I’d like to draw a few pictures of batted balls. If you’ve read the Hardball Times Annual 2006, you know what I’m talking about. If not, you can read the Batted Ball Leaderboards article posted a couple of weeks ago to get the gist.
Here’s a nutshell: we applied the relative run values found in Tom Ruane’s excellent Retrosheet article to the outcome of every batted ball over the last four years. That allowed us to assess the value of every batted ball type, explore team defense and analyze park factors, among other things, in the Annual. However, to perform some more definitive player analyses, I had to develop fully regressed batted-ball park factors. I finished that task last week.
So now I can tell you that Barry Bonds averaged .34 runs more than the average plate appearance for every outfield fly he hit, adjusted for ballpark. I can also tell you that every outfield fly Einar Diaz hit averaged .18 runs less than the average plate appearance, adjusted for ballpark. I can tell you this sort of thing for every major league player.
And now, because I am an insufferable graphing fool, I am going to show you some pictures of the player-specific, park-adjusted, batted-ball data. Ready? First up is something called a “box-whisker” graph. It shows the spread of the average run values for outfield flies, ground balls and line drives, by batter. I included all batters with at least 1,000 plate appearances over the last four years (a total of 300 batters).
The blue line in the middle of each red box represents the median relative run value of each type of batted ball (for this sample). The red box encompasses the quartiles above and below the median for each type of batted ball (so 50% of the players are contained in the red box), while the lines extending out from the red box encompass the top and bottom quartiles. For each set of data, there are outliers identified by red triangles.
As you can see, the run values of outfield flies have a much wider variance than ground balls or line drives. Much, much wider. There’s a little truism in this graph: For line drives and ground balls, the key is how often you hit them. For outfield flies, the key is how you hit them.
Like all truisms, this one shouldn’t be applied indiscriminately. Ichiro gets a lot of good value out of his ground balls, and Jason Phillips should avoid them like the plague. Frank Thomas and Russell Branyan certainly get a lot out of their line drives.
But the chasm between the best and worst flyball hitters is wide. An outfield fly by Bonds is as good as a line drive by the average major leaguer. On the other hand, an outfield fly by the aforementioned Diaz is just as bad as a Phillips grounder.
What makes a good flyball hitter? Swinging hard, at the risk of missing. To illustrate the point, here’s a graph of strikeout rates and outfield fly values for these batters. I’ve added a “fitted line” to the graph to highlight the relationship between the two.
The line tells the story. The cost of hitting outfield flies well is striking out more often (for most everyone not named Bonds, Pujols or Vladimir). I’ve labeled some of the players who deviate the most from the fitted line. Those farthest above the line are among the very best hitters in baseball. The ones farthest below it aren’t, including Diaz (who obviously just needs to strike out more!) and shortstop Ramon Vazquez. Branyan is a special case, by the way. His power shows up relatively more often in his line drives.
In case you’re wondering (and who isn’t?), the R-squared of the relationship between striking out and hitting valuable outfield flies is .40. If you remove Bonds from the sample, it’s .43. Maybe you weren’t wondering about that.
You know how some people say that strikeouts aren’t that bad for batters, but they’re great for pitchers? And you know how some other people say that makes no sense? Let me show you the basic difference between the two with another graph, a duplicate of the strikeout rate/outfield fly run value graph, but for pitchers instead of batters (same sample size).
Now scroll up to the batter graph. See the difference? For batters, the line goes up; for pitchers, the line goes down. Although the relationship isn’t as strong (an R-squared of .13), it’s there. Strikeout pitchers tend to give up lower-impact fly balls. Strikeout batters tend to hit higher-impact fly balls.
This is essentially the same conclusion J.C. Bradbury reached when he took “Another Look at DIPS” earlier this year. He found that strikeout pitchers tend to have favorable fielding stats, and he concluded that DIPS “works” because the formula includes strikeouts.
John Burnson, in the 2006 Graphical Pitcher concluded that strikeout pitchers also have a lower home run-per-fly ball rate. I didn’t reach the same conclusion, but I think that’s because I calculate home run rates per outfield flies. I did find that strikeout pitchers have a higher rate of infield flies per total fly balls (R-squared of .10), which would yield the same result as Burnson’s because infield flies don’t go over the outfield fence. Strikeout rates don’t have a measurable impact on the run values of line drives or ground balls, as far as I can tell.
In general, however, batters have a larger impact on the outcome of a batted ball than pitchers do. Following is a combined box-whisker graph, which includes the outcomes of each type of batted ball for both batters and pitchers. As you can see, the width of each box/whisker is wider for batters than pitchers (much wider for outfield flies), which means that pitchers don’t differ as much as batters in this regard.
