November 13, 2009
Prospectus Hit and Run
Digging the Long Ball
One of the six or seven factoids about the World Series which somehow escaped my notice when writing this site's epic preview was that the matchup between the Yankees and Phillies marked the first time since 1926 (Yankees vs. Cardinals) that the two teams who led their respective leagues in home runs faced off in the Fall Classic. The Phillies hit 224 homers and featured a quartet of hitters-Ryan Howard, Chase Utley, Jayson Werth, and Raul Ibaņez-who each hit at least 30, the 12th such combo in history. The Yankees swatted 244 homers and had seven players with at least 20 homers, the fourth team with such a widespread distribution of dingers.
Given those numbers, I had intended to set aside time to follow up my late-April look at home-run rates, particularly with Nu-Yankee Stadium's homer-promoting qualities the subject of so much discussion this year. Alas, it slipped through the cracks, but that only means we've got something else to warm our hands with as the Hot Stove season kicks off.
Back in April, amid so much breathless commentary from the mainstream media regarding rising homer rates, I noted that on a per-game basis, homers were up a whopping 20.8 percent relative to the previous year's comparable timeframe (the late start in the year made an April-to-April comparison less valuable), and 7.7 percent relative to overall 2008 rates. That trend leveled off, though as in years past, April's data foretold the overall direction of change. Looking at overall rates in the context of the long ball-happy Wild Card Era:
Year HR/G Change 1995 1.012 -2.1% 1996 1.094 8.2% 1997 1.024 -6.4% 1998 1.041 1.7% 1999 1.138 9.3% 2000 1.172 2.9% 2001 1.124 -4.1% 2002 1.043 -7.2% 2003 1.071 2.8% 2004 1.123 4.8% 2005 1.032 -8.1% 2006 1.109 7.4% 2007 1.020 -8.0% 2008 1.005 -1.5% 2009 1.037 3.3%
I've expressed the per-game rate as per team per game, a preference that jibes with our tendency to talk of run-scoring environments in terms of a single team (as in normalizing statistics to a 4.5 RPG environment). Overall, home runs increased by 3.3 percent over 2008, a rather ho-hum change that's notable primarily because 2008 marked the lowest rate of the era, with 2007 the being second-lowest. Excluding the shortened 1995 season, this year's rate ranks 10th in the current era.
As it turns out, that modest increase was actually the result of sharply divergent trends in the two leagues:
Year NL change AL Change Gap 1995 0.952 -0.2% 1.071 -3.7% 0.119 1996 0.979 2.8% 1.210 13.0% 0.231 1997 0.954 -2.6% 1.094 -9.6% 0.140 1998 0.988 3.6% 1.102 0.7% 0.114 1999 1.117 13.0% 1.163 5.6% 0.046 2000 1.159 3.8% 1.187 2.0% 0.028 2001 1.139 -1.7% 1.106 -6.8% -0.033 2002 1.003 -12.0% 1.088 -1.6% 0.085 2003 1.046 4.3% 1.101 1.2% 0.055 2004 1.099 5.1% 1.150 4.4% 0.051 2005 0.995 -9.5% 1.075 -6.5% 0.080 2006 1.097 10.3% 1.123 4.5% 0.026 2007 1.043 -4.9% 0.993 -11.6% -0.050 2008 1.008 -3.4% 1.001 0.8% -0.007 2009 0.958 -4.9% 1.128 12.7% 0.170
In the National League, home runs declined for the third straight year, falling to levels not seen since the late 1990s, and falling back below American League rates for the first time since 2006, though lower rates in the Senior Circuit are actually the norm. In the AL, home runs rose sharply to their highest level since 2004, that via the sharpest increase in per-game rates since 1996. The gap between the two leagues, expressed above as the AL's advantage, was the widest seen since 1996, after several years of the two leagues tracking fairly closely:
