OK, here's a long and rambling essay I've pounded out about why I like Wins Above Replacement, not as the end-all, be-all*, but as a pretty good place to start when trying to figure out a player's value.
*"Be-all and the end-all" as phrase comes from Shakespeare, from MacBeth specifically, when MacBeth used to the phrase basically to mean "All there is to it." He was referring to killing King Duncan of Scotland, and he was essentially saying that if killing him the be-all and the end-all -- if he could kill the king and have that be the end of it -- it would be done. But because there could be consequences in the afterlife, there are more issues involved.
We tend to use end-all, be-all differently, as a phrase to describe the ultimate, the very best we can get, all we would really need.
But before going to WAR (alert: I will try to avoid war puns from now on), let me talk a bit about this little discussion I had with Bill James.
Here was the starting point: Poker. Let's say you're playing three down, five-card stud poker (first three cards down, last two up). OK? You've been dealt four cards so far, and this is what you have:
2H 2D KS 8C*
*I'm suspect you know this but that's the two of hearts, two of diamonds, king of spades and eight of clubs.
Your opponent, though you don't know it, has this:
JC 8S 8D JS
So, basically, you have almost nothing, a pair of twos. This seems really good to you for reasons that are not especially clear to anyone else. Your opponent has two pair -- jacks and eights. You have not read him right and you are sure he's bluffing and so you put everything into the pot -- your house, your car, your big screen TV, your prize copy of The Machine (now in paperback!), everything. You have a very serious gambling problem, my friend, and you should look into that.
Then it's down to the last card. He gets the ace of spades. You, against ludicrous odds, get the 2 of clubs, which gives you three deuces -- YOU WON! Woo hoo! The Giants win the pennant!* You run around, do a dance, act like a big jerk, and the guy across the table is crying because he put all of HIS prize possessions into the pot (including his copy of The Soul of Baseball), and he knows that he just got beaten by someone who does not know what he/she is doing.
*Top 32 sports calls is coming this week.
From here, Bill asks a simple question:
Which card is the MVC -- Most Valuable Card?
It's a fun question that you can look at many different ways. Looking at it one way, the final 2, the 2 of clubs, HAS to be the most valuable card. I mean it showed up out of nowhere and hit the walk-off homer or got the ninth inning save -- it was the difference maker! It delivered the win! Zack Greinke -- after all I've tried to impart upon him -- would probably pick the 2 of clubs because, as he so eloquently told Bob Dutton, the two most important stats are probably "innings and wins."*
*Zack ... Zack ... Zack ... even with all the hard work Brian Bannister and I have put in, you're still showing Joe Morgan announcing signs at age 26. It's not good. Justin Verlander, Roy Halladay, Felix Hernandez, C.C. Sabathia ALL had more innings and wins than you did last year. I know you didn't enjoy winning the Cy Young last year, and I know you've grown overwhelmingly frustrated with pitching for the losing Royals City, but come on.
Back to the card game. So, some people would say the last 2 is your MVP, your Cy Young winner, whatever. Trouble is, how do you differentiate that 2 of clubs from the 2 of hearts and 2 of diamonds? They're all of the same importance. So what's the difference? Timing? Clutch performance? RBIs? Wins? And anyway, how can a TWO be the most valuable card in the game? It's a two.
Looking at it another way, the king should be the most valuable player card. It was the best card on the winning team, and the MVP award is often just that -- best player on a winning team. Trouble is the king didn't really play much of a role in winning the game and anyway it wasn't the best card in the game.
So that's another way to look at it: The ace has to be the most valuable card because it's the highest value card. It clearly had the best season. Trouble is, the ace played for a losing team. And even on that losing team, he did not play much of a role.
Looking at it still another way, the 8 of clubs could be the most valuable player card. How do I come up with that one? Well, if the 8 of clubs had gone to the other team, he* would have been the difference maker, he would have made that the winning hand (three 8s would have been decisive). By having the 8 of clubs, you prevented the other team from getting him. Trouble is, you really have to kind of stretch your mind to even come up with that philosophy.
*Eight is a very masculine number.
What's the right answer? That's the beauty of this thing ... there is no RIGHT answer. This is all just a goofy Batman could beat Superman talk. Hands, not cards, win poker games. The first hand won not because of one card but because of the three twos. The second hand lost because three of a kind beats two pair. It's likely that if you ever won a game like that with that much on the line, yes, you would always view the 2 of clubs as your favorite card. You would come away thinking that's your lucky card, you'd have a T-shirt made with the 2 of clubs on it, the 2 of clubs would become famous among your circle of friends. But it took a whole series of events, and a collection of cards to win. And remember, if that last card had been the 2 of spades or another king, you still would have won.
You can bring a little of that to baseball too ... we try to calculate individual players' value because it's fun and because we want baseball cards to increase in value, and we want posters to hang on our walls. But it isn't individual players who win games. Put the 1923 Babe Ruth on the 2010 Pittsburgh Pirates -- even assuming that he would have the same kind of year, the Pirates will still stink. They will stink less, certainly, but teams win games and teams win championships and it's telling that many of the greatest seasons in baseball history ended without the team even making the playoffs:
Walter Johnson's 1913 season? Senators finished second.
