The Problem of the Pick-off Throw

The first time I really thought about the pickoff move was in 1995.  My beloved and beleaguered Seattle Mariners had finally reached the postseason, and in the second game they faced a young, unspellable left-hander by the name of Andy Pettitte.  I was used to seeing lefties lob the ball to first, almost as a warning shot; Pettitte snapped the ball to first like a rubber band in a motion that looked like a cross between a balk and a dance move.  It struck me, an unbiased observer, as unfair and possibly inhuman.  Pettitte picked off two runners that game, and I found myself in unconscious awe.  What was he doing that made him so incredible?  Why wasn’t anyone else doing it?

The next day’s newspaper article made no mention of the two pickoff throws.  It’s hardly surprising, because there was plenty to talk about, especially Jim Leyritz’s game-winning, fifteenth-inning home run.  But it’s also not surprising because the pickoff only sort of exists.

Nobody likes the pickoff throw.  The fans detest it; I don’t know what the level of tolerance used to be, but at the game I attended last weekend, the crowd booed with every single toss to first.  The statisticians hardly bother to track it.  The analysts don’t care for it either, because of the way it hampers the rhythm of the ballgame and inserts dead air into the proceedings.  Opposing coaches gnash their teeth as weary hurlers cast the ball back and forth to the first baseman, buying time for a reliever to limber up.  The runners themselves can’t be too thrilled about having to dive back all the time, either.

From an aesthetic standpoint, however, I enjoy the pickoff.  I find the deception in the windup and the suddenness of the motion thrilling.  Added to this is the appeal of a battle of wills between baserunner and pitcher, who is already locked in combat with the batter at the plate.  It’s a combination of threats, the physical appearance of a pitcher slowly surrounded like a go piece thrust in atari, flailing back at his tormentors.

But beneath these surface considerations, something bothered me.  There is something fundamentally wrong with the pickoff throw, beyond its effect on the pace of the game.

One of the most beautiful aspects of baseball is its reliance on mixed strategy.  Mariano Rivera throws a decent cut fastball, but if he throws it for every single pitch, the batter will expect it and hit it more often.  If he throws too few, he’ll be sacrificing some opportunities to use his best pitch to get the batter out.  What results is a careful equilibrium that seeks to optimize the output of a player’s performance by adding enough variety to prevent the hitter from getting comfortable.  When the pitcher can’t do this, because his breaking ball isn’t working or he falls behind in the count, the hitter gains the advantage and his chance of success increases proportionately.

Not only must the pitcher (and the batter, guessing which pitches he is likely to see) optimize his arsenal, but he must randomize it.  Mariners fans of 2011 are well aware of Felix Hernandez’s past penchant for relying too heavily on the fastball early, leading to many first-inning struggles.  Randomization is not an easy thing; the human brain tends to work in patterns.  Unpredictability is necessary for gaining the upper hand.

The running game itself provides excitement in execution and its own mixed strategies, not just in the evaluation of a single game element, but in the overall strategy by which a general manager builds his team and searches for skills in his players.  For teams that lack firepower, the stolen base becomes a viable alternative for scoring runs.  Based on run-scoring environment of each era, the running game waxes and wanes in popularity.   Players with certain skill sets become under or overvalued, creating market inefficiencies and fostering creative ways to develop championship teams.  As a self-regulating system, it’s pretty amazing.

And that’s where the tragedy of the pickoff lies: it’s a dominant strategy.  From the perspective of winning ballgames, there is simply no reason why the pitcher shouldn’t continue to throw to first base ad infinitum whenever a runner steps off the bag.

Dan Malkiel at Baseball Reference undertook some painstaking and invaluable research regarding pickoffs, and the evidence is somewhat surprising.  To summarize his findings: The pickoff throw does not distract the pitcher and make it harder to throw strikes.  In fact, there is slightly more evidence that it is the hitter, not the pitcher, who loses concentration during multiple pickoff throws.  Nor does the pickoff actually deter the runner from running: because of the heavy correlation between multiple pickoff attempts and faster baserunners, we see higher steal rates after a runner is sent back to the bag a couple of times.

What we’re left with, then, are the two outcomes that change the state of the game: a successful pickoff, and an error.  Because an out is almost always worth more than a single base, it would take several times as many errors to create a risk worthy of deterring hopes of a pickoff, but the numbers lean the opposite way: a pickoff throw is three times as likely to result in an out as an error.  The pickoff is simply too dangerous a weapon.  You rarely see it succeed, but you see it succeed too often.

There is nothing in the rulebook that constrains a hypothetical continuous pickoff strategy, save for 9.01(d), which allows umpires to eject players for unsportsmanlike conduct.  Instead, the play is handled by baseball’s unwritten rules, which serve repercussions for such behavior in the secret underground bunker each Sunday evening.  Different proposals have been made: Bill James recommended reducing the number of “free” pickoff throws to two an inning, and charging a ball to the pitcher for each unsuccessful attempt thereafter.  The trouble with this lies in three-ball counts, where the mixed strategy will crumple to pieces.

There are two primary ways to alter the pickoff situation: to restrict them, or to make them less appealing as a strategy.  Most of the discussion centers on the first, but I find myself drawn to the latter: by balancing the strategy into a mixed one, with potential benefits and costs, the game not only speeds up, it becomes more interesting at the same time.  The way to do this is to alter the ratio of successful pickoffs to errors, either by lowering the first statistic or raising the latter.  On top of this, it would be helpful to do it in such a way that the runner’s leadoff isn’t allowed to expand, which might play with stolen base numbers.

The only ways to reduce pickoff success, without altering the length of a runner’s lead-off, would be to somehow make the pitcher’s throwing action more difficult, potentially by requiring an extra step.  This is troublesome, however.  The other option is to increase error rate.  This can be done by leaving the pitcher alone and instead making the play more difficult for the first baseman, by preventing him from camping at the bag.  Force him to run in from his regular position to perform a pickoff, and not only does the play become more exciting and demanding, but more errors are likely to occur.

Would it be enough?  We couldn’t know until we try.  But as a proposal it has a few virtues, not least of which being its subtlety.  A rule that proposes the first baseman move fifteen feet is more likely to find traction with the conservative baseball folk than one that creates new statistics, or creates a new type of walk.  I’d like to see it in action.  Not only would we get to keep the pickoff, but it might be a little more exciting.