Estimating Baseball's All-Time Worm Burners
We at the Hardball Times like to track a lot of new age stats. Line drives, zone rating, groundball percentage - all those numbers allow us to evaluate today's players more accurately and with a breadth of information at our disposal.
But often, it's tough to understand what those numbers really mean because they have only become available over the past few years. So Derek Lowe allows groundballs on two-thirds of all balls in play, but what does that really mean? Is he the greatest groundball pitcher of all-time, or have there been better worm burners?
Unfortunately, due to lack of data, this is not a question we can answer directly. But I think I have come up with a method to do so indirectly.
My first step was to run a regression using regular statistics that are available throughout history and groundball percentage, which we have for the past few years. Three variables turned out to be highly significant:
The correlation between predicted groundball percentage and actual was very good, with an "r" of .47. With a weighted average of just 371 batters faced in the sample, what that means is that after four-to-five seasons, the correlation should be around .9, or almost perfect.
Again, I should point out that model is slightly biased towards pitchers who strikeout a lot of hitters, but beyond that, it is very solid. Among those for whom we do have data, the guys at the top of the predicted career groundball percentage list are indeed groundball pitchers, and the guys at the bottom indeed do allow a lot of fly balls.
So now for the fun: Who are the greatest groundball pitchers of all-time? To determine this, I first adjusted each statistic that goes into the predicted groundball formula for the league average. Because batters faced data is shoddy for the first 45 years of professional baseball, I only used data from 1916 on, so 19th and early-20th century guys are excluded. Actually, that's probably for the best, because I'm not sure how well the formula I devised would hold up in that time period.
In the chart that follows, I have listed two statistics: "Index," which is sort of like ERA+, except for groundballs, and "SD+," which measures how many standard deviations above average the player was for his career. I figure that the slightly more math-inclined may appreciate "SD+" a bit more than "Index."
First Last Debut Index SD+ Rube Foster 1913 1.14 4.01 Doug Sisk 1982 1.12 3.38 Ed Klepfer 1911 1.12 3.36 Aaron Cook 2002 1.10 2.96 Pat Clements 1985 1.09 2.59 Dale Murray 1974 1.09 2.50 Greg Minton 1975 1.08 2.45 Benny Frey 1929 1.08 2.41 Frank Linzy 1963 1.08 2.39 George Cunningham 1916 1.08 2.37
(Minimum 1,000 batted balls between 1916-2005.)
Ladies and gentlemen, Rube Foster is your winner! He allowed 14% more groundballs than the league average pitcher during his career (well, excluding its first three years, which actually means we're looking at just two years of data). In over 300 innings in 1916-17, Foster did not allow a home run, though admittedly, his team, the Boston Red Sox, allowed just 22 over those two years.
We actually do have some data for Doug Sisk, and it turns out that between 1980 and 1997, he was the fourth-most extreme groundball pitcher in baseball. That's a pretty big feather in our system's cap!
On the Ultimate Mets Database website, one commenter had this to say about Sisk: "The only man I know that would load the bases with no one on and one out just to throw a double play grounder." Reminds me of Derek Lowe...
We also have actual groundball data for Aaron Cook. Last year, Cook was third in the National League in groundball percentage; had he qualified, he would have been third in 2005 as well.
Let's now turn to the pitchers who have allowed the fewest groundballs in baseball history, on a percentage basis:
First Last Debut Index SD+ Jack Coombs 1906 0.87 -3.69 Bill Caudill 1979 0.88 -3.60 Wayne LaMaster 1937 0.88 -3.56 Al Jurisich 1944 0.88 -3.48 Allan Russell 1915 0.88 -3.47 Armando Benitez 1994 0.88 -3.35 Herb Score 1955 0.89 -3.22 Troy Percival 1995 0.89 -3.19 Luis DeLeon 1981 0.89 -3.18 Oliver Perez 2002 0.89 -3.18
(Minimum 1,000 batted balls between 1916-2005.)
Most of Jack Coombs' career came in years before our database starts, but in 1916-17, when Foster allowed a grand total of zero home runs, in fewer innings, Coombs allowed 10. Given that he played in a park that was neutral for home runs, Coombs seems like a pretty safe choice to be a fly ball pitcher.
For the guys for whom we have data, Perez posted a 30.1% groundball percentage last season, while Benitez allowed groundballs on 32.2% of all balls hit off him. Had they qualified, Perez and Benitez would have placed second and third from the bottom in groundball percentage last year, respectively.
If you're interested in seeing more new age statistics extended into the past, you can click here to download a spreadsheet with estimated groundball rates for all pitchers with 1,000 batted balls allowed between 1916 and 2005.