- FanGraphs Baseball - http://www.fangraphs.com/blogs -

Why Was Allen Craig Running?

Hidden among all the intentional walks and sacrifice bunts during the tic-tac-toe game that was game five of the World Series, there were two caught stealings that may have turned the game. In both cases, Allen Craig was gunned down during an Albert Pujols plate appearance. Before the face palm napalm dropped, there was a question hidden in the initial stunned silence, heard all the way up into the booth:

Why was Allen Craig running?

In the top of the seventh, in a tie game, Craig was at first and Pujols was facing Alexi Ogando. On a one-and-one count, Craig took off for second base. This is a man who stole all five of his attempts in the regular season, but it’s also a man that never stole more than eight in a full minor league season. Plus, he was stealing in front of Pujols. There was a chance that, even if he stole the base, the Cardinals’ reward would have been the same intentional walk that Ogando issued once Craig was caught stealing. Open base and all.

But the idea seemed misguided from a win probability standpoint, too. In the top of the seventh, in tied road game, the Cardinals had a win expectancy of 50.3%. If he steals the base, they have a 53.2% win expectancy. The caught stealing dropped their win expectancy down to 42.9%.

The break-even point for that stolen base was 71.8%. What with Craig’s 19 combined stolen bases in over 2500 career professional plate appearances, it seems like folly to expect that high of a success ratio. The average baseball player managed a 72.2% success rate, but in 119 career games in the major leagues, Craig’s Bill James speed score is 3.6, and the average number in that category is 5.0. It just doesn’t make a lot of sense for a player with below-average wheels.

Pujols only struck out in 8.9% of his at-bats this year (9.5% career). Once the ball is in play (about 80% of the time for Pujols), batters hit into double plays about 11% of the time in your typical double-play situations. At most, there was a 20% chance that the Cardinals ended up in the two worst possible situations if they don’t attempt the steal. Managing to avoid that double play seems to be a strategy that focuses too much on a slight negative.

Which segues us nicely to the second Craig caught stealing. In the top of the ninth, with the team down by two, Craig at first, and Pujols at the plate and facing a full count, Tony La Russa opted to send the runner. This one may have made more sense.

Back to that 11% chance of hitting into a double play. This statistic assumes that there is a double play opportunity and that the ball is put into play. With the count full, this all comes down to one pitch. Pujols has a 86.5% contact rate, and so the chance of a double play was somewhere around 9.5%. If someone like Matt Holliday was up there (77.7% career contact rate), that percentage drops to 8.5%. Mike Napoli (71.1% career) would have had a 7.8% chance.

But Pujols is Pujols and he’s likely to make contact on a pitch that needs to be in the zone. On a pitch in the zone, he was 92.2% likely to make contact (and 10.1% likely to hit into a double play). So maybe you send the runner to avoid a 10% chance at the super-bad situation. The strike-em-out-throw-em-out double play was only about 2.5% likely given the players involved.

Still, those were two caught stealings that made a nation scratch their collective head. One may have been worse than the other, but viewers will remember them in tandem. Add in the fact that this was the World Series and you can add insult to injury: Not since Billy Martin in 1955 has a player been caught stealing twice in a World Series game.