October 2, 2012
WARP for People Who Didn't Like Math Class
Over the weekend, there were plenty of end-of-season retrospectives from columnists who cast non-existent ballots for the MVPs, Cy Young award winners, and Rookies of the Year. As might be expected, many of the columnists brought up the WARP (Mike Trout) vs. Triple Crown (Miguel Cabrera) angle. There was a common theme running through the pieces that argued for Cabrera: WARP is a complicated and math-heavy stat, and because it is so complicated, how can we be sure that Trout was actually the better player?
WARP (Wins Above Replacement Player) does take a little bit of math to arrive at, and not everyone enjoyed math class in high school, but it's actually a pretty simple theory. In the spirit of fairness, I will lay out the basic idea behind WARP. You can make up your own mind from there.
I promise, there won't be many gory mathematical details.
W is for Wins
The first step in generating WARP is figuring out how many runs each player has contributed to his team or taken away from the other team.
It's true that a double brings you halfway around the circuit, but what a double really does is give a team a better chance to score a run than it had before. If, before the double, the bases were empty and there were two outs, the chances of the batting team scoring a run were low (say 10 percent—I'm making numbers up for illustration). By reaching second, a batter improves his team's chances of scoring in the inning to 40 percent (again, fake number). Just by getting to second, he’s added 30 percent of a run (0.3 runs). This gets credited to his account.
If there were runners on base, a double is also good because any runners on second or third go from being potential runs to actual, scored runs(!) Maybe the guy on first scores too. The batter didn't put those ducks on the pond, so he doesn't get credit for them. But if a runner on second has a 40 percent chance of scoring before the double, he has a 100 percent chance of scoring after the double. The batter who hit the double added 60 percent of a run.
Here's an important thing to note. Let's say that there are two fictional players, Smith and Jones. Smith plays on a team with a bunch of guys who can't hit. Smith hits a lot of doubles, but there's never anyone on to drive in, and no one hitting behind him who can drive him in. Jones is lucky and hits behind a couple of guys who are always on base. Jones' team scores more than Smith's. But Smith and Jones both hit the same double. Should we penalize Smith for the fact that his teammates are terrible? WARP says no.
As far as WARP is concerned, Smith and Jones get the same amount of credit for their doubles (usually the average value around the league that a double adds to a team's chances of scoring). In this way, we can compare apples to apples, and Smiths to Joneses.
There are other ways to add value on the bases. Going from first to third on a single is like "stealing" an extra base. So is going from second to third on a groundout. Then again, you might be thrown out on the basepaths and take away a chance for your team to score.
When evaluating baserunning, we usually compare a player's performance to the rest of the league’s. If on a single, about 70 percent of runners across baseball go from first to third, and you get to third 80 percent of the time, you have added value above what the average player would have done.
No fielder will get to every ball. But there seem to be a lot more balls that trickle into center field with some shortstops than with others. There are some center fielders who seem to have a lot of putouts, rather than just being the guy who fielded the base hit. Every time you throw a guy out, you get the credit that comes with stopping the other team from scoring. Every time you let a ball through or make an error, your account gets docked. Usually, fielders get compared to what we would expect from the league-average defender.
Summing it all up
Often, the number of runs that a player is responsible for is converted into wins. The rough rule of thumb is that 10 runs equals one win. It changes a little bit from year to year, for reasons that we won't get into here. The point of that is so that we can compare players across years. If you are comparing two players from the same year (say Miguel Cabrera in 2012 to Mike Trout in 2012), it's not that big a deal. But that's why you'll often see wins above replacement, rather than runs above replacement.
ARP is for Above Replacement Player
The team would find the next-best player it had to play that position. He might be the team's utility infielder/fourth outfielder. He might be a hot-shot prospect (or an "insurance" veteran) from Triple-A. He might be a guy on the waiver wire trying to catch on. He won't give you zero production, but there's a reason that he's either on the bench or a journeyman. This is a "replacement" player. The nice thing in baseball is that these fourth outfielders and utility guys do get to play sometimes, and we can see how well they produce. The important thing to note here is that position matters. It's a lot easier to find a guy who can play first base than one who can play shortstop (and not embarrass himself). Brendan Ryan can hit below .200 and still have a job because he's that good on defense and he plays short. No first baseman would ever be allowed to do the same.
Each player is compared to the average backup player in baseball that plays his same position. So, at the end, we can say that Smith is X number of runs (and wins) better than some backup who also plays his spot.