Please explain WAR to me

[The Truth]

I've got an open mind, but I don't understand WAR. Looking at these all-time rankings, a few very questionable things jump right out:

 

http://www.baseball-refer...leaders/WAR_career.shtml

 

First of all how is Albert Pujols ranked below people like Frank Robinson & Eddie Matthews when his career numbers in every category are higher? Seriously, have you looked at Matthews' splits vs LHP?

 

Secondly, how is Cal Ripken Jr ranked that high? Over Pujols? With a career .788 OPS?

 

How is it possible that Scott Rolen is ranked higher than Jackie Robinson!!!!?

 

How is Bob Feller ranked at only #128, below Kenny Lofton?

 

How do Bert Blyleven and Fergie Jenkins get ranked ahead of Pedro Martinez?!

[The Truth]

In what statistical category, or stretch of the imagination (!) is Phil Niekro better than Pedro Martinez?

[RockCoCougars]

In what statistical category, or stretch of the imagination (!) is Phil Niekro better than Pedro Martinez?

It doesn't matter what the reputation of these guys was, or whether you watched them play with your own eyes. If Niekro had a higher WAR, he was a better pitcher, case closed.

[DuWayne Steurer]

For one thing, WAR is cumulative, it's not a rate stat.

 

A guy who's played slightly above average baseball for 20 years will have a higher WAR than a guy who's been an all star for 3-4 years.

 

As I understand it, a 2.0 WAR is an average player for 150 games. a 5.0 would be an all star caliber performance, and an 8.0 WAR is an MVP/Cy Young level performance.

 

So if you have a guy who's 1.0 WAR after 75 games, that doesn't mean he's below average, that means he's on pace to be just about average.

 

Now a guy like Ripken will have a high WAR based on his defense. A shortstop can have MUCH higher fluctuations in defensive WAR than a 1st baseman. Also, take into account that the league OPS during Ripken's time would have been a lot lower than during Pujols, time, and WAR is compiled relative to your competition.

[The Truth]

OK, RockCoCougars, I see you wrote that in Blue...

 

Martinez - career 2.93 ERA, career 1.054 WHIP, Career 10 K/9, won 3 Cy Youngs

Phil Niekro - career 3.35 ERA, career 1.268 WHIP, Career 5.6 K/9 , zero Cy Youngs

 

I don't get it...

[The Truth]

Now a guy like Ripken will have a high WAR based on his defense. A shortstop can have MUCH higher fluctuations in defensive WAR than a 1st baseman. Also, take into account that the league OPS during Ripken's time would have been a lot lower than during Pujols, time, and WAR is compiled relative to your competition.

Ripken was pretty good defensively, but he was no Ozzie Smith....

[Funketown]

OK, RockCoCougars, I see you wrote that in Blue...

 

Martinez - career 2.93 ERA, career 1.054 WHIP, Career 10 K/9, won 3 Cy Youngs

Phil Niekro - career 3.35 ERA, career 1.268 WHIP, Career 5.6 K/9 , zero Cy Youngs

 

I don't get it...

Niekro threw twice as many innings, but Martinez still put up 79 WAR to Niekro's 98. Pedro was far more dominant than Niekro during his tenure in the majors, but Niekro stuck around for way longer.

 

You see similar stuff if you look at all-time batting averages. Pete Rose has the record for hits, despite hitting "only" .303, while Ty Cobb hit .366. Again, the difference is simply in the fact that Rose had 2,500 more at-bats than Cobb.

[RockCoCougars]

Niekro pitched for a lot of bad teams, but he was not a very good pitcher. The only reason that he was able to accumulate the stats that he did was that he started 40 some games a year in his prime, never got hurt, and pitched until he was 50. The whole 'best knuckleballer ever' thing doesn't play for me either. Wilbur Wood was better in the 70's, but he got drilled by a line drive in the leg, basically ending his career.

 

Martinez on the other hand was basically a late 60's Bob Gibson clone smack in the middle of the roid era.

[rluzinski]

In what statistical category, or stretch of the imagination (!) is Phil Niekro better than Pedro Martinez?

