April 19, 2013
The Search for the Missing Steals
There was a time when goliaths roamed the Earth. Runs were plentiful, and with few natural enemies, these behemoths could grow larger and larger, more and more sedentary. They had no need to run, so they lumbered about, leisurely returning home unmolested by predator or foe.
But the metaphor could not last forever. Eventually, offense in baseball went back down to more typical levels, and the game once again welcomed back the singles hitter, the glove man, the productive out. And, as runs went down, stolen bases went up. I gave the following graph, showing leaguewide scoring and stolen base totals by year, a pun title.
That’s the narrative, and it’s a good one. No problems with that narrative. Offense goes down, stolen bases go up. With runs at a premium, home runs less available, and basestealers getting better and better at avoiding caught stealings, the small-ball tactic takes on great significance.
What’s weird is that 2013 has disrupted the narrative, so far at least, and it’s not easy to say why. Here’s the same chart above, but with runs and stolen bases (prorated across a full season) added for 2013:
Brett Gardner is still looking for his first stolen base; Carlos Gomez got his first on Thursday. Michael Bourn, Mike Trout, and Jose Altuve have one each. Last year, teams attempted .89 steals per game; this year, they've attempted .69 steals per game.
So what’s to blame? Easier to rule out what’s not to blame, or probably not to blame, and see what’s left.
1. A gradual league-wide shift in style of play is not to blame.
As noted in the graphs above, the trend has been toward the stolen base, rather than away from it. If we go back to the beginning of the century, the most steal-happy seasons—both by attempts and by successful thefts—have come in the past two years:
So this is an abrupt turnaround.
2. Basestealers being more choosy aren’t to blame.
Technically, they are choosing not to steal, so that is literally the reason. But it isn’t like baserunners all realized that getting caught's not worth it and decided they're not going to steal unless it’s a sure thing that they'll be safe. The league’s 73.4 percent success rate through the first 13 games is about identical to the league’s 73.5 percent success rate through the first 13 games of years 2007 through 2012; it’s only slightly better than the 73.1 percent overall success rate since 2007. Which is all just to say that this isn't a stolen bases phenomenon but a stolen base attempts phenomenon.
3. The particular characteristics unique to the month of April are not to blame.
You might guess that the first two weeks would be a slow time for stolen bases—weather, and injury concerns, and players working their way back to full speed. It’s not. From 2000 to 2012, teams attempted .827 steals per game in their first 13 games. They attempted .827 steals per game in their final 149 games.
All the same, if we compare 2013 thus far to the previous dozen years at comparable points (which is to say, the first 13 games of each team’s season, 13 being the number I’m choosing because there are teams this season that have played only 13 games), we have to go back a full decade to find a season as uninterested in basestealing as this one:
Comparing apples to apples, there are 87 stolen bases missing since last year.
4. The limited predictive power of mid-April statistics is probably not to blame.
The correlation between the league’s first 13 games and its final 149 games over the past decade is .63. There’s certainly room there for this to turn out to be a fluke, but that’s a solid correlation.
5. A lack of stolen base opportunities is not primarily to blame.
If, for instance, there just weren’t as many baserunners reaching first base (and OBP is down from last year, a tick), or if there weren’t as many baserunners reaching first base without a runner on second base, or if there weren’t as many baserunners reaching first base with two outs and without a runner on second or third (the most common basestealing situation) then that would explain it. That doesn’t explain it, because that’s not the case.
There are 24 base/out situations. Ten of them are either completely non-stealing situations (no runner on base) or de facto non-stealing situations (runner on third, bases loaded, etc). Of the other 14, baserunners are attempting to steal less frequently in 13 this season.
For instance, with a runner on first and two outs last year, baserunners stole second base or were caught stealing second base 10.76 percent of the time. In 2013, a runner has been on first with two outs 1,039 times. If 10.76 percent of those baserunners attempted to steal, there would be 112 attempts; in fact, there have been 94 attempts. Repeat that math for every situation, and it adds up to 97 missing stolen base attempts, based on this year's stolen base opportunities.
6. A preponderance of blowouts, if it existed, would not be to blame.
There have been somewhat more plate appearances this year in blowouts (a lead or deficit of five runs or greater), but even if it were a very significant disparity (it’s not) it wouldn’t be to blame. The league is stealing less often in tie games, in one-run games, in two-run games, in three-run games, in four-run games, and in blow-outs.
7. High-profile basestealers not getting on base are not primarily to blame.
Some of them might be, I guess. But there are 239 baserunners who have attempted to steal less often this year than they did last year, and just 90 who have attempted to steal more often. If every individual runner who reached first base stole at exactly the same rate that he stole last year—that is, Trout stole as often as Trout stole last year; Kelly Shoppach stole as often as Kelly Shoppach stole last year; etc.—there'd be roughly 83 more steal attempts than we've seen.
8. The temperature is not to blame.
A popular theory among BP authors is that the cold weather this spring would make baserunners more cautious, for fear of injury. It’s true that there have been more games in the 30s and 40s in the first two weeks of this year than in the first two weeks of the past three years; it’s not all that close:
Except that stolen base attempts have been higher in low temperatures. Attempts per game at each temperature tier:
And so what ideas does that leave? It leaves dampness and precipitation, for one. There's been a lot of dampness and precipitation! But the weather data at Baseball-Reference is much less helpful on dampness and precipitation than it is for temperatures. It leaves some heretofore unnoticed trend in pitchers holding baserunners on more closely or more effectively, which can't be confirmed because it hasn't yet been noticed. It leaves ... well, you tell me. What's it leave? You suggest it, I'll try to check it out.
Thanks to Colin Wyers for research assistance.