:27 PM	#17
endgame Hall Of Famer Join Date: Aug 2002 Posts: 16,842	These exchanges always seed education. I welcome them, particularly between individuals who not only seem well-versed in the area, but who also can demonstrate that knowledge with civility and not immediately draw lines and build fences. Commendable. As for me, the first thing I learn is always how much I have to learn. __________________ "Try again. Fail again. Fail better." -- Samuel Beckett _____________________________________________

04-21-2018, 11:49 PM	#19
Shoeless' Socks Minors (Double A) Join Date: Jul 2017 Posts: 162	IMO at the major league level, my theory is that defense only matters at the marginal extremes. Unless a defender is elite, or atrocious, it doesn't matter much. Nothing will ever convince me otherwise. The lower you go, the more defense matters. Coach pitch little league it is vital. At the ML level, it matters very little, IMO. I do not believe that there is a player alive that can save as many runs with his glove that he can create with his bat. Not given 500+ ABs.

04-24-2018, 09:37 AM	#20
Déjà Bru Hall Of Famer Join Date: Apr 2009 Location: Long Island Posts: 10,991	This, from Baseball Reference, is key - to me, that is. For others used to dealing with WAR for years now, it's probably basic but to me this was quite illuminating. I am assuming that the concept is the same in OOTPB even though I know and accept that OOTPB's calculations vary somewhat from Baseball Reference and fWAR. Attached Images __________________ - Bru