If Niekro had a higher WAR, he was a better pitcher, case closed.

The guy is asking for an explanation for what WAR is and you respond with a snarky comment that only highlights your ignorance of it. WAR doesn't tell you who was "better"; that's a very subjective term in the first place.

WAR stands for wins above replacement. Replacement level is the theoretical performance you would get from a typical AAA "lifer", which is considered "freely available". If you don't like the whole concept of replacement level, just pretend it's "wins above average" because it doesn't matter what the baseline is when you are just comparing players.

A win is toughly equal to 10 runs so WAR basically is adding up all the runs scored or saved above a base line for the player's career. There is one TERRIBLY important category that Niekro crushes Pedro in and that's IP (nearly twice as many!). That might not make Niekro "better" but some consideration should be made for career length. I consider WAR more of a value stat than a "better" stat. A 4.0 ERA starting pitcher with 225 IP is more valuable to a team than a reliever with a 3.5 ERA in 50 IP.

"Better player" comparisons usually center around a player's prime years production.

 

[Funketown]

Niekro pitched for a lot of bad teams, but he was not a very good pitcher.

He wasn't exceptional, but he was above average and he threw an incredible number of innings. WAR doesn't say he was dominant, simply that he was a pretty good pitcher over the course of a very long career.

[MJLiverock]

So how on earth are they calculating defense for guys that played 30 years ago? I can't believe it is is too accurate for that metric. The current system has questions, let alone putting a number on the old timers.

[bjkrautk]

"Wins Above Replacement" is a cumulative stat, comparing the player to the mythical "replacement-level" player at the same position. Because it is cumulative, the more a player plays, the greater the positive effect.

 

Take a player with decent offensive numbers, have him not miss a game in seventeen years, with him spending most of that time playing a position where offense is at a premium (particularly in the era when he played)....and the number gets large enough to explain Cal Ripken.

 

Blyleven started 685 games in 22 seasons; Pedro started 409 games in 18 seasons.

Jackie Robinson didn't play a single game in the majors until he was 27 or 28 (based on his 'age' following the baseball-reference link above).

 

 

Comparing WAR across careers is interesting, but to my mind (and I'll be the first to admit that I don't know how it's calculated) the stat has more value on a season-to-season basis, as an indicator of performance relative to a player's positional peers over a fixed length of time. If you're just looking at a 162 game chunk (or even projecting a partial season out to a full 162), it's a far more useful tool in evaluating players than it is in the aggregate....largely for some of the reasons your examples have highlighted.

[TheCrew07]

So how on earth are they calculating defense for guys that played 30 years ago? I can't believe it is is too accurate for that metric. The current system has questions, let alone putting a number on the old timers.

Without doing research my guess would be some sort of system like Total Zone which has enough holes you could drive a dump trunk through it.

 

 

[DuWayne Steurer]

So how on earth are they calculating defense for guys that played 30 years ago? I can't believe it is is too accurate for that metric. The current system has questions, let alone putting a number on the old timers.

Probably just using range factor, which is a terrible way to judge defense.

[rluzinski]

So how on earth are they calculating defense for guys that played 30 years ago? I can't believe it is is too accurate for that metric. The current system has questions, let alone putting a number on the old timers.

It's probably more accurate over a career but I share still your concerns. Of course, nothing stops you from just comparing batting or pitching WAR and ignoring defense altogether. Hell, Fangraphs and Baseball Reference doesn't even use the same methodology for pitching WAR. WAR is just a basic framework; people can mold it all different kind of ways. There is nothing absolute about them.

A link to Fangraph's WAR primer:

http://www.fangraphs.com/blogs/index.php/glossary/#winvalues

 

[RockCoCougars]

In what statistical category, or stretch of the imagination (!) is Phil Niekro better than Pedro Martinez?

If Niekro had a higher WAR, he was a better pitcher, case closed.

The guy is asking for an explanation for what WAR is and you respond with a snarky comment that only highlights your ignorance of it. WAR doesn't tell you who was "better"; that's a very subjective term in the first place.

