Introducing the Weighted Net Steal

We’ve seen an absolute array of statistics revolutionized by the sabermetric era of baseball. All sorts of acronyms that confuse the average fan–UZR, wOBA, and FIP–are becoming more and more accepted among baseball communities. But those statistics only cover defense, hitting, and pitching–leaving out one of the major baseball tools. The statistical side of the final segment of baseball (base running) has hardly been changed since Day One when the stolen base stat was drawn up.

As an advanced thinker of baseball, this bothered me. Sure, there was the net steal (simply stolen bases minus times caught stealing) and the newly-christened Ultimate Baserunning, but neither covers the whole story. Net Steals don’t appropriately value the damage done by being caught stealing, and UBR doesn’t even factor in stolen bases.

Many teams–including the much publicized Moneyball Athletics–have stopped trying to steal bases in general. Why? Because the risk is so high, taking an extra base is often a gamble with a low return. Successfully stealing one base in every two tries doesn’t break even, that extra out created ultimately doesn’t justify taking an extra base. Having a man on second is clearly much more valuable than having a man on first, but there is value enough in just having a man on base–without using up one of your 27 outs.

According to studies done by James Click on stealing bases, a runner needs to steal successfully about 73% of the time to break even. Now, this number fluctuates depending on which base is being attempted at and how many outs there are. But the point is that the rate you need to make it into the black is much higher than the previously used 50%.

In simpler terms, this means that for every three times a runner is caught stealing, he needs at least eight stolen bases to not be hurting his team. Even eight out of twelve swipes is a failure. And with that in mind, it gives us the formula for Weighted Net Steals (abbreviated wNS):

wNS = SB – (8/3)CS

Before we break down the leaders in weighted net steals, let’s take a look at the top 10 base stealers of 2011 based on stolen bases:

From just looking at stolen bases, Michael Bourn is by far the best runner, followed by Gardner, Kemp, and a pack of seven equally good runners. But once you weigh the disadvantage of being caught stealing, the rankings shake up dramatically. Kemp, Bonifacio, and Stubbs fall out of the top 10, and Ellsbury is so inefficient that he receives a negative grade for wNS.

Here are the top twenty base stealers of 2011 based on net steals:

Net steals are much better than pure stolen bases when it comes to assessing the best base stealers in the game. But as discussed earlier in this article, it doesn’t put enough importance on being caught stealing. Gardner, Stubbs, Bonifacio, Kemp, Andrus, Upton, and Ellsbury were all caught stealing double-digit times, yet they remained in the Top-20. Unfortunately, Net Steals doesn’t exactly measure efficiency for steals, it just slightly penalizes overly-aggressive base stealers.

Here are the top twenty base stealers of 2011 based on weighted net steals:

What we see now is that raw stolen base totals aren’t so important. Craig Gentry cracked the top 10 list, and he didn’t even swipe 20 bases all year. He just was never caught. Will Venable and Eric Young made it in at #6 and #10 respectively, despite only taking 53 bases combined. But they were only caught 7 times all year.

Your strength as a baserunner doesn’t really depend on how many times you steal a base. Runner X who steals 40 bases and gets caught 15 times has the same wNS score as a runner who takes eight bases while getting caught three times. It’s about efficiency.

Weighted Net Steals also somewhat takes a page out of the book of Simpson’s Paradox. The stat doesn’t revolve around just efficiency, or Adrian Gonzalez would be the best baserunner in the league with one steal in one attempt. The aforementioned Craig Gentry may be 18-for-18 in steals, but he’ll get a lower wNS score than Cameron Maybin, who went 40-for-48, since Maybin has stolen many more bases, even after being caught eight times. Of course, the difference between their scores (18 and 18.7) is minute, but it shows that it takes more than never being caught to be a good baserunner.

Base stealing is one of the last fronts in baseball that still has room to grow in terms of advanced stats. But I believe that wNS is a step in the right direction. The stat doesn’t measure speed; home-to-first times are more valuable if that’s what you’re looking for. Rather, wNS shows just how much value a player adds to his team. Each whole number is an extra base added (just like stretching a single into a double), and a score of zero means the player had no impact (positive or negative) on the basepaths.

Is Jacoby Ellsbury an efficient base stealer? In 2011, no. But for his career, very much so.

These values, though, are not meant to be predictive. In fact, they can even fluctuate greatly between seasons. But what they do illustrate is the efficiency of the player’s baserunning.

For example, you can read into Jacoby Ellsbury’s 2011 wNS score of -1 and see a tick-below-average runner. That may be true for last season, but you can’t extrapolate that number to say he’s an average runner overall. In fact, for his career, he’s a wonderfully efficient runner with a career wNS of 71–good for 14.2 per season. Before last season, he posted wNS scores of 4.3, 38, 20.7, and 9.

Some teams have sworn off stealing because of the repercussions of being caught stealing. But that’s not the best strategy. When done efficiently, stealing is an easy way to compensate for a lineup that may be lacking in pop. Maybe that’s why the Moneyball A’s never won anything.