Rogers Hornsby's 1924 season? Cardinals finished sixth.
Ted Williams' 1941? Red Sox finished second.
Stan Musial's 1948? Cardinals finished second.
Steve Carlton's 1972? Phillies finished last.
Dwight Gooden's 1985? Mets finished second.
Barry Bonds 2001? Giants finished second and out of playoffs.
And these are some of the very greatest seasons in baseball history. Poker HANDS win. Baseball TEAMS win.
Still ... it's a fascinating exercise to think about how much less the Pirates would stink if they replaced Lastings Milledge with the 1923 version of Babe Ruth. Murray Chass on his non-blog blog printed a "letter," from one of the "obviously intelligent, articulate people," who agree with him on the absurdity of advanced stats like WAR. A few of the the letter's obviously intelligent, articulate words:
When an MVP-level candidate is rated 6 to 8 (wins above replacement), you can’t help but shake your head. Because, of course, it means that an MVP is ONLY WORTH six to eight more wins than a "replacement’ player." If that’s the case, then let’s stop going to Major League games, and cheer for AAA players, who must be only one or two games less worthy than their average peers in the Majors. Indeed, we’d all save a lot of money at the ballparks if we only embraced a WAR-inspired commitment to Triple A.
Sigh. It's amazing how irrational people get when it comes to making arguments against things they don't like. How many wins WOULD this person say that an MVP candidate is better than a replacement player? Twenty wins? Thirty? Has he done the math on this? His intelligent and articulate calculations would seem to assume that a team of Class AAA players would win zero games in the big leagues -- they wouldn't. The Pirates are essentially a team of replacement plyaers plus Andrew McCutcheon and Neil Walker and a couple of other guys -- but they're not going to lose 162. They're going to lose 100, maybe even 110. A full team of replacement players might lose 120, like the 1962 Mets.
More, the letter's author seems to assume that a team with, say, three MVP candidates and the rest filled with Class AAA players would win 100 to 130 games. They wouldn't. There are a million examples of this -- but let's just throw out the 1996 Mets. he Mets had three players who had great years. Todd Hundley mashed 41 home runs. Bernard Gilkey was fabulous, hitting .317 with 30 homers, 117 RBIs, 108 runs scored. Lance Johnson was excellent, hitting .333 and scoring 117 runs. The Mets had some other pretty good players, much better than Class AAA players. And the 1996 Mets still lost 91 games.
Great players don't win games by themselves in baseball. They don't. I thought that was one of the beautiful points of baseball, that teams win games.
Look: One of the great things about watching and enjoying sports is that there are no rules. You can believe what you want to believe. It's supposed to be fun. Dan Quisenberry said the best thing about baseball is that there's no homework ... I would add there are no pop quizzes. If you want to believe that baseball is won entirely by heart, that RBIs and wins are the two most important numbers, that defense can be measured best and entirely by what you see, that Jack Morris and not Bert Blyleven belongs in the Hall of Fame, that numbers deaden the sport, you are absolutely entitled -- more than entitled, you are empowered to see the game as you want to see it, enjoy it as you want to enjoy it. I'm not sure what you're doing here, 2,000 or so words into this essay about WAR, but you absolutely should watch all sports for the joy of it. I just hope you're not running a team I like.
But I like WAR. I like it because while it is complicated -- complicated enough that Baseball Reference and Fangraphs come up with different, often very different, results -- it is also extremely simple and entirely sensible. And it attempts to answer the poker question of value and the Babe Ruth question of how much difference a player can make.
Here's what I mean: Let's say you wanted to calculate a player's value. How would you do it? For an every day player -- and we'll stick with those for now -- you would try to determine the player's offensive and defensive value to his team, right? Maybe you would look at a players RBIs, stolen bases and errors. Maybe you would look at a players batting average, home runs and fielding percentage. Maybe you would look at a players wOBA, his Bill James base running numbers and his UZR. But, really, it's all going for same thing. I think the last one is more precise than the first two, but we're still going for the same thing.
And that's really all WAR is -- looking at a player's value offensively and defensively. I know the replacement thing throws people -- but I think you need to have a baseline to compare the player. You could do this any number of ways. You could pick a really HIGH baseline -- you could make your stat read Wins Below Willie Mays (WBWM) or Value Under Albert Pujols (VUAB). But that wouldn't be much fun to do and would probably tell us more about Willie Mays and Albert Pujols than the players themselves. You could look for value above average -- that what OPS+ and ERA+ do, and those are valuable stats. But there are so many positions and so many roles that "average" is kind of a tricky concept in baseball (which is why ERA+ doesn't make a lot of sense for relief pitchers, and a 100 OPS+ is really good for a shortstop, quite lousy for a left fielder).