WAR stands for wins above replacement. Replacement level is the theoretical performance you would get from a typical AAA "lifer", which is considered "freely available". If you don't like the whole concept of replacement level, just pretend it's "wins above average" because it doesn't matter what the baseline is when you are just comparing players.

A win is toughly equal to 10 runs so WAR basically is adding up all the runs scored or saved above a base line for the player's career. There is one TERRIBLY important category that Niekro crushes Pedro in and that's IP (nearly twice as many!). That might not make Niekro "better" but some consideration should be made for career length. I consider WAR more of a value stat than a "better" stat. A 4.0 ERA starting pitcher with 225 IP is more valuable to a team than a reliever with a 3.5 ERA in 50 IP.

"Better player" comparisons usually center around a player's prime years production.

 

Niekro and Martinez have basically the same WAR/year ratio. That seems incredibly flawed to me. I know that Martinez was basically injured the last few years of his career, but Niekro didn't pitch much at the end (or the beginning) as well.

[theavrock]

http://www.beyondtheboxscore.com/2010/5/11/1465549/inspired-by-ross-barnes-best-war

 

I just did a quick google search for this. I am sure there are better tables, but this ranks position players with most WAR's per 700 PA's.

So this takes out the guy who was slightly above average for a long time and focuses on those that provided the most wins per ~ 1.5 seasons IE: you get a better idea of someone who consistently has a high WAR value each season.

For instance in regards to one of your examples: Scott Rolen doesn't even sniff the top 25, but Jackie Robinson is #15.

[Funketown]

http://www.fangraphs.com/graphs/1010978_1177___sgraph_%20_7_14_2011.png

Fangraphs has a somewhat fun tool called a WAR Graph that you can use to plot player's nth best season against their WAR in that season. The above graph is for Pujols and Ripken. From the graph, you can see Pujols accumulated more WAR in his ten full seasons than Ripken did in his ten best seasons (combined).

 

I'm not terribly fond of their use of defense on a season by season basis, as I just don't think it accurately measures a player's performance on a season by season basis. Numbers for purely offensive WAR are easy to find on Baseball-Reference as well.

[Funketown]

Niekro and Martinez have basically the same WAR/year ratio. That seems incredibly flawed to me. I know that Martinez was basically injured the last few years of his career, but Niekro didn't pitch much at the end (or the beginning) as well.

Pedro threw an average of 162 innings per season over 18 years. Niekro threw an average of 225 innings per season over 24 years. Good pitchers accumulate more value when they throw more innings.

[igor67]

Note that Niekro averaged 14 more innings per season, which would provide a fair boost to his WARP every season.

[RockCoCougars]

Managers don't allow guys to throw more innings like they did before the early 80's. Wouldn't that artificially deflate WAR for modern pitchers?

[Funketown]

Note that Niekro averaged 14 more innings per season, which would provide a fair boost to his WARP every season.

I saw that on Baseball-Ref too (per 162 games) and thought it was way low. I'm not sure how they calculate that but if you look at their careers Niekro actually averaged more like 60-70 more innings per season than Pedro.

[topper09er]

WAR is a counting stat first and foremost and people need to realize that. It is like hits, HRs, etc. There are some players who have a lot because they got these stats at a high rate in their careers (Cobb, hits) and other players who climbed the leaderboards due to longevity (Rose). If you can understand that although Pete Rose has the most hits he should not be considered the greatest hitter of all-time, then you should be able to understand that Phil Niekro was not a better pitcher than Pedro Martinez. No one who likes WAR thinks Niekro was a better pitcher than Pedro. In fact, what you could do is turn the counting stat into a rate stat, like hits/ABs and would see Cobb rates at .366 and Rose is at .303, or say WAR/IP. If you do that you see Niekro = 98.6/5404 = 0.0179. Pedro rates at 75.9/2827.1 = 0.0268 (using BBref WAR), or nearly 1.5 times better.

