I was talking to a friend the other day who pointed out that, had Johnny Cueto not been knocked out in the first game, and had not Mike Leake been the Reds' uninspiring only option to replace him, the Giants probably wouldn’t have won the NLDS or, consequently, the World Series. That seems reasonable:

  • Cueto: 20.4 pitching VORP in 2012, best ERA+ in the National League
  • Leake: 2.9 pitching VORP in 2012, seventh-best ERA+ among Mikes

That describes most of what each pitcher does on the baseball field, but not quite all of it. There’s also the hitting, the fielding and, occasionally, the baserunning. Those are small things for pitchers. They rarely come into play. But they do come into play, and it’s how we get this:

Cueto still has the edge, and it’s not all that close, but it’s a lot closer, and if you believe in it it puts Leake in a different class of pitchers. By pitching VORP, he’s 281st in baseball, down with the relievers you thought retired: Vinnie Chulk, Chris Resop, Chad Gaudin. By WARP, incorporating all the things he does, he’s 53rd, tied with Tommy Milone and Colby Lewis. This doesn’t make him a great pitcher. But it makes him a lot different pitcher. The difference between Mike Leake, the best hitter/fielder/runner, and Tommy Hanson, the worst hitter/fielder/runner, was worth about $10 million in 2012.

Yet this is the sort of detail that (in my experience) rarely comes up when we talk about free agent pitchers, or when we evaluate trades. So, with Mike Leake as a starting point, the question is: Should we?

(Quick note: for the correlations below, we’re using as our sample the 53 pitchers who got at least 30 plate appearances in 2011 and in 2012.)
Spread from best to worst in 2012: about three runs
Equivalent to the value of: almost no measured skill in baseball is worth less potential value than pitcher baserunning. The spread between the best and worst catchers by non-throwing, non-passed ball fielding errors was about four runs in 2012, so that’s close. Pitcher baserunning is worth a bit less than the ability to not make non-throwing, non-passed ball fielding errors as a catcher.
Repeatability: The correlation between our pitchers’ 2011 and 2012 baserunning values was .006, so no correlation demonstrated.

Between playing in just one-fifth of their team’s games, batting roughly half as often in each game as the typical hitter, reaching base far less often than the typical hitter, and running with a cautious, station-to-station mindset, pitchers can’t stand out much on the bases. Among all NL pitchers last year, Stephen Strasburg was the worst baserunner (at 1.8 runs worse than average) and Clayton Richard was the best (at 1.1 runs). There are almost always going to be statistical outliers unlikely to be repeated, and it’s probably fluke more than skill that put Strasburg and Richard at the far ends of this list. As a comparison, the best baserunner in 2011 was Clayton Kershaw, at 0.7 runs, and the worst was Roy Halladay, at -2.5 runs. Halladay measured out as the better baserunner the next year.

Ninety-five of the 122 pitchers were within one run of each other. The spread between the very best and the very worst baserunners is so small that this one’s not really worth caring about, other than as an extra tenth of a win slipped into WARP.

Spread from best to worst: about eight runs
Equivalent to the value of: taking extra bases on teammates’ hits. The gap between Mike Trout, at the top of the hit-advancement leaderboard, and Prince Fielder, at the bottom, is 8.3 runs.
Repeatability: The correlation between our pitchers’ 2011 and 2012 fielding values is .30. By comparison, the correlation between our pitchers’ 2011 and 2012 pitching VORPs is 0.46.

It’s not surprising that there is a bigger spread here, and it’s not surprising there is more correlation, than in baserunning. R.A. Dickey, for instance, gets on base 10 times a year; he might have a half dozen opportunities to really make a difference, positively or negatively, as a runner. But he gets 60 or 70 chances a year as a defender, is involved in at least a handful of high-leverage fielding tries (on bunts, where the range of outcomes includes retire lead runner and start double play and throw the ball down the left-field line), and can swing plenty of innings by picking off runners. And so Dickey’s fielding stats don’t look randomly generated at all:

  • 2010: 6.3 runs, second in the NL
  • 2011: 8.3 runs, first in the NL
  • 2012: 4.7 runs, second in the NL

Of course, 60 chances is virtually nothing; even a year of fielding stats for a shortstop is suspect, and he’ll get more than 10 times as many. So there’s a high probability of flukiness here: Tim Hudson, for example, was worth -0.4 runs on defense in 2009, 7.6 runs in 2010, 3.2 in 2011, and -3.1 in 2012. But the demonstrated ability to shave two or three runs per year on defense—as with Dickey, Mark Buehrle, and Jake Westbrook—is worth putting in a free agent’s dossier.

Spread from best to worst: about 16 runs
Equivalent to: the spread between the third-best defensive first baseman in 2012 (Casey Kotchman) and the third-worst (Allen Craig). Or the spread between the best baserunner in 2012 (Michael Bourn) and the third-worst (Miguel Montero).
Repeatability: By batter VORP, it’s about the same as defense: 0.30. But by TAv, it drops all the way to .14, which suggests that a large portion of the correlation is playing time.

This is where we’d expect to see the biggest spread, and we do. But we don’t see all that much repetition in the leaderboards, which isn’t surprising given, once again, the nature of pitchers: they’re getting only a few dozen plate appearances; and a fair portion of those plate appearances are dedicated to sacrifice bunting (which counts, of course, but presents less opportunity for true talent to become evident). It’s not really worth paying extra money to a pitcher for his offense if you can’t count on his offense, and Mike Leake’s 12 BVORP in 2012 is a pretty good bet to regress in 2013. Not all the way to zero, necessarily, but to something less exciting. Here are the batter leaders from the past 10 years, and how they did the next year:

Year Pitcher Value Follow-up
2003 Russ Ortiz 13.5 2.5
2004 Livan Hernandez 15.4 12.4
2005 Jason Marquis 14.2 4
2006 LIvan Hernandez 9.4 8.7
2007 Micah Owings 16.9 8.1
2008 Carlos Zambrano 15.3 10.4
2009 Carlos Zambrano 10.4 5.2
2010 Yovani Gallardo 12.4 6.7
2011 Daniel Hudson 8.9 1.5
2012 Mike Leake 12

And the worst hitters in each season, and their follow-up performances:

Year Pitcher Value Follow-up
2003 Mark Redman -5.7 N/A
2004 Doug Davis -6 2.7
2005 Ben Sheets -3.9 -1.1
2006 Tony Armas -2.2 -1.1
2007 Brett Tomko -2.1 0.1
2008 Jason Bergmann -4 0.2
2009 Jose Contreras -2.3 N/A
2010 Rodrigo Lopez -2.2 -2.3
2011 Ted Lilly -3.6 1.5
2012 Aaron Harang -3.4 0.2

Which is just to note that, even if some pitchers can hit well enough and often enough to pay them for it, lots of regression is required.

So, taken as a whole:

Pitcher's Batter WARP (includes fielding, baserunning, batting)
Spread: About 1.9 wins
Equivalent to the value of: Carlos Beltran in 2012, according to WARP
Repeatability: A correlation of .35, less than the correlation of year-to-year pitching, but not irrelevant, either. Paying for some fraction of a pitcher’s batter WARP seems a reasonable decision on the free agent market, and paying a pitcher like Tommy Hanson a bit more to pitch in the AL, or a pitcher like Mike Leake a bit more to pitch in the NL, is a justifiable market decision.