Replacement level makes sense to me. On the most simplistic level, a replacement player is just what the word suggests -- a player you could easily get in an emergency to replace someone. He is the sort of player that is lingering on the end of the bench or is pretty good in Class AAA -- the sort of player who gets paid the minimum or could be picked up for "cash considerations" or the dreaded "player to be named later" in a minor trade that nobody would notice. Every team has loads of replacement players, even if they would not call them that. Some are starting for teams like the Royals.
Joe McEwing -- he's a replacement player. In many ways, he's the perfect replacement player. He played nine seasons for four different teams, posted a 71 career OPS+, played every position but catcher, tried really hard and was always available. If you were going to put a name to replacement player, I would say Joe McEwing is perfect* -- and it doesn't hurt that his career WAR is 0.0. The stat could easily be WAM -- Wins Above McEwing.
*I saw someone on Baseball Reference used Willie Bloomquist as his example -- same principle. Others long-timers who fit the tag: Danny Bautista, Jeff Huson, Doug Strange, Matt Mieske, James Mouton, Luis Rivera, and so on and so on and so on. It should be noted that most replacement players don't have the longevity of the players listed above. Those players offered SOMETHING that inspired people to keep bringing them back. Replacement players, necessarily, tend to have short careers.
Once you figure out what replacement level is, WAR then compares the player's value to that level -- depending on position, depending on ballpark, etc. The offensive part can be involved, but it's the easier part. People disagree about the specifics of offense, but in a larger sense people agree that it's about getting on base, it's about advancing bases, it's about avoiding outs, you know, all that stuff. Fangraphs WAR uses wOBA and Baseball Reference uses Rbat, but they're both built around linear weights (a single is worth this, a home run is worth that, etc). Base running is also factored in -- stolen bases, caught stealing, etc. Pretty straightforward.
Defense -- that one's tougher. Here's the problem: Defense has played a very small role in the mainstream definition of a player's value. Yes, sure, people give it lip service. GMs will constantly tell you that they are trying to improve the team's defense. Then they will go out and sign some 90 RBI outfielder who can't move. Media types like myself will say that defense is important and then vote defensive liabilities for MVP. You want to see something fascinating? I looked back at the MVPs in both leagues for the last 40 years. They break down like so:
First base: 17
Left field: 15
Right field: 11
Third base: 10
Center field: 6
Second base: 6
OK, does that order look familiar at all? If you're a fan of Bill James, it should. That order is almost a precise replica of Bill James Defensive Spectrum, which ranks the positions from easiest to hardest:
The Defensive Spectrum:
Look at that: Except for a couple of extra MVPs to the glamour position of shortstop, the MVPs have gone in perfect order from easiest position to hardest. Why? Because the MVP is an offensive award. Because our idea of value in large part revolves around offensive contributions -- and not just any offensive contributions but certain KINDS of offensive contribution. The last MVP to hit lower than .295 was Kevin Mitchell way back in 1989. We all know the MVP has been skewed toward RBI men. From what I can tell, the last position player to win an MVP award largely because of his defense was, I don't know: Ryne Sandberg in 1984? Zoillo Versalles in 1965? Phil Rizutto in 1950?
All three of those players, you should know, led position players in their league in WAR.
And this is one of the big things I love about WAR -- it really does attempt to look at the whole player. Defensive statistics are more advanced and more controversial than ever, and because of that Baseball Reference WAR and Fangraphs WAR can fluctuate pretty wildly (Baseball Reference uses Total Zone as their defensive stat; Fangraphs uses UZR). But the point to me is that WAR is TRYING to figure the defensive contribution. It's TRYING to get at the whole player, and not just the obvious stuff. It's TRYING to make an educated guess about a player's total value. I don't have to agree with all the conclusions. I don't have to like the inconsistencies, don't have to like that someone like Josh Hamilton is ranked as by far the most valuable player in all of baseball (by Fangraphs) and as the fourth most valuable player in the American League (by Baseball Reference). I don't have to agree with all the conclusions. I can live in shades of gray, I really can. I can decide for myself what stats are worth. I can like WAR without having it determine every single thing I believe or like about baseball.
The question I asked a few hundred words back was this: How many more wins would this year's Pittsburgh Pirates win if they had 1923 Babe Ruth in right field instead of Lastings Milledge? It's a make-believe question without a real answer. Maybe the Pirates would pitch Ruth on days he didn't play right. Maybe teams would intentionally walk Ruth every single time (which would actually ADD to his value). Maybe other players would be inspired by Ruth. Maybe they would play even worse because he was around. Maybe none of that would happen.
At the moment, the Pirates are on pace to win 55 games, lose 107. WAR says Ruth in 1923 had a WAR of 14.7. WAR says that Milledge, in 2010, is actually worse than replacement by 1.1 wins. So adding the numbers together, WAR makes the claim that 1923 Ruth would make the Pirates 16 wins better -- taking them from 55-107 to 71-91. That seems to me about as good an answer as we can find right now.