 

When trying to determine who was "better" you need to understand WAR is a career stat, so often people are interested in some combination of peak prime value + career value. Here is one example of this using WAR,

 

http://darowski.com/hall-of-wwar/

 

where players get extra credit for WAR over a certain threshold to double count parts of prime years. So if player A has 5 seasons of 8 WAR and player B has 8 seasons of 5 WAR, if you count WAR over 6 as extra value for a dominant season, then player A gets 5*8 + 5*(8-6) = 50 weighed WAR for those seasons but player B just gets 8*5 = 40 weighted WAR for those seasons.

 

Another important thing to understand is that WAR is all about value, not necessarily talent or greatness. It is comprised of runs added/saved from hitting, base running, defense and has positional and replacement level adjustments. So for example, just being able to play SS has great value for your team based on how difficult it is find good ones, even though you may actually suck, you still provide some value by being able to play SS because if you sucked at 1B you would be in way of a better player in waiting. This year is the perfect example of this, Yuni sucks at SS but we dont really have great in house replacements, but if was at 1B we could bring Gamel up, or move Hart to 1B and bring Boggs up, etc. For example out of Derek Jeter's ~ 70 WAR about 11 have come only because he is a SS, independent of how good he was at offense and defense as a SS. This is because it is valuable to a team if you can play SS because it is then easier to find big bats at easier positions. If Jeter played 1B his whole career then the Yankees wouldn't have gotten Giambi or Teixeira.

 

Personally, I think offense is the same way too. Although having a highers OBP is more valuable to a team, and a guy that just sits there and walks as much as possible like Bobby Abreu is very valuable to a team, I don't feel like that necessarily means he is a better hitter than guys with higher SLG%. This is mostly my take from actually playing and realizing I could maintain a high average by trying to slap singles and walk a lot and that was valuable for being a leadoff hitter on my team, but the best players were the ones who hit the dingers. If I was in charge of the Brewers I would want all the high OBP guys because that is more valuable to a team, which WAR reflects, but I don't feel like it as difficult to do as maintain a high SLG.

 

Remember if you just look at the WAR leaderboard and use that to determine who was better it is not that different from looking at the career hits list to determine who the best hitter is, but if you can turn it into a rate stat you can then try to compare players based on everything they did, offense, defense, baserunning etc, whereas you cant do that with any other rate stat.

[Thurston Fluff]

I have long been at war with WAR. I think stats are very useful when uses properly. War does not do that in it's calculations. Even assuming it did it is so misused/misunderstood that it causes more confusion than it clears up.

 

WAR doesn't tell you who was "better"; that's a very subjective term in the first place.

 

So why do people continue to use a player's WAR to assess who is having a better season or who is the better player? That happens all the time.

 

I have several problems with WAR.

 

One is it is continuously misused. See above.

 

Two is it doesn't use the stats in it's calculations the way they are supposed to be used. The defense in particular is used wrong IMHO. You can't use one season's worth of UZR for example when it is commonly accepted that it takes three years to come to something useful. It is silly to use something known to be inaccurate and expect the outcome to be accurate. Then you have the issue of weighing each component properly which leads me to the nest issue.

 

Three nobody really seems to understand precisely how it derives it's outcome. We seem to know what goes in but not sure what measure of each and such. Hell Rluz once said he wasn't exactly sure how it was derived and he's one of the most astute stats guy here. It's to baseball stats what derivatives are to banking.

 

My final issue is more based on my skepticism of the very concept that one stat can be all encompassing than on WAR itself. I understand the desire for an easy to read, easy to understand number that encompasses everything we need to know about a player or when comparing two players. Problem I see is it isn't possible to do with any amount of accuracy. At least it hasn't been done yet. I applaud the attempt. It may turn out I am wrong and it can be done. In that respect I have no problem with the attempt. Just the outcome of said attempt as useful for anyhting more than a stepping stone toward something intrinsically useful on it's own.

[CheezWizHed]

Managers don't allow guys to throw more innings like they did before the early 80's. Wouldn't that artificially deflate WAR for modern pitchers?

 

But isn't WAR relative to that season? So it would normalize based on how players are generally used in that season. Or even other factors like improvements in the ball or bat, park factors, or general fitness (I'll avoid 'roids here).