For all three types of batted balls, the outcome has more to do with who’s batting than who’s pitching. This is the “DIPS insight,” if you will. But you also may notice that there are a few more outlier triangles for the pitchers. The outliers tell a story of their own, so let’s take a closer look at the pitching box-whisker graph, labeling the outliers.
The two outliers on the top of the outfield fly data are Brandon Webb and Jimmy Anderson, who also happen to be two of the most extreme groundball pitchers in the sample. Another truism is that when groundball pitchers yield outfield flies, they tend to have more run value (R-squared of .11 between groundball rates and outfield fly values).
The bottom two outliers are Francisco Cordero and Shigetoshi Hasegawa, two of the better relievers of the past four years. You may have also noticed that Mariano Rivera is the low outlier in line drive run values. Which leads me to my last, surprising truism: great relievers do have some demonstrable control over their batted balls. They are the “anti-DIPS.”
Consider this: Billy Wagner’s career Defense Efficiency Ratio is .750 by my calculations. Rivera’s is .745 if you exclude his first year. These are much higher figures than their respective team DERs, which implies that Wagner and Rivera’s batted balls were more catchable than average.
One could argue that we’re dealing with small sample sizes, but these two guys have pitched 630 and 740 innings in their careers, respectively. That’s equal to three-to-four full seasons of a major league starter.
Or consider this table of the pitchers with the lowest outfield fly relative run values in our sample. If you can look past Felix Rodriquez, you’ll see that it includes some of the best relievers of the last four years.
Last First OF Run Value Cordero Francisco -0.099 Hasegawa Shigetoshi -0.079 Rodriguez Felix -0.060 Nathan Joe -0.060 Foulke Keith -0.058 Rincon Juan -0.051 Marte Damaso -0.045 Wagner Billy -0.042 Hawkins LaTroy -0.041 Politte Cliff -0.040 Fultz Aaron -0.039 Walker Jamie -0.038 Smoltz John -0.037 Donnelly Brendan -0.037 Speier Justin -0.034 Benitez Armando -0.033 Stanton Mike -0.033 Quantrill Paul -0.032 Burnett A.J. -0.031
Why do top relievers tend to be anti-DIPS? Well, there is probably some selection bias in the data. Also, some relievers are saved for specific batter matchups, which undoubtedly helps. But many of the players on this list are closers or primary setup men; they typically enter a game regardless of who is batting.
My guess is that we’re also seeing something fundamental to the relief role. Pitchers who only pitch an inning or two throw harder and only use one or two pitches. The best relief pitchers, like Wagner and Rivera, leverage this situation to overpower batters, even when they hit the ball. Whatever it is that makes them best suited for relief also gives them an upper hand with outfield flies.
This suggests that, when you’re looking for top relievers, you might investigate not only their strikeout and walk rates, but their outfield fly run values too. Although there are some wild swings from year to year, the truly elite relievers will tend to be above average over the long term.
Here’s a review of some of the game’s other top relievers:
- Francisco Cordero is just dominant in this category. Over the past four years, his figures have been -0.138, -0.094, -0.097 and -0.082.
- If he had pitched enough to qualify, Chad Cordero would be third-best on the list of outfield fly run values, at -0.065 runs below the average plate appearance. It’s true that RFK kills flyballs, but these figures are adjusted for ballparks. In other words, both Corderos are keepers.
- Eric Gagne: 0.006 outfield fly run value. Good, but not Corderoesque. When healthy, Gagne is so dominant in the strike zone that it doesn’t matter.
- Jason Isringhausen: -0.022 for the last four years, but he’s been up significantly the last two years. He appears to be a risky proposition going forward.
- B.J. Ryan: Four-year average of 0.017, albeit -0.019 last year. Not exactly confidence-building.
- Brad Lidge: Outfield fly run values of -0.080 in 2003, 0.046 in 2004 and 0.035 in 2005. That’s not a good trend, as the Astros learned to their regret this past offseason.
- Huston Street registered -0.109 last year, which is Cordero territory. Derrick Turnbow, on the other hand, came in at 0.042, which isn’t encouraging. Turnbow shows signs of following the Rivera pattern, however. His line drive relative run value was only 0.206 last year.
References & Resources
After posting this article, one reader asked if I adjusted relievers’ outfield fly values by their strikeout rate. The answer is no, but every pitcher listed in the final table had outfield fly values lower than that predicted by their strikeout rate.