Gears of WAR, Part II: One Stat to Rule Them All

What gives a team more value, Marlon Byrd's glove or Milton Bradley's bat? Answer: they're about even. Who should we expect to contribute more to the Cubs successes in 2010: Ted Lilly or Derrek Lee? Answer: Ryan Dempster. How important was the bad baserunning of the 2009 Cubs? Answer: it cost them about one and a half wins. Whose 2009 on-field performance was the most damaging: Alfonso Soriano's or Geovany Soto's? Answer: Alfonso Soriano's -- by a LONG shot. How many wins can we expect the Cubs to total in 2010? Answer: about 85. Has Kosuke Fukudome been worth his contract thus far? Answer: roughly speaking, yes (but you could also make an argument of "close, but not quite" h/t D98).

Believe it or not, I answered all of these questions using one statistic: WAR (Wins Above Replacement). There are two things about WAR that make it the one statistic I choose above all others when I want to make a quick assessment of a player's potential impact:

1.) WAR has the same unit of measurement for everything. This allows us to compare, for example, the net impact of Marlon Byrd's glove to the net impact of Milton Bradley's bat (they're about even). It also allows us to include all aspects of a player's on-field contributions, leaving us with a very comprehensive view of a player. I'll use Carlos Zambrano as an example of this in a future post.

2.) WAR stands for Wins Above Replacement. Thus, the unit of this metric is the most important number in baseball (or any sport, for that matter): wins. It is literally a measurement of the number of wins a player contributes to his team.

Follow me more below the fold to see how WAR is calculated, and how it can be used to answer the questions above.

In my first post on WAR, I showed you how to translate individual game events into runs. Today, we'll start the conversation with a much simpler calculation: converting runs into wins. This calculation is so simple I'd wager that just about everyone reading this blog can do it in their heads. Are you ready? Here goes:

Wins = Runs/10. (Hah! I screwed this up originally! I guess it isn't that easy. Thanks for the correction, false cognate.)

It's that simple! Take the number of runs a player creates or prevents, and move the decimal place one to the left. (Aside: my dad used to buy things in groups of 10, to make it easy to figure out how much the total cost would be.) I should put the caveat out there for aficionados of these stats that the conversion rate is dependent on things like run scoring environment, and the conversation rate of 10 is more of a general rule than a fundametnally true conversion. But even most sabr-heads will convert by multiplying runs created/prevented by 10 when analyzing things on the fly. Now the real trick isn't in converting runs to wins; the real trick is in calculating the runs created/prevented in the first place. In Part I, I showed you how this was done using linear weights. This lets us calculate the number of runs created by hitters and baserunners. I'll explain how we calculate runs saved by pitchers and fielders in the next installments in this series. But I wanted to take a quick break from run creation and prevention to help motivate those pieces with an eagle's view of WAR.

WAR can answer a variety of questions such as the ones in the intro to this story. Using it, you can compare things that previously seemed like apples and oranges. For example, WAR can be used to compare the value of a good pitcher to that of a good hitter or the relative importance of a players's contributions from hitting, baserunning, and fielding. You can also add these values together to figure out team totals. And because those team totals will be in terms of wins that allows you to take projections of individual player WAR and figure out how many wins a team will tally.

Let's do a concrete example. You can subtract one player's WAR and add another player's WAR to see what the effects of a proposed move would be. For example, one can consider a trade for Kerry Wood, as Al suggested this morning. Let's say for argument's sake that the Cubs ship John Grabow and a Josh Vitters to Cleveland in exchange for Wood. The team would replace Grabow's performance - projected to be 4.36 runs allowed/9 IP - with Wood's performance - projected to be 3.73 runs allowed/9 IP - over the 54 innings Wood is projected to pitch. That's a difference of ~4 runs, or ~0.4 wins. Even if you include leverage to account for the importance of late-innings on game outcomes, the difference between the two will be 6.8 runs, or less than 1 win over the course of a season. That's not a lot, and the Cubs would be expending a lot of resources to get that incremental improvement. The team's salary would increase by $7.8M this year and $6.2M next year, AND they'd be giving up a prospect. That's far too steep a price for a < 1-win improvement on the 2010 roster. This is a great example of the utility of WAR. It lets us take a consistent approach to team decisions such as this one.