Where have the double-plays gone?

The years 2005-2007 were something of a golden era for the twin killing in Arizona. Over the first seven seasons, the team averaged 109 double-plays per year, with a high of 120. However, likely triggered by the blossoming of  Brandon Webb as a ground-ball pitcher, there has been an explosion over recent seasons. We surged to a record 141 in 2005, and have come  close to matching that figure in both seasons since, notching 139 in 2006 and 138 last year.

The trend has come to an abrupt halt of late. So far in this season, through the first 44 games we have only 25, which would be on pace for just 92. This would be easily a franchise record low, beneath the previous worst of 97, set in 2002. And yet, as shoewizard recently pointed out, our defensive efficiency [the percentage of balls in play converted to outs] is excellent. Let's take a closer look, and see if we can find out what has happened to cause such a sharp reduction.

Stepping back to last year, we had 138 DP's in a total of 5,513 at-bats: one every forty, more or less. Perhaps surprisingly, Webb was below average, getting a double-play ever 44.2 at-bats; Doug Davis was the most impressive among the starters, with 23 in 750 at-bats, a rate of 32.6 . Here are the rates for the other pitchers who had 200 at-bats or more last season, and 60 or more this season.

A couple of numbers stand out there. Replacing Hernandez 2.0 with Dan Haren has resulted in a significant hit to the double-play rate. On the at-bats for Haren so far, Livan 2007 would have been expected to roll up six double-plays; Haren has only two. This is somewhat surprising, as I'd have thought his splitter was a great pitch to induce ground-balls - his career rate is 48.3, so based on that, we should expect some more from him going forward.

There are also a few "Inf", which means the pitchers concerned have yet to record a double-play this season. Micah Owings is the leader there, 187 at-bats this year, and 47 of those came with a man on first, but no double-plays have resulted. That's somewhat remarkable, given his hit trajectory shows 36% of plate-appearances result in a ground-ball. Between, Owings, Lyon and Johnson, there have been 416 at-bats, and the first ground-ball double-play for any of them only took place yesterday, when Johnson got one on the opening inning. While the Big Unit has never been reliant on such things [his career rate is over 60], Owings and Lyon had perfectly respectable GIDP rates last season, so their sudden phobia of them is surprising.

Let's also look at where those double-plays were hit, and compare the figures to last year, to see if there has been any change in the distribution. To do this, I looked at the distribution from 2007 - the positions listed below are the ones which started the groundball double-plays [I'm excluding line-drive or fly-ball double plays, but including bunt attempts]:

Based on the DPs turned last year, we'd expect to have 38 thus far, rather than the 25 actually received. Using the 2007 splits, we can divide the expected 38 up by position, to predict a number of double-plays for each position, and then compare that to the actual which have been started there:

This suggests that while there have been fewer almost all around the diamond, the double-plays which aren't being turned, are mostly missing from the left-hand side of the infield.

It's hard to say if this is just some kind of statistical quirk over the first quarter of the season. It may be that this is partly tied to Arizona's impressive DER, because a double-play presumably only counts as one for DER purposes. I note that Pittsburgh, who lead the majors in double-plays, have the worst DER in the NL. On the other hand, Cincinnati, second-worst there, have second-fewest double-plays too, so I am wary of drawing any conclusions in this area. Certainly, it's something worth monitoring as we go forward.