WAR - The meaningless statistic

[Jim Klag]

I know this subject must have been discussed here before, but I want to know why the baseball people who use statistics as what they optimistically call "analytics" use WAR as some sort of benchmark statistic. I have seen most of the many definitions of WAR and realized it is total mathematical fiction - i.e., no mathematical basis whatsoever. There is no mathematical definition of a "win" and no mathematical definition of a "replacement", so I call bullshit! Here is an example of WTF when it comes to WAR. Mike Minor, the Texas Rangers pitcher has a 2019 WAR of 5.1. Freddie Freeman, the Atlanta Braves firstbaseman has a 2019 WAR of 3.2. See where I'm going? Freeman is a plus defender and in the top 10 of virtually every offensive statistic. Minor is an okay pitcher having a really good year, who is in 3rd place among qualifiers for the ERA lead. He is in the top ten in no other pitching statistics except complete games and shutouts and his total for those two stats is one (1) complete game and shutout. Someone explain to me how a pitcher who participates in only 1/5 of his team's games has a "superstar" WAR rating and a perennial all-star firstbaseman's WAR is only 3.2. Think WAR might contain a certain amount of voodoo and subjectivity?

 

[Timby]

fWAR is completely broken because it weights certain defensive statistics and xFIP entirely too heavily, but bWAR is a mathematically sound concept. I'm heading out the door so I can't do a long writeup about it, but what I'll say is that I hate "adding up the WAR" as a way of valuing a player, but it is a legitimate metric of performance.

 

[The Nth Doctor]

Frankly, I ignore all of that crap. It baffles me that some people are so taken in by all of those special analytics.

 

[Timby]

Frankly, I ignore all of that crap. It baffles me that some people are so taken in by all of those special analytics.

If you say that you think RBI and pitcher wins are a valuable metric, I might ban you.

Might.

 

[The Nth Doctor]

Gods, no! I'm certain I've commented multiple times on how much I absolutely loathe wins in particular.

 

[Mr. Laser Beam]

WAR. What is it good for? Absolutely nothing. Say it again...

 

[Herbert]

I think WAR has some value but it's not the be-all and end-all that some people want you to think that it is. Given that there can be multiple ways to calculate it, makes me discount it.
It's not going to get any better though because it's all analytics now and nerds with slide rules are sitting in offices right now trying to come up with more "WAR" type stuff. Bet on it.

Oh, and I like pitcher wins and RBI so ban away, Timby.

 

[Jim Klag]

I think WAR has some value but it's not the be-all and end-all that some people want you to think that it is. Given that there can be multiple ways to calculate it, makes me discount it.
It's not going to get any better though because it's all analytics now and nerds with slide rules are sitting in offices right now trying to come up with more "WAR" type stuff. Bet on it.

Oh, and I like pitcher wins and RBI so ban away, Timby.

I think WAR has zero real value. It is an artificial "statistic" based on unknowns like "win" and "replacement." SABR-metricians use it because they've got nothing else. When the "analytics" crowd rule MLB, you get things like Jeff Bagwell in the Hall of Fame. Now, before you yell at me about Bagwell (who was a fine ballplayer but not HOF material), compare him to the next two guys who will likely be going into the Hall as firstbasemen - Albert Pujols and Miguel Cabrera and tell me which of the three deserve enshrinement. Players like Ted Williams, Joe Dimaggio, Al Kaline, Hank Aaron, Willie Mays and the aforementioned Pujols and Cabrera were discovered and signed without the benefit of "analytics." Wonder how they did that?

 

[Herbert]

I'm certainly no WARrior but Bagwell would be in regardless. There are plenty of players who are in the hall and have been for decades that are suspect and probably shouldn't be there. If those guys are there, then Bagwell deserves a spot. However, I don't want this to devolve into a "Hall of Fame" Vs "Hall of Very Good" discussion.

 

[Jim Klag]

I'm certainly no WARrior but Bagwell would be in regardless. There are plenty of players who are in the hall and have been for decades that are suspect and probably shouldn't be there. If those guys are there, then Bagwell deserves a spot. However, I don't want this to devolve into a "Hall of Fame" Vs "Hall of Very Good" discussion.

Needless to say I am a small Hall guy. I believe in the Hall of the truly great players. The guys who were considered the very best players in their day. Mike Trout is the very best player in MLB right now and has been for a few years. Regardless of what he does from here on, he will be a HOF enshrinee. Albert Pujols and Miguel Cabrera both had stretches where they were the Mike Trout of that time. I don't think anyone could say that about Jeff Bagwell. I don't even think he was the best player on his team. Your comment about those guys who probably shouldn't be in the Hall proves my point - Jeff Bagwell is "metrically" comparable to those guys who shouldn't be there and that is why he is going in.

 

[1001001]

If you say that you think RBI and pitcher wins are a valuable metric, I might ban you.

Might.

I understand that Pitcher Wins are essentially meaningless (they control so little of the outcome), but I'm wondering about RBI.

I'm assuming the objection is no one controls how many base runners get on ahead of them? Good hitters on bad teams might have artificially low RBI numbers because everyone else on the team can't get on base?

