Starling Marte was caught stealing for the fifth time this season in the fifth inning Wednesday, but he has hardly been a liability on the basepaths this year. In fact, his 23 steals are tied for the most in baseball. Of players who have swiped at least 20 bags, only Billy Hamilton has been caught less than him.
Common sense would say that aggressiveness has created plenty of extra runs for the Pirates. Twitter’s collective think tank thought so, at least.
Future @BucsDugout story:— Alex Stumpf (@AlexJStumpf) July 9, 2018
Starling Marte has 22 stolen bases and only 4 CS this season. Going based on Baseball Prospectus' run expectancy matrix, how many expected runs do you think his SB attempts have created?
But common sense is a liar (sometimes). Marte’s stolen bases actually have resulted in only a fraction of a run. In fact, despite being a roughly average club in terms of stolen bases and steal percentage, the Pirates may have been better off if they never tried to swipe a bag this year.
I’m going to back-up that claim by creating two fictional players: Player A and Player B. Player A has 25 stolen bases and Player B has 30. If I presented you no other information besides that and asked you which one was the better base stealer, you’d probably say Player B. But if I told you Player A is 25/30 in stolen base attempts and Player B is 30/50, you’d probably say Player A is the superior base stealer. But how much better is he?
If you want a general rule of thumb, a player needs to steal three bases for every caught stealing for it to be worth it. But that’s a broad generalization since not all stolen bases are created equal. Sure, stealing third with two outs helps...a little. A wild pitch or balk yields a run now, but a base hit would have scored the runner from second anyway. That stolen base pales in comparison to a steal of second with nobody out.
So how can we measure the impact of each stolen base? Fear not, for we have the run expectancy matrix! This Baseball Prospectus Godsend tells us how many runs a team is expected to score in each situation they could face based on the number of outs and runners on base. A team is expected to score 2.2ish runs when they load the bases with nobody out, roughly only one-tenth of a run when the bags are empty with two away, and everything in between.
This is the matrix:
This chart is great for finding out the impact a stolen base has. If a runner goes from first to second with one out, his team went from expecting to score 0.5217 runs to 0.6592. That stolen base created 0.1375 expected runs. If he was instead caught stealing, that 0.5217 figure drops to 0.1002, meaning he cost his team -0.4215 runs.
Assuming the runner attempting to steal is the only man on base, here is how a team’s run expectancy is impacted when a player tries to swipe a bag:
Let’s look at some examples. The most valuable potential stolen base is going from second to third with one out. That situation opens up a lot of new opportunities to score, including by wild pitch, sacrifice fly or a single past a drawn in infield. Swiping that bag adds nearly a quarter of a run to a team’s run expectancy. To look at in another way, if a player did that four times without being caught, his team would, on average, score one more run that would not have happened had he stayed at second base.
Getting caught stealing in this situation could cost his team over half a run, but if he is successful just 2.3 times for every time he fails in that situation, he breaks even. Meanwhile, it would take 8.6 steals of third with two outs to recoup for making the third out there once. The most costly caught stealing is trying to swipe third with nobody out. The old adage that you shouldn’t make the first or third out at third holds up: it’s not worth the risk. The second out, though? That might be worth the gamble.
Context is important when you look at the impact of each stolen base. Let’s go back to our two theoretical players. Player A has 25 stolen bases, but if they were all steals of second with two outs, he’s only created about 2.5 expected runs. If his five caught stealings were all at second for the first out of the inning, he cost his team roughly 3 expected runs, meaning despite his fantastic stolen base rate, he’s half a run in the red.
Meanwhile, Player B stole second base with nobody out 30 times, creating roughly 6 expected runs, and was caught stealing second for the third out of the inning 20 times, costing his team roughly 4.4 runs. So despite being thrown out four times as often as Player A, Player B’s stolen base attempts were worth about 2 more expected runs than his. Baseball makes no sense.
Before I go any further, I should preemptively respond to the inevitable rebuttal: what actually happened after every stolen base? It’s a fair question. If Marte stole second with two outs and then scores on a base hit (and that is the last hit of the inning), then you can argue he created a whole run rather than just one-tenth of one. I’m going by the rate at which a team would normally score. So if Marte did that ten times and never scored (or scored regardless of the stolen bag, like on a home run), his contributions should still be recognized.
Pitchers can also change their attack plan based around stolen bases. In Austin Meadows first game on May 18, he stole second with two outs. While the pitcher had been attacking Jordy Mercer up to that point, he then decided to walk him intentionally since the pitcher was on deck. Did that stolen base help the team because Meadows got into scoring position or hurt it because the pitcher had to bat?
Ok, that’s enough talk about make-believe players and theoreticals. How about we take a look at a real Pirate? Here is the breakdown of Marte’s stolen base attempts and the impact it made on the Pirates’ run expectancy.
Well, there you go. The math says Marte’s 82% success rate has yielded less than half an expected run. Why?
14 of Marte’s 23 steals have come with two outs, meaning they didn’t have that large an impact on the Pirates’ run expectancy (only about 1.3 runs). Meanwhile, he has an egregious failed steal of second with a runner on third and one out, costing the Bucs over four-fifths of a run. That one failure basically negates nine two-out stolen bases.
It is not really Marte’s fault that he has stolen so many bases with two outs. Six of those swipes came in the first inning when he batted third, meaning the first two batters made an out, he gets on base and then steals second. Had he done that as a leadoff hitter with nobody out, he would have doubled his run expectancy added. The Pirates have moved him around in the order recently, but the Bucs’ lineup construction hurt him for the first half of the season. It’s tough to be a valuable base stealer as a No. 3 hitter. It’s why, for better or worse, managers still tend to put their speediest players at the top of the order.
The rest of the Pirates are a combined 17/31 on stolen base attempts, which is downright bad. For simplicity’s sake, I listed just their success/attempt rate and their impact on the matrix.
Even with Marte, Pirates stolen base attempts have, on the whole, cost the team two expected runs. Their stolen bases have not been in very impactful situations, and the times they’ve been caught with their hand in the cookie jar have hurt more than usual. It’s a bad combination.
Abandoning the run would be very drastic, but unless they pick better times to run or are more successful, stolen base attempts are going to hurt more often than they help.