Johnny Cueto's left knee buckles. To the runner on first, it appears, if only for an instant, that Cueto is about to lift his front leg and deliver a pitch. Instead, it's Cueto's back foot that moves first, two lightning-fast steps rotate his body nearly 90 degrees, and he throws a bullet to first base across his body. This Cueto two-step is the single best defense against baserunning in baseball, more effective than Yadier Molina's fantastic arm and quick release, and capable of making stolen base attempts untenable for even the fastest runners in the game.
The rate of pickoffs, perhaps unsurprisingly, is closely tied to the intent to steal, and when calculating success rates, it makes sense to include them in our calculations. Since the beginning of 2011, Cueto has nabbed eight runners with his move to first base -- note that all pickoff numbers in this article do not include pickoff-caught stealing -- while another 13 have been caught stealing second. Only two runners have stolen second base successfully on him. In other words, runners have been successful in less than 9 percent of their attempted steals of second base on him.
John Lackey highlights the other end of the spectrum. Lackey hasn't picked off a runner from first base since 2003. Over the past three-plus years, 65 runners have stolen second base on him, while only nine have been caught, a success rate of 88 percent.
So how much of this difference between these two pitchers can be explained by their battery mates, the runners they've faced, the game situations they've pitched in and just plain ol' luck? Well, independent studies by Thomas M. Loughin and Jason L. Bargen, John Dewan and Max Weinstein agree that pitchers have considerably more control over the running game than catchers do. John Dewan estimated that pitchers have 65 percent of the control, with the other 35 percent going to the catchers. In fact, pitchers have more control over the runner's chance of success than the runner himself.
To identify the most effective pitchers over the past three-plus seasons and to see just how successful they've been, I followed in Loughin and Bargen's footsteps and built two mixed-effects logistic regression models. The first model predicts the probability of a stolen base attempt based on the pitcher, the catcher, the runner, the inning and the score as well as the lineup position of the batter. The second model predicts the chance of that stolen base succeeding based on the players involved. These models provide estimates of how much each runner, pitcher and catcher affects both the probability of a stolen base attempt and the chance of success.
More on the Cueto-Lackey divide
While Cueto has enjoyed having above-average catchers, according to our model, he is easily the dominant factor in his own success. Paired with an average catcher and pitted against a typical runner, Cueto cuts the typical chance of success -- 67 percent -- in half, to 33 percent. Lackey, in the same situation, increases the chance of success to 82 percent.
Steal probabilities
The best and worst pitchers in terms of estimated success rate of a typical runner with an average catcher (includes pickoffs).
Best SB success rate
Johnny Cueto 33%
Mark Buehrle 44%
Zack Greinke 44%
Clayton Kershaw 46%
Craig Breslow 48%
Worst SB success rate
John Lackey 82%
David Robertson 81%
Dale Thayer 81%
Drew Pomeranz 81%
Luke Gregerson 80%
To put Cueto's prowess -- and Lackey's weakness -- into perspective, the best catcher, A.J. Ellis, drops the usual 67 percent success rate to only 57 percent on his own, and the least effective catcher, John Jaso, merely brings the probability of success up to 72 percent.
The model also allows us to explore hypotheticals. For instance, a Cueto/Ellis battery would bring the chance of success down to a dismal 25 percent. Even when the most efficient runner in the game, Jarrod Dyson, attempts to steal against our imaginary Cueto/Ellis battery, he can be expected to succeed only 36 percent of the time. Meanwhile, an average runner would enjoy an 86 percent success rate against a Lackey/Jaso battery, and Dyson could be expected to succeed better than 91 percent of the time.
The ratings that our model assigns to baserunners line up closely with what we had expected based on their times to second base. According to Baseball Info Solutions, Dyson, Carlos Gomez and Jacoby Ellsbury had the fastest average stolen base times in baseball last year, and our statistical model places them first, second and eighth, respectively, in expected success rate after accounting for the batteries they've run against.
While Johnny Cueto is by far the most effective at catching runners, there are a number of pitchers who have been more effective at discouraging baserunning. In the most typical stolen base situation -- a runner on first with second and third base empty -- runners have attempted a steal 10.5 percent of the time in recent years. However, with Erik Bedard, Josh Tomlin or Bartolo Colon on the mound, our model estimates that the probability of a stolen base attempt drops to 4.1 percent. With Mike Adams, David Robertson or Lackey on the mound, it jumps to 25 percent.
Most obvious stolen base situations of 2014
According to our model, the most obvious stolen base situation of the year occurred in the top of the 11th inning in a May 3 game between the Dodgers and Marlins. After a Carl Crawford two-run home run, Dee Gordon beat out a ground ball to first base and found himself on first with Carlos Marmol on the mound and Jarrod Saltalamacchia behind the plate. Our model predicted an 87 percent chance of a stolen base attempt thanks primarily to the combination of Gordon, the fourth-most aggressive base stealer in the game, and Marmol, who struggles mightily holding runners on. To no one's surprise, Gordon took second on the first pitch.
The surest stolen base occurred on May 5 when Alcides Escobar stole second base against Dale Thayer and Yasmani Grandal. Thayer is the third-easiest pitcher to steal against in the game, Escobar is one of the most efficient runners and Grandal is the second-most ineffectual catcher following John Jaso. Our model estimated a 90 percent chance of success. With an excellent jump, Escobar took second base rather easily. To demonstrate just how confident he was, two batters later Escobar stole third with two outs.
The most foolhardy attempt of the year, according to our model, occurred on April 6, when Adam Eaton made his first stolen base attempt of the year against James Shields and Salvador Perez. Our model gave him a 40 percent chance of success and, indeed, Eaton was thrown out, if only barely, despite Perez needing to reach for a pitch at his toes.
If nothing else, please keep this information in mind when setting your daily fantasy lineup. If your team's speedster is facing Cueto, don't expect much.