 

[Jim Klag]

I understand that Pitcher Wins are essentially meaningless (they control so little of the outcome), but I'm wondering about RBI.

I'm assuming the objection is no one controls how many base runners get on ahead of them? Good hitters on bad teams might have artificially low RBI numbers because everyone else on the team can't get on base?

Regardless of how RBI's are achieved, they represent something a player actually did at bat. Aside from WAR, the current darling of the "metricians" is On-base percentage. Of those times on base, how many came with exactly zero input from the hitter? A hitter can fail to dodge a pitch, get hit, and add to his on-base percentage.

 

[Herbert]

Needless to say I am a small Hall guy. I believe in the Hall of the truly great players.

I agree but that we aren't getting that horse back in the barn. What are you gonna do?

 

[Jim Klag]

I agree but that we aren't getting that horse back in the barn. What are you gonna do?

Bitch about it, I guess. I'm pretty good at bitching - futile though it may be.

 

[JirinPanthosa]

WAR's not a perfect statistic and can sometimes be overrated, but it certainly means more than things like pitcher wins and RBIs that are two thirds based on things outside your control. It measures the things within the player's control that really contribute to victory. If it didn't yield results, people would have stopped using it by now.

You could come up with some more useful RBI statistic like "Percentage of runners in scoring position batted in", but all you're really doing there is reducing your sample group and letting in more standard deviation.

The player may not be totally in control of walks, but smart, patient batters get a whole lot more of them and they're just as good as a single when there aren't runners in scoring position.

Also, what other stats take into account things like grounding into double plays? Things with a huge negative impact on the game, that the runner is in control of? Should those count against your RBI total? What about fielding efficiency? Errors are only counted if you messed up a play that anyone in the league should have made, there's no error if there's a borderline play that some fielders could have made and others would not.

Sports commentators whose goal is entertainment should talk about WAR less, because they don't tell a good story. (I HATE when ESPN Gamecast shows the 'percent to win' stat). But for stats nerds, it's a very useful stat.

 

[JirinPanthosa]

https://sports.theonion.com/baseball-statisticians-unveil-new-analytics-model-measu-1828577747

 

[Jim Klag]

WAR's not a perfect statistic and can sometimes be overrated, but it certainly means more than things like pitcher wins and RBIs that are two thirds based on things outside your control. It measures the things within the player's control that really contribute to victory. If it didn't yield results, people would have stopped using it by now.

You could come up with some more useful RBI statistic like "Percentage of runners in scoring position batted in", but all you're really doing there is reducing your sample group and letting in more standard deviation.

The player may not be totally in control of walks, but smart, patient batters get a whole lot more of them and they're just as good as a single when there aren't runners in scoring position.

Also, what other stats take into account things like grounding into double plays? Things with a huge negative impact on the game, that the runner is in control of? Should those count against your RBI total? What about fielding efficiency? Errors are only counted if you messed up a play that anyone in the league should have made, there's no error if there's a borderline play that some fielders could have made and others would not.

Sports commentators whose goal is entertainment should talk about WAR less, because they don't tell a good story. (I HATE when ESPN Gamecast shows the 'percent to win' stat). But for stats nerds, it's a very useful stat.

Exactly what does WAR measure? There is no mathematical definition of a "win" or a "replacement." There are so many different "formulas" for WAR that I can't even find them all. All the stats people worship at the altar of WAR but no one can tell us what it means.

 

[Herbert]

https://sports.theonion.com/baseball-statisticians-unveil-new-analytics-model-measu-1828577747

 

[JirinPanthosa]

Exactly what does WAR measure? There is no mathematical definition of a "win" or a "replacement." There are so many different "formulas" for WAR that I can't even find them all. All the stats people worship at the altar of WAR but no one can tell us what it means.

Using 'Win' as the unit is just a futile attempt to make it more understandable to non-stats nerds.

It represents how many more wins the team can expect to get with the player on the roster than with the next available replacement (By definition, the best minor league player).

So the best minor league player is defined to have a WAR of 0.00, and the stat is how many more or less wins the team will get than that player.

 

[Jim Klag]

Using 'Win' as the unit is just a futile attempt to make it more understandable to non-stats nerds.

It represents how many more wins the team can expect to get with the player on the roster than with the next available replacement (By definition, the best minor league player).


So the best minor league player is defined to have a WAR of 0.00, and the stat is how many more or less wins the team will get than that player.

So, what if the real next available "replacement" is a major leaguer who is not a whole lot worse than the player in question. For instance, the next available replacement for Javier Baez at SS for the Cubs would be Addison Russell, not a mythical minor leaguer but a former starting SS and current starting 2B who would be replaced by Ben Zobrist. Isn't that a much more likely scenario (in fact 90% more likely on most teams) than replacing Mike Trout or any other starting player with a minor leaguer. It seems to me for a "stat" to work, it must work in every situation. Wins or RBI may not be fully within a player's control, but they have real definitions (only one definition at that - at least as "wins" concern starters) and everyone knows exactly what they mean.

It represents how many more wins the team can expect to get

How? Show me the math.