When it comes to base-stealing strategy, I love aggressive. I'm not so fond of stupid.
The key to this sentiment, of course, is defining what's meant by "aggressive" and "stupid." In search of enlightenment, let's turn once again to the pages of the baseball stat-geek's Buddhavacana. In The Book: Playing the Percentages in Baseball, Tom Tango, Mitchel Lichtman, and Andrew Dolphin ran the base-running numbers from 1999-2002. Here are three key points from their conclusions:
(1) If there were fewer than two outs, having a runner on first base who didn't try to steal led to an average gain of 14 points of wOBA for the man at the plate. The effect was even greater for veteran hitters and for left-handed batters.
(2) If there were fewer than two outs, having a runner on first base who did try to steal led not only to the loss of the 14-point advantage, but also eight additional points of wOBA on average -- a 22-point wOBA penalty overall.
(3) This third point is so key I'm just going to quote it: "The break-even point for the stolen base is highly dependent on the inning and score. The most desirable situations are tied games in the later innings or ones in which the batting team is ahead. The least desirable situations are down by at least two runs in the later innings."
The Book also gives a few examples of break-even points, for reference. Though the study uses a relatively high run environment as a basis -- five runs per game (RPG) -- the break-evens are also worth citing here, for reference:
Overall, regardless of inning and score: The break-even point as about 69 percent.
Down one run: The break-even point as about 70 percent, throughout the game.
Tied: The break-even point as about 69 percent early in the game, approximately 66 percent in the seventh inning, 63 percent in the eighth inning, and 60 percent in the ninth inning (and any extra innings beyond it).
These numbers are now generally accepted by sabermetricians. There's also a standard objection to these numbers, however: sure, that's the break-even based on (5 RPG) run expectancy, but what about the havoc that stolen-base attempts wreak on pitchers and defense?
The answer: the numbers from The Book are the best accounting we have on those phenomena, and they're pretty clear: stolen-base attempts actually disrupt the batter more than they do the pitcher or fielders. In fact, the evidence shows that holding the runner on first causes more disruption than the actual stolen-base attempt.
Now, obviously, if a team never steals, there'd be no reason to hold the runner. But nobody's ever said that teams should never steal. Quite the opposite: most analysts have concluded that teams should aim to employ the stolen base down to the score-and-out-dictated break-even point, with occasional attempts (or at least bluffs) in less-than-optimal situations just to keep opposing teams honest. The central point is that teams shouldn't steal bases just for the sake of stealing, or fall so in love with the stolen base that they throw the percentages out the window.
OK. So there's our baseline. Let's turn to Texas.
Last season, the Rangers stole 149 bases on 185 attempts. That's 80.5 percent success. Aggressive? Yes: the Rangers logged the third-most stolen-base attempts in the American League, trailing only the Rays and Angels. But also smart: Texas' 80.5 percent success rate was not only well above the average break-even point, but also far and away the best in the American League. Aggressive and smart: that's a great combination.
But 2009's history. How's Texas doing in 2010?
On the surface, things don't look too terrible, thought they're certainly not as shiny and happy as last year. So far, in 50 games, the Rangers have 49 stolen bases in 71 attempts. (That includes Kinsler's theft of third and Andrus being snuffed out at second against the Twins on Friday night.) The 71 attempts are again third in the AL, so far -- but the 69 percent success rate is only 9th out of the 14 AL teams. Not great, but given the AL run environment this season, 69 percent is right about (or even a couple points better than) the break-even overall. Based on those numbers, assuming the Rangers are running at more or less the right times, they haven't been stupidly giving runs away with stolen base attempts, and the stolen base might have even slightly increased their offensive output.
Yeah, yeah: I saw Elvis Andrus get thrown out on a pitchout Friday, just like you did. That looked pretty stupid, didn't it? But anecdotal evidence doesn't really do it for me. I want systematic evidence of stupidity, or at least the closest I can get to it.
So let's break down the numbers a little further. In particular: let's revisit the Rangers' series against the Red Sox, way back on April 20th, 21st, and 22nd. Here's what we find:
On April 20th, the Rangers stole nine bases -- three each for Andrus and Nelson Cruz, two for Vladimir Guerrero (!), and one for Julio Borbon. Not a single Rangers runner was caught trying to steal. On April 21st, Texas logged three more thefts (two for Borbon, one for Andrus); again, nobody was thrown out trying. And on April 22, Nelson Cruz stole two bases in the Rangers' two attempts. That's 14 SB and 0 CS in three games.
Running the numbers again: in the 47 games the Rangers haven't played against Boston this season, they've stolen 35 bases in 60 attempts. That's 58 percent success. Definitely not break-even.
But wait: every team's going to be worse if you take away their performance against one of the weakest performers in baseball, right? Sure. But that misses the point. The goal of base-stealing is not to be bad just as long as you're better than other teams, or to be good enough in five percent of your games to be bad in the other 95 percent. The goal's to gain as many runs as possible while giving as few away as you can. And right now, the Rangers are giving away runs, plain and simple.
(But for what it's worth: the team leading the AL in steal attempts -- the Rays, again -- have stolen 12 bases in 12 attempts against the Red Sox in seven games. Throw those out, and the Rays are still at 72 percent success. In 10 total games against the Red Sox, Tampa and Texas have racked up a full 46 percent of the stolen bases that Boston has surrendered in 50 games this season.)
As great a 2010 as he's having, Elvis Andrus has been a particular culprit. Outside of the Boston series, he's 14-for-22 (63 percent) on stolen-base attempts -- especially disappointing considering the batters following Andrus in the Rangers order. But isn't Michael Young prone to hit into double plays? Isn't trying to get Andrus to second worth it, even given the cost?
Well, yes and no. Young's hit into six double plays this season, which leads the team -- but given his chances, his rate of hitting into double plays is not quite as bad as some have made it out to be. He's hit into three double plays with a runner on first this season, out of 37 PA; one with runners on first and second out of 12 PA; and one with runners on first and third out of 5 PA. (The last of his six double plays came with the bases loaded, which doesn't seem like an especially good time for the runner on first to try to steal.)
That's five double plays in 54 PA, or about 1 in every 10 chances. So let's say 10 percent of the time Young hits into a double play (which seems just slightly high given his career splits). Does that really justify a 40 percent caught-stealing clip from Andrus? When answering, keep in mind that Young's currently sporting a .368/.429/.658 line with a runner on first and less than two out. Yes, small sample size applies to those numbers (it's only 42 PA) -- but for his career, Young is a .314/.351/.499 hitter with a man on first and less than two out (1,110 PA). Those numbers simply don't come close to justifying the sort of base-stealing strategy the Rangers have been employing.
In some precincts, the Rangers' hijinks on the basepaths (which, of course, haven't been limited to stolen-base attempts) have been justified via comparison to the aggressiveness of their division rivals from southern California. But: in the last five years, the Angels haven't once logged an SB success rate lower than 70 percent, and have generally been in the 72-73 percent range. (This year, so far, Los Angeles is at 64 percent success, which is, of course, not good at all, and certainly not a mark Texas should be striving to equal.)
Again, aggressiveness is good. But aggressiveness on the base-stealing front has to be accompanied by a certain degree of unpredictability. As The Book notes, managing decisions in base-stealing (and in sacrifice bunt and many if not most other in-game) situations basically boils down to a matter of game theory. Preventing your opponent from successfully anticipating your strategy is quite literally a crucial component of those equations.
In robbing opposing teams blind on the base paths last year, the Rangers consistently kept opposing batteries off guard. But opponents' scouts, managers, and players sure seem to have caught on to Texas' tactics this season. It's no longer enough to go with the gut (if it ever really was). The Rangers need to make the necessary adjustments, and play the prevailing percentages. To do otherwise would be -- well, stupid.
And I'm really not so fond of stupid.