The Cost of Bad Defense

In last week’s NLDS preview, in addition to lauding the Reds’ defense, and that of Jay Bruce in particular (boy did I feel like a moron), I wrote the following:

“Postseason lore is littered with defensive exploits, both Willie Mays-like and otherwise. But apart from those freak occurrences, fielding, over only a few games, doesn’t tend to swing a series”

Well, we had a big ol’ mess of those freak occurrences in the first round, not just in the Phillies-Reds series, but in the Atlanta-San Francisco series, where Brooks Conrad’s four errors may have swung the entire direction of the series, and in last night’s deciding Rays-Rangers game. These defensive gaffes–particularly in the two National League series–became one of the overriding story lines, and in concert with the tremendous pitching performances on both sides of both series, errors became one of the few ways the Reds, Phillies, Braves, and Giants could score.

To evaluate the impact of these errors, we’ll use Win Probability Added (WPA), a measure that determines, as the name might suggest, how individual plays affect a team’s chances of winning. This is based not on some statistical voodoo, but on data collected over the course of years that say that certain base-out states lead to certain average winning percentages. Since both teams start off with 50% win probability, if one player registers a WPA of .500, that ought to be enough to win the game, assuming his teammates’ contributions are neutral. This, of course, is not always the case. The single-game record for WPA by a player is 1.503 (enough, in theory, to win the game three times), by Art Shamsky of the Reds in 1966. Shamsky homered in each of his three at-bats, in situations of high enough leverage to generate a record WPA, and the Cincinnati still lost, 14-11.

Digressions aside, let’s look at three defensive culprits (or would-be culprits) from the divisional round: Chase Utley, Jay Bruce, and Brooks Conrad and see how much their missteps stood to cost their teams. One note before the break: WPA measures the difference between win probability before the play and after. So Chase Utley’s first error costs the Phillies 3.3% chance of winning compared to before the play happened, not compared to the state if he had made the play.

Chase Utley, 2B, Phillies, Game 2, NLDS

Utley’s two errors, because of their closeness and because the Phillies were already losing, looked worse than they actually were. The errors occurred early, the Reds managed to score only one run in the inning, and, of course, the Phillies came back to win the game, so no harm done. The total win probability cost of Utley’s two errors was less than 8%–not a trivial amount, but it could have been worse.

Jay Bruce, RF, and Brandon Phillips, 2B, Reds, Game 2, NLDS

This one, even in a game with six errors, was about as bad as a bad fielding sequence can get. Two errors on one play, resulting in a lead change in the late innings. In moments, the Phillies went from an underdog to more than a 4-to-1 favorite to win the game. What’s more, Scott Rolen’s slow release and decision to go to second base instead of first on the play before added another .073 WPA to the Phillies’ cause. From Ryan Howard’s strikeout in the 7th to Mike Sweeney’s pinch single—five batters—the Phillies started as a 2-to-1 underdog and ended as a 9-to-1 favorite. This play kept alive the rally that killed the Reds’ season.

Brooks Conrad, 2B, Braves, Game 3, NLDS

This play got even more press than the Jay Bruce’s Flying Circus two nights before, because Conrad, whose late-inning batting heroics had made him a folk hero in Atlanta, had been having serious defensive problems for more than a week, necessitating his move to second base, where he was fourth on the depth chart to start the season. The error was Conrad’s third of the night, and it would be his last—Bobby Cox yanked him for Game 4. But it fundamentally changed the Braves’ season. Up 2-1, with two outs in the ninth, Cox went to reliever Mike Dunn. Dunn allowed a game-tying single to Aubrey Huff that left runners on first and second (and itself took the Giants from a 14.4% win expectancy to 48.2%) before Moylan came in and allowed what should have been an inning-ending grounder to Conrad.

Even tied, with the sky falling and the winning run on second, the Braves still found themselves with a more-or-less even chance of winning because they still had the bottom of the ninth to score. Conrad’s error, however, put them in an impossible situation and just about ended the game.

Utley’s errors, because they came early and because Roy Oswalt was able to pitch around them, didn’t wind up costing the Phillies much of anything. Conrad’s gaffe, and the Bruce/Phillips double-whammy, came in the late innings and resulted in lead changes. Those are the kind of errors that can turn the tide of not only a single game, but as has been demonstrated, an entire season.