Insightology

First let's discuss leverage and then get to (the Philadelphia Phillies outfielder) Burrell. Leverage can briefly be described as the ability to exert influence. When a person has the ability to influence another or a situation, then that person is said to have leverage. It is a term often used in financial dealings. Have you seen it applied in sports such as baseball? Here's the set-up. In baseball are all runs equal? In other words, do runs scored in the early innings have the same importance as the runs scored in later innings? Sabermetricians will argue that they are not, simply because runs scored later have lower probabilities of being overtaken.  (Click here for an introduction to sabermetrics).

In fact, win probabilities can be constructed for every stage of a baseball game by using the average of winning percentage of all past teams that have been in that position before. A game starts with two teams each having a 50% probability of winning (assuming equal strength). If one team scores a run in the top of the first, its win probability goes up based on the average winning percentage of every team that has scored a run first. Generally it is a little above 50%. If the other team then scores two runs in the bottom of the first, then its win probability goes up higher and so on. A high leverage situation is one where the win probability can swing pretty dramatically with an event. So bottom of the ninth, home team down by a run, runner on first and two outs, is a very high leverage situation because an out or a home run on the next pitch can end the game, but with completely different outcomes. In the same situation when the home team is down by 8 runs, it is a low leverage situation. Or if the same situation occurred in the third inning it is still a lower leverage situation. These ideas were first introduced by the sabermetician Tom Tango.      

Players who do really well in high leverage situations can have a big impact on their team's win probability. Adding up all of those impacts a metric called Win Probability Added (WPA) has been developed that provides a clear measure of how impactful a player has been for his team. Pat Burrell, who seems to be having a good season, actually turns out to be having a terrific season. As of this writing, he is second among all of the major league players in WPA indicating that when given a situation with leverage he is using it really well. In fact he is the only Phillies hitter in the top 25.        

Other implications of understanding leverage are also interesting. In modern baseball it is almost an article of faith that the closer comes into the game in the ninth (or sometimes late eighth) inning when the game is relatively close. Usually this is a high leverage situation. But what if there had been an even higher leverage situation earlier in the game in the say, sixth or seventh inning? Let's say in the sixth inning the team was ahead by one run, but the other team had runners on first and second with no outs with the middle of the order coming up. If that is a higher leverage situation shouldn't that mean the closer (who is usually the team's best reliever) come in then instead of waiting for the ninth and allowing the situation to get out of hand? Sabermetricians make that argument, but few teams do that.  

Another interesting measurement that comes out of the idea of leverage is whether a player is a "clutch" player. Rather than simply saying that anyone who performs above their average in late innings (usually chosen as eighth or later) is a clutch player, one could add up the contributions of a player throughout the game in various leverage situations and measure how clutch they are depending on the sum total of their contributions. This would be a continuous and fairer measure than an arbitrary one that punishes, for example, seventh inning heroics. In fact, the WPA measure mentioned previously is a good measure of how "clutch" a player is as it accounts for performances in both low and high leverage situations.         

Does the idea of leverage apply to other sports too? It won't be as easy since other sports like football and basketball are not as discrete as baseball and hence win probabilities will be hard to calculate. But the general principle can be applied. Simply put, quarterbacks who throw more touchdowns in high leverage situations (when the game is close and it is late) are higher impact players than ones who run up the score. Similarly, all else being equal, basketball players who score more (or defend better) late in a close game have higher impact and are presumably more "clutch" players. 

A similar approach can also provide clues on when a game is "over". Bill James who is the patron saint of sabermetricians has actually done this  for college basketball games and in fact provides a helpful lead calculator that will tell you how safe a particular lead is. Check it out!