Of course, there's a fairly obvious reason for the current split. You may have heard somewhere that 2009 featured the introduction of two new ballparks in the Big Apple, two parks that had very different effects on home-run rates. An MLB-high 237 homers were hit at Nu-Yankee Stadium, 1.463 per team per game-a staggering rate, but nonetheless a far cry from the 2.0 per game in April which prompted meteorologists to chip in their two cents about wind tunnels. Meanwhile, just 130 homers were hit at Citi Field, the majors' sixth-lowest total. Isolating the impacts of the new parks and their predecessors from their leagues:
Split 2009 2008 Change MLB 1.037 1.005 3.3% MLB - NYs 1.031 1.003 2.8% AL 1.121 0.997 12.4% AL - NuYank 1.094 0.998 9.7% NL 0.964 1.011 -4.6% NL - CitiF 0.975 1.007 -3.1%
Note that the league figures here differ from the ones above because they're tallied by ballpark, not by team offense. Controlling for the new venues, we see that homers in AL parks increased by almost 10 percent in 2009, while those in NL parks fell by just over three percent, which means that the split is more than just an effect of the two new parks. As it turns out, those parks didn't even produce the largest year-to-year changes at either end of the spectrum:
Split 2009 2008 Change Yankees 1.463 0.988 48.1% Rangers 1.327 1.259 5.4% Phillies 1.278 1.167 9.5% Orioles 1.265 1.275 -0.8% Brewers 1.241 1.084 14.4% Angels 1.198 0.951 26.0% Blue Jays 1.198 0.772 55.2% White Sox 1.185 1.378 -14.0% Reds 1.154 1.321 -12.6% Twins 1.152 0.852 35.3% Red Sox 1.148 0.907 26.5% Rays 1.142 0.987 15.7% Tigers 1.105 1.247 -11.4% D'backs 1.074 0.975 10.1% Rockies 1.062 1.074 -1.1% Marlins 1.043 1.049 -0.6% Astros 1.012 1.218 -16.9% Nationals 1.000 0.925 8.1% Cubs 1.000 1.148 -12.9% Mariners 0.963 0.833 15.6% Pirates 0.883 0.944 -6.5% A's 0.864 0.791 9.2% Indians 0.852 0.951 -10.4% Royals 0.827 0.759 8.9% Mets 0.802 1.074 -25.3% Giants 0.796 0.741 7.5% Padres 0.796 0.840 -5.1% Dodgers 0.784 0.741 5.8% Braves 0.765 0.889 -13.9% Cardinals 0.741 0.994 -25.5%
Home runs in the Blue Jays' Rogers Centre domicile saw the majors' sharpest increase, as a team that lacked a player with more than 20 homers in 2008 suddenly found itself with two who had at least 35 in Aaron Hill and Adam Lind, neither of whom spent the full 2008 campaign with the Jays. Meanwhile, homers in Busch Stadium III fell by a slightly larger margin than they did in Queens-one homer on either side of the ledger would have been enough to turn things. That drop had much to do with a Cardinals staff which, thanks to the return of Chris Carpenter and the development of Joel Pineiro's sinker, had the majors' highest ground-ball rate and yielded just 54 homers at home. The team also lost nearly 40 homers from 2008 to 2009 via injuries and falloffs from Troy Glaus, Rick Ankiel, and Ryan Ludwick. Both extremes serve to remind how volatile year-to-year home-run rates can be from park to park, as personnel changes, weather, and the kind of good ol' randomness inherent in even an 81-game sample can have their effects. That's why useful park factors rely upon multiple years of data.
Excluding the New York parks, the correlation between 2008 and 2009 per game rates is just 0.59, and the standard deviation of the percentage changes is 17.2 percent. That's fairly typical of recent years; the correlation from 2007 to 2008, excluding the new Nationals Park and its predecessor, RFK Stadium, was just 0.45, and the standard deviation of the annual changes was 16.8 percent.
Also worth revisiting are the distance numbers from Hit Tracker, which reports both True Distance (actual home-run distