February 22, 2008
Prospectus Hit and Run
Having a Ball
Amid the ongoing swirl of steroid stories this past week, I came across a bit of research that had me dusting off something I wrote three years ago. In 2005, I contributed a chapter to Will Carroll's The Juice (which arrived more or less on the eve of baseball's day in front of Congress in 2005), which analyzed some alternative explanations to the theory that steroids had been responsible for the home run increases which typified baseball after 1992. I examined the effects that expansion, interleague play, the changing strike zone, and new ballparks may have had on the rising homer rates, and wound up concluding at the time that none of them were likely to have driven the surge.
Now the estimable Tom Tango has added some support for that viewpoint, at least with regards to parks and expansion. Comparing matched sets of head-to-head plate appearances between hitters and pitchers in the same park against all other pitcher/hitter/park combinations, Tango found virtually identical changes in home run frequency (HR per contact PA) from 1987 to 1988, and from 1992 to 1994. That is, both the matching combo and the unmatched combo saw their homer frequencies change at comparable rates during the same periods, first from 1987 to 1988, when a one-year home run spike came and went, and then from 1992 to 1994, a span in which homer and scoring rates escalated to levels that would be common over the next decade.
As I discussed recently, my Juice chapter pointed to equipment changes, namely the well-publicized changes in bat composition and the more under-the-radar changes in the ball's manufacturing process, as a more likely culprit. In particular, MLB-funded tests at the University of Massachusetts-Lowell identified a 49-foot gap between identically struck balls at either range of Major League Baseball's tolerance specifications, suggesting that juiced baseballs, not juiced sluggers, may have driven the increase. More recent revelations about a synthetic ring unaccounted for within the specifications provided additional support to a conclusion that was admittedly somewhat anecdotal. Here I should note that I was hardly the first person to suggest that the ball had been juiced; indeed, my own thinking on the matter was fueled by Eric Walker's essay entitled "The SillyBall: the Juiced Baseball." Walker has continued to follow the bouncing ball on his comprehensive Steroids and Baseball website.
In Tango's piece, he turns his attention to the ball as well, and to the UMass-Lowell testing in particular, focusing on testing director Dr. James Sherwood's report of an 8.7-foot difference in flight distance between tested major league balls and minor league ones, which differ in the compositions of their cores. Extrapolating from data provided by Greg Rybarcyzk of HitTracker Online, Tango finds that, lo and behold, an 8.7-foot decrease would reduce home run rates to almost exactly where they were in the decade prior to the surge. A tidy little explanation for where those extra long balls might have come from, right?
Not quite. Tango implies that what took place may have been as simple as MLB and Rawlings, the ball's current manufacturer, replacing balls made with a pure cork center (as specified for the minor league balls) with ones made with a compressed-cork center (a composite of cork and ground rubber, known as cushion cork or cushioned cork, which is part of MLB's official specifications for the ball). In actuality, the cushioned cork center ball is decades old: according to information provided by the Spalding company (which manufactured the balls up through 1976), it was officially adopted in the major leagues way back in 1926. Oddly enough, the words "cushioned cork center" imprinted on MLB balls were removed in 1999, the year before the UMass report was published, although the report notes that rubber continues to be added to the pill, the innermost element of the ball (emphasis added):
The cork comes from Maryland Cork, who in turn gets the cork from Spain and Portugal. The virgin cork from the supplier is ground to a fine consistency. Cork breaks down over time, and this breakdown implies the need for the rubber outer cushion.
Though rubber and cork are still in the pill, its exact composition appears to have changed over the past couple of decades. A team from Universal Medical Systems confirmed this last summer, when they compared computerized tomography (CT) scans of baseballs from different eras. Whether simply due to technological advances incorporated into the manufacturing process or a calculated desire to produce more home runs, the pill has increased in size and density over the years. And that's without considering the aforementioned synthetic ring, or the increasingly synthetic composition of the yarn used to wind the ball, something a University of Rhode Island study identified back in 2000. While Sherwood and company continue to test balls on an annual basis for MLB and have even shown some teeth by criticizing the outdated specifications of the testing, they've remained conspicuously quiet as to the impact of the composition changes, to say nothing of MLB bulldozing its own published specifications.
