Thursday, February 22, 2007

2006 Reds Hitting Review - Addendum

This post makes a few additions/corrections to my 2006 Reds hitting review.

Clutch vs. Choke, Revisited


The goal of this part of the 2006 hitting review was to evaluate how the players' total offensive production, in terms of the total number of singles, homers, etc, they hit, compared to their game situation-specific impact on winning games as measured by Win Probability Added. Last night, as I was trying to fall asleep, I realized that the way I attempted to do that, via a comparison of VORP and WPA was flawed. The reason is that VORP corrects its offense estimates for position, such that players at low-offense positions will receive a higher runs estimate than players at high-offense positions. This is one of the great things about VORP, but in this case, since WPA doesn't make such an adjustment, VORP isn't directly analogous. Some players, like Brandon Phillips, have quite high VORP's due to their position, but their overall offensive production is fairly low compared to players like the similarly VORP'd Adam Dunn.

Therefore, instead of VORP, I'm re-doing this analysis using a simple version of Runs Created, RC = OBP * TB (THT's stats use a more complicated version of RC that includes information about hitting well with runners in scoring position--I didn't want to use that because I wanted any measurements of situation-specific impact in this analysis to be indicated solely by WPA). The only drawback of this approach is that Runs Created doesn't incorporate steals, but as we'll see, that would only worsen the situation for the relevant players.

Comparing Win Probability Added (WPA) to Runs Created gives this figure:
Trend line is the regression line between RC and WPA for Reds' players only. I can't easily get the data to calculate the league-wide regression line.

Observations:
  • As before, Edwin Encarnacion and Ken Griffey come out as looking very "clutch." Their WPA totals were far higher than would be predicted by their more classic "scorebook" statistics, which determine Runs Created.
  • Rich Aurilia, Dave Ross, and Chris Denorfia also come out looking good in this analysis, though with substantially smaller residuals than we saw with EdE and Griff.
  • Brandon Phillips and Austin Kearns still look to have not been very good in terms of converting their production into wins for the Reds, but they are now joined by Ryan Freel, Scott Hatteberg, and Jason LaRue as players who "choked" last season.
    • I expected to see Hatteberg move down below the line now that he's not getting downplayed by VORP, but I'm honestly surprised that Phillips is still rated as poorly as he is.
    • Nevertheless, I had a much warmer recollection of Hatteberg's performance in high-leverage situations than these data indicate. Looking back at my monthly reviews, however, he only had one strong month WPA-wise (July WPA = 0.84; he won my "statistical hitter of the month" award), and this seems to have been negated by poor showings in May (-0.35) and September (-0.56). In other months, he had close to zero net effect on games.
  • For all the moaning about Adam Dunn's performance in the "clutch," his contribution to wins seems to be right in line with his production. ... Of course, a hit like this won't hurt one's WPA totals...
Walks vs. Strikeouts

What follows could probably be its own post, but I thought it was a nice complement to the bb vs. k-rate analysis I did in the hitting review:

I was doing some reading this evening in the Hardball Times 2007 Annual and ran across an article by Dave Studenmund that reported, in terms of runs scored, the average effect of walks (0.355 runs) and strikeouts (-0.113 runs; for those interested, he also reported these figures for other batted-ball outcomes). Strikeouts are, not surprisingly, the most negative batted-ball outcome. But I was struck by how much larger the relative benefit of a walk could be. Therefore, I thought it might be fun to look at the direct impact of walks and strikeouts on Reds players:

Orange bars indicate total estimated runs contributed by walking, blue bars indicate runs given up by striking out. Black dots represent net gain in runs, calculated as the difference between the two bars.

Observations:
  • Despite giving up more than 20 runs via strikeouts over the season--which is very significant, and is worth ~2 wins--Dunn's incredible ability to draw a walk more than compensated for this, resulting in a net 17.8 runs scored via walks and strikeouts.
  • Hatteberg, of course, was amazing in this regard, with a high walk rate and yet the lowest strikeout totals of any Reds regular last season.
  • Freel's walk totals were third on the team last season. In contrast, his strikeout totals were second overall.
  • Despite having the lowest walk rate of any regular, Aurilia had the fifth highest net k/bb runs total because he doesn't strike out very often.
There's a way in which it's probably not that informative to directly contrast walks with strikeouts as I've done here, especially without reference to the other batted-ball outcomes. Nevertheless, it's a fun exercise, and I think indicates (once again) that while strikeouts are certainly bad things, their negative effect on the outcome of games is often exaggerated. Furthermore, the benefits of taking a walk are still underappreciated, at least by many fans and broadcasters.

2 comments: