Putting the V in MVP
In the next month, there will undoubtedly be a ton of debate surrounding the MVP award, in both leagues. While people will have many angles for their choice, from a sabermetric perspective, we should all be on the same page, or at least understand where each other are coming from.
In this entry at The Book Blog, there is an in-depth discussion of last year's MVP award, and, more importantly, the process (or processes) one should use to pick the winner. What follows is my attempt to convert that lengthy thread into an article, and hopefully add my own twist. So, thanks in advance to Tango, MGL, and all the commenters over there for their help in shaping my opinion on this matter (of course, if I screw something up, which is almost inevitable, don't blame them!).
Anyway, let's get back to the discussion. There's one word here that really throws everyone off, and that is value. How do we define value? Well, there isn't a simple answer. If you read the above linked thread, there are three views that come up most often:
Examples: Batting Runs, VORP, Runs Created, etc.
If you're in this group, you believe that clutch performance shouldn't be considered in the MVP voting process. A home run in a 10-0 blowout is worth just as much as a walk off homer in the 9th. An example of a context neutral batting stat is Pete Palmer's Batting Runs (which we've discussed here before). As you can see from the formula, each homer (or any event, for that matter) is worth the same (1.4), regardless of when it occurs.
Example: Win Probability Added
Now, we're looking at "clutch" performance or, more generally, context. WPA looks at how much each event changes a team's chance of winning. So, by WPA, a solo homer in a 10-0 blowout may be worth, oh, let's say .01 WPA points (or virtually nothing) and a walk off homer in the 9th might be worth around .5 WPA points. There's a huge difference there. So, if a player does well (or poorly) in clutch situations, it's going to impact his MVP candidacy, under this process. Note that the player's team doesn't necessarily have to win; the team can lose but a player can still gain WPA points, or contribute to a theoretical win.
Must Contribute to Real Wins
Example: This may be a good attempt
Unlike the above process, here you're only counting performance that directly affects the team's win total. If a player hits three home runs in an 8-6 loss, he doesn't get any credit. This is certainly going to favor players on winning teams and players that do good in wins.
It's important to note that these are just three general groups. There are surely others out there that can be considered, and of course there are sub-groups inside of these groups and so on. The point is, as Tango says, you've got to pick a position and stick with it. There's a good chance that there are three or four reasonable MVP candidates in each league, depending on your stance.
Fielding and Other Stuff
As you'll note, we've only talked about offense so far, really. We can't ignore fielding, and base running, and the other facets of the game. At this point with fielding, we're almost always going to have a "context-neutral" stat, whether we use UZR, PMR, THT's stats, or whatever. There's no clutchness factor in any of the fielding metrics (you wonder why Derek Jeter doesn't fare well ; ). Until someone makes a WPA-like fielding stat, we're going to have to use what's available. Also, there are a slew of other things to consider, like, as mentioned, base running, positional adjustments, park adjustments, and so on.
When a writer talks about a player's huge hits in big wins, you're probably going to be shaking your head, as his overall numbers may not be that great. But, remember, that writer may just be onto something. While that player may not have been the best player in the league, he just may have added the most value.