Win Shares for MVP

My good buddy Aaron Gleeman recently handicapped the American League MVP race with a variety of sabermetric stats. He included stats like Win Shares Above Average (WSAA), Wins Above Replacement Player (WARP), Value Over Replacement Player (VORP), Runs Created Above Average (RCAA) and Runs Saved Above Average (RSAA). A veritable what’s what of sophisticated baseball stats.

The thing is, none of these stats could agree on an MVP. Much to Aaron’s delight, the guy who came out ahead, on average, was Johan Santana. But each stat picked a different MVP, and some had wildly different rankings. There’s a reason for that. These stats all suffer from the same infliction baseball writers do — they use different criteria to determine what makes a player valuable.

Most of these stats incorporate only those things that make ballplayers consistently great over time. Conversely, they leave out those performance skills that tend to wax and wane on a random basis — things that are typically attributed to luck. For instance, very few players have shown a consistent ability to hit better with runners in scoring position. So most run estimation tools, like WARP and RCAA, omit clutch hitting stats.

Although clutch hitting may not be a consistent skill, players sometimes have very good years in the clutch. And this can have a big impact on a team. In fact, sometimes clutch hitting can be key to a team’s success. So you might want to consider referencing stats that include clutch hitting when readying your MVP ballot.

Of course, choosing a stat would be easier if the BBWAA were clearer about its MVP criteria. That is unfortunately not the case. Instead, the BBWAA says (among other things):

There is no clear-cut definition of what Most Valuable means. It is up to the individual voter to decide who was the Most Valuable Player in each league to his team. The MVP need not come from a division winner or other playoff qualifier.

The rules of the voting remain the same as they were written on the first ballot in 1931:

1. Actual value of a player to his team, that is, strength of offense and defense.
2. Number of games played.
3. General character, disposition, loyalty and effort.
4. Former winners are eligible.
5. Members of the committee may vote for more than one member of a team.

You are also urged to give serious consideration to all your selections, from 1 to 10. A 10th-place vote can influence the outcome of an election. You must fill in all 10 places on your ballot. Keep in mind that all players are eligible for MVP, and that includes pitchers and designated hitters.

Only regular-season performances are to be taken into consideration.

Well, that doesn’t really help, does it? The key words are “actual value of a player to his team…” But how do we determine value?

Let me posit this: the ultimate goal of a team is to win games. That’s it, end of story. Therefore, a player’s value is founded on what he does to help his team win.

Most voters seem to agree with this philosophy. Just look at the emphasis they put on division winners, or RBI’s, or wins for a pitcher. These stats suggest which players have contributed the most to help their teams win, but they’re superficial indicators at best and sometimes they’re downright misleading. For MVP, you need a stat that does a better job of measuring each player’s contribution to winning games.

And, of course, that’s what Wins Shares does.

I would like to make the case that Wins Shares are the best stat to use for the inevitable MVP debates. Win Shares are not good for a lot of things, like predicting a player’s future value, or defining his “true talent,” but they are excellent at determining how many wins a player contributed to his team. That’s why they’re called Win Shares. Let me give you some specific examples:


A Hardball Times Update
Goodbye for now.

The Win Shares system actually goes to great lengths to make sure the total contribution from all players equals the number of games his team actually won. None of the other stats do that.

Why is this important? Well, let’s look at the New York Yankees, who have the most wins in the American League (though they are barely hanging onto the lead of the American League East). They’ve scored 5.5 runs a game, and allowed 5.1 runs a game. Their offense, in particular, has been extremely good and their pitching/fielding has been average at best.

But if you take their runs scored and runs allowed, and run them through another theoretical formula called the “Pythagorean Formula”, they’d have about ten less wins — as many wins as the Rangers. In other words, they not only wouldn’t lead the division, they wouldn’t lead the wild card.

All the theoretical baseball stats leave those extra ten wins out. They do it for a good reason — because pythagorean variances tend to be a result of luck rather than skill — but how can you leave ten wins out of your MVP consideration? Because of this, every other stat undervalues Yankees. On Aaron’s list, only one Yankee hitter was consistently on the list (Sheffield), and no stat other than Win Shares ranked him in the top four in the league… DESPITE the fact that the Yankees have the most wins in the league, and are obviously led by the strength of their hitting. Even if you hate the Yankees, that just doesn’t seem right.

Clutch Hitting and Scoring Runs

I’ve already mentioned clutch hitting. Many sabermetricians like to say that clutch hitting is random, and shouldn’t be used to judge a player. They have a point. But when players DO hit well in the clutch, their teams score more runs. And most theoretical stats don’t take this into account.

Let’s use another example. The Chicago White Sox are hitting .267 overall, but .294 with runners in scoring position. In particular, Magglio Ordonez and Timo Perez (!) have hit extremely well with RISP, as have several other players. As a result, they have scored 698 runs.

However, the standard Runs Created formula, which does not include an adjustment for clutch hitting, projects 656 runs for the team — over forty runs LESS than they actually scored. If you refer to this Runs Created formula or most other sabermetric stats, you are leaving 30 to 40 runs out of your MVP debate.

If you add clutch hitting stats to Runs Created, the White Sox’s runs created total increases to 678. It makes up half the difference. What about the other 20 runs?

Well, this is where Win Shares also does something that no other stat does — it essentially credits runs to players, based on the team total of runs scored. It doesn’t shortchange them if Runs Created don’t equal the total. It balances out the difference, and credits players appropriately.

Most other sabermetric stats will shortchange Sox hitters by all forty of those runs. Win Shares makes sure those runs aren’t lost, and the hitters receive their due.

Pitching, Fielding and Preventing Runs

It takes two things to stop the opposition from scoring: good pitching and good fielding. It is extremely difficult to separate the relative impact of pitching and fielding, and many sabermetric stats do not try. But Win Shares does try. And Win Shares goes a step further by making sure that all the run prevention credit given to pitching and fielding actually balances with the number of runs allowed. As far as I know, no other sabermetric stat does both those things.

This is a VERY complex subject, so I’m not going to get into it here. But the net effect is that pitchers who rely a lot on their fielders receive less credit than those pitchers who control the game regardless of their defense; Randy Johnson comes to mind. And full-time position players receive credit for both their bats and their gloves.

Win Shares are not perfect. We’ve made a few changes to Win Shares on our site, and I’m sure Bill James has his own changes in mind. The three changes we’ve made are:

  • We’ve credited players with negative Win Shares, which Bill James doesn’t do.
  • We’ve toned down the credit that Win Shares gives players for saves, because we think it’s more accurate in today’s save-happy environment.
  • We’ve added a stat called Win Shares Above Average, which we believe is a better way of defining a player’s value (thought it is admittedly only a step on the way to reaching Win Shares Above Replacement, which would be better still).

I believe these three changes address several of the biggest complaints about Win Shares, and make it more accurate as a value measure. But more can be done to make the model better, particularly the split between pitching and fielding Win Shares.

Win Shares are certainly not perfect, and they shouldn’t be used for a lot of different things. But for determining a league’s Most Valuable Player, they can’t be beat.

References & Resources
Our Win Share stats are available on our stats page. You can also find sortable Win Shares at Bryan Donovan’s site.

Dave Studeman was called a "national treasure" by Rob Neyer. Seriously. Follow his sporadic tweets @dastudes.

Comments are closed.