Don't Fence Me In
One more note about the home run increases as they relate to my back pages. It's generally taken as an article of faith that shorter fence distances due to the influx of new ballparks are partly responsible for this rise in homers. The problem is that the fence distances aren't actually getting shorter, either via new parks or changes to existing ones. I first examined this in The Juice, using data from Ballparks.com, team web sites, and Josh Leventhal's Take Me Out to the Ballpark (not all of which agreed). Using 1990--a year in which there were no new parks or dimension changes--as a baseline, here's what I found:
NL 1990 2004 Change LF 331.3 333.0 1.8 LCF 375.8 375.5 -0.3 CF 402.6 404.5 1.9 RCF 375.8 379.6 3.8 RF 331.0 332.5 1.5 AL 1990 2004 Change LF 327.1 328.7 1.6 LCF 378.1 377.9 -0.1 CF 406.1 403.3 -2.9 RCF 374.9 374.6 -0.3 RF 323.1 324.4 1.3
The distances down the foul lines actually got longer in both leagues, while the rest of the changes were a mixed bag. Updating the data through 2007 and combining the leagues for a composite "then and now" view:
MLB 1990 2007 Change LF 329.6 332.0 2.4 LCF 375.5 376.6 1.2 CF 404.9 404.9 -0.1 RCF 376.0 377.6 1.6 RF 329.1 329.3 0.2
With the exception of the teensiest of fractions for straightaway center field, fence distances have actually increased during the wave of building that's put 21 clubs (including four expansion teams) into new ballparks. What has decreased during the time period in question--indeed, what may be confusing the issue--is smaller park capacities. In 1990, the average ballpark held 53,057 patrons; last year it was 48,219, a drop of about 10 percent. So yes, parks are smaller, but not in a way that carries any ramifications for home run levels.
Tandem II: Harmonic Boogaloo
Two weeks ago, on the occasion of the Johan Santana trade, I made an attempt to quantify the best starting pitcher tandems in baseball history, using each pitcher's five-year peak (based on the five best WARP scores from his career). The results were a bit... half-baked, even for a list that was manually compiled. A few omissions quickly came to light, and I scurried back to my spreadsheet and found even more, many pertaining to either Hal Newhouser or pre-World War I big guns like Cy Young and Ed Walsh.
In updating the list, I decided to revise the methodology as well. I raised the cutoff for the lesser of the two pitchers from 40.0 to 45.0 WARP, thus putting them closer to Santana's five-year score of 48.3. Second, I took a page from Bill James (via his Power/Speed Number), and used the harmonic mean of their peaks to rank the pairings, providing a slight edge to ones where the two pitchers are closer together in score:
