The Value of Baserunning

In evaluating players' talent and impact, we have always had biases. Certain areas of the game receive disproportionate attention while others tend to get overlooked or dismissed. It's how a player can finish second in the MVP voting with as ordinary a season as Ryan Howard's, and even get a significant chunk of first-place votes over such a standout candidate as Albert Pujols. It's how much better players like Chase Utley and Hanley Ramirez can finish far behind in the voting. And I'm assuming, though I'm not certain because the formula has been carefully guarded from the public eye for decades now, that this is how the Elias Sports Bureau can submit their official annual player rankings to MLB naming Mark Teixeira as the best player in baseball.

Perhaps the most overlooked area of the game (the anti-RBI, if you will) is baserunning. I've lost count of the number of well-meaning but ultimately misguided fans have regurgitated the supposedly sabermetric idea that baserunning holds no value. As far as I can tell, this sentiment comes from two separate tenets of sabermetrics-

1. that stolen base totals are a very poor indication of baserunning value, and that it is very possible for high stolen base totals to be completely negated by a reasonable number of caught stealings, and
2. that, in the absence of a solid objective understanding of the value of some part of the game, it is incredibly difficult to accurately assess the value subjectively. In other words, how many runs or wins is Derek Jeter's leadership worth? Since we can't answer that question with any certainty, we can't really consider it in an objective estimate of his overall value.

Since it is nearly impossible for the average fan to find any decent objective valuation of baserunning skills besides SB, CS, and SB%, it is often assumed that the second of these comes into play here, which, of course, it does. However, the extension of this principle to assume that baserunning actually holds no value because you have no measure of it steps beyond the realm of sabermetrically-minded claims. The real issue here is that sabermetrics is a constantly moving field, and so fans who accept what they have read (and sometimes misinterpreted, as with inappropriately extending the second tenet above) as unchanging fact are not, in fact, thinking sabermetrically, even to the point where they will vehemently argue with actual sabermetricians who they feel don't get the new wave of statistics. Even for someone who enjoys irony as much as myself, these conversations can often be the most frustrating.

And so it goes with the topic of baserunning. As daunting as the task may seem given the lack of resources available, it is actually possible to measure baserunning ability in terms of runs contributed, and that means we can consider it with hitting ability with appropriate weight in considering a player's offensive value. The first issue to overcome in undertaking such an assessment is that stolen bases are, in fact, a very poor measure of baserunning value, and that more value actually presents itself in how well a runner does on the base paths on balls put into play by his teammates. There are a few difficult-to-find metrics that attempt to measure non-SB baserunning value, such as first-to-third percentage and +/- rating for baserunners, but these lack both a proper consideration of context and a usable translation to run value. By looking at comprehensive play-by-play data, we can do better than these and come closer to assigning a true baserunning value. This value is largely concentrated into the following four situations:

-going from first to third on an outfield single
-scoring from second on an outfield single
-scoring from first on a double
-scoring from third on and outfield fly ball out with fewer than 2 outs

There are, of course, other places where baserunning affects a play, but they are either too uncommon or too fluky to be reliably judged from just play-by-play data, so I focused only on these situations. By compiling a database of every time one of these situations occurred in 2007, I was able to apply a linear weights formula to arrive at a run-value for baserunning. Basically, this means I calculated the run value of each base taken based on how frequently a runner scored from that base with that number of outs, and then calculated the difference in value of the bases a given player took on the base paths to what an average runner would have taken.

Beyond just looking at those 4 situations, I had to further break each situation into the number of outs on the play to get an appropriate linear weight value because their is a significant difference in the value of the same play on the bases with a different number of outs. Unlike with hitting, this difference can't be ignored for baserunning for 2 main reasons:

1. there are not enough occurrences of each situation throughout the year for the numbers to reliably even out across the possible number of outs, and
2. the difference in value of taking the extra base and of possibly running into an out in different situations affects the decision-making of a baserunner in ways it does not affect a hitter.

Furthermore, the first situation had to be broken down by which outfielder fielded the ball since it is much easier to go first to third on a single to right field than on one to left.

Once the methodology is established, all that remains is the long, tedious process of actually carrying it out. Fun stuff. The findings-

-Baserunning on batted balls is more important than stolen bases. Good baserunners can more reliably advance beyond the closest safe base without getting thrown out on balls in play than on balls not in play. The result is that the extra bases they take are much less dampened by the risk of getting thrown out.
-Stolen bases do not necessarily correlate well with baserunning value. I ran separate calculations for SB and non-SB components of baserunning value using the same methodology. The good baserunners were generally better than the bad baserunners in both, but among the good and bad baserunners, a number of them had their value distributed significantly differently among the SB and non-SB components.
-The difference between a very good baserunner and a very bad baserunner can be at least 12 runs. I only computed values for 15 players because of the time-consuming nature, but the sample included both good and bad baserunners, all of whom were very good hitters and on base a lot. The top baserunners in my sample from 2007 (Sizemore and Rollins) were about 12.2 runs ahead of the worst (Helton and Fielder). This may sound insignificant, but it's not. Ten runs are generally considered about 1 win, and anything over a win from 1 player is a significant contribution. For comparison, that is a greater difference than the difference in VORP between Adam Dunn and Aaron Rowand this season.
-The subjective consensus of a player's baserunning value is generally good at separating good from bad baserunners, but there are some exceptions. For example, David Wright gets more credit as a good baserunner than Alex Rodriguez, but they were pretty much dead even in value in 2007. As an even bigger surprise, Matt Holliday ranked ahead of Hanley Ramirez (as did Wright and A-Rod). David Ortiz, who would probably be a lot of fans' pick for worst of the bunch, ranked ahead of 4 players in the sample. This also weighs in on controversial baserunners, such as Pujols, for whom exist staunch proponents of both his great value and his great detriment on the base paths due to his aggressiveness. Turns out, he's pretty much average (shocking, huh?).

There are still some issues with this measure. It's not comprehensive. It suffers from sample size issues, and would require a few years of data to really be reliable. It does, however, show how baserunning is a skill that can be measured just like any other area of the game, and that the value in good baserunning is too significant to simply be ignored.

0 comments:

Post a Comment