Rk Score Pitchers Team (Years) 1 64.1 Hal Newhouser (66.2), Bob Feller (62.1) CLE (1954-1955) 2 62.3 Walter Johnson (81.7), Stan Coveleski (50.3) WS1 (1925-1927) 3 60.5 Hal Newhouser (66.2), Bob Lemon (55.7) CLE (1954-1955) 4 59.3 Roger Clemens (61.8), Tom Seaver (57.0) BOS (1986) 5 58.9 Walter Johnson (81.7), Clark Griffith (46.0) WS1 (1912-1914) 6 58.7 Bob Feller (62.1), Bob Lemon (55.7) CLE (1946-1956) 7 58.6 Hal Newhouser (66.2), Dizzy Trout (52.5) DET (1939-1952) 8 58.3 Walter Johnson (81.7), Jesse Tannehill (45.3) WS1 (1908-1909) 9 57.9 Lefty Grove (61.6), Wes Ferrell (54.7) BOS (1934-1937) 10 57.5 Cy Young (74.5), Eddie Cicotte (46.8) BOS (1908) 11 56.8 Bob Gibson (57.4), Steve Carlton (56.2) SLN (1965-1971) 12 56.3 Cy Young (74.5), Jesse Tannehill (45.3) BOS (1904-1908) 13 56.1 Ed Walsh (70.1), Eddie Cicotte (46.8) CHA (1912-1916) 14 55.3 Ed Walsh (70.1), Red Faber (45.6) CHA (1914-1916) 15 55.2 Greg Maddux (63.6), Tom Glavine (48.7) ATL (1993-2002) 16 54.9 Christy Mathewson (69.6), Joe McGinnity (45.3) NY1 (1902-1908) 17 54.4 Robin Roberts (56.4), Fergie Jenkins (52.5) CHN (1966) 18 54.3 Dazzy Vance (56.5), Dizzy Dean (52.3) SLN (1933-1934) 19 54.0 Roger Clemens (61.8), Mike Mussina (48.0) NYA (2001-2003, 2007) 20 54.0 Steve Carlton (56.2), Phil Niekro (52.0) CLE (1987) 21 53.8 Warren Spahn (54.3), Juan Marichal (53.4) SFN (1965) 22 53.6 Hal Newhouser (66.2), Early Wynn (45.0) CLE (1954-1955) 23 53.5 Randy Johnson (56.7), Curt Schilling (50.6) ARI (2000-2003) 24 53.4 Warren Spahn (54.3), Gaylord Perry (52.5) SFN (1965) 25 53.1 Warren Spahn (54.3), Phil Niekro (52.0) MLN (1964) 26 52.9 Juan Marichal (53.4), Gaylord Perry (52.5) SFN (1962-1971) 27 52.6 Roger Clemens (61.8), Mariano Rivera (45.8) NYA (1999-2003, 2007) 28 52.5 Fergie Jenkins (52.5), Gaylord Perry (52.5) TEX (1975) 29 52.2 Steve Carlton (56.2), Bert Blyleven (48.8) MIN (1987-1988) 30 52.2 Bob Feller (62.1), Early Wynn (45.0) CLE (1949-1956) 31 51.9 Pedro Martinez (53.2), Curt Schilling (50.6) BOS (2004) 32 51.5 Robin Roberts (56.4), Jim Palmer (47.3) BAL (1965) 33 50.9 Pedro Martinez (53.2), Tom Glavine (48.7) NYN (2005-2007) 34 50.7 Randy Johnson (56.7), Mariano Rivera (45.8) NYA (2005-2006) 35 50.6 Pedro Martinez (53.2), Johan Santana (48.3) NYN (2008) 36 50.6 Gaylord Perry (52.5), Bert Blyleven (48.8) TEX (1976) 37 50.4 Tom Seaver (57), Nolan Ryan (45.2) NYN (1969-1971) 38 50.3 Phil Niekro (52), Tom Glavine (48.7) ATL (1987) 39 50.1 Dazzy Vance (56.5), Burleigh Grimes (45.0) BRO (1922-1926), SLN (1933-1934) 40 49.8 Bob Lemon (55.7), Early Wynn (45.0) CLE (1949-1957)
In all, seven of the top 14 rankings were missing from my previous attempt. Six of those date to the Deadball Era, when the job of pitching was significantly different from what it is today. Also missing from that upper echelon was the longest-running combo from this list, that of the Tigers' Newhouser and Trout. A couple of the other omissions were blindingly obvious (Martinez-Schilling and Johnson-Rivera) while the remainder were near-misses from the previous list that snuck back in (Perry-Blyleven, Niekro-Glavine, and Lemon-Wynn). Amazingly, Niekro makes the list in his first and final seasons, thereby bridging Braves history from Warren Spahn to Tom Glavine.
Meanwhile, it's clear that as good as Santana has been, he'll have to improve that peak score even to pass Pedro's other pairings. As noted before, there's a good chance of that happening, but if he can't top his 2003 WARP of 6.9 sooner or later, he'll be run out of the Big Apple like he was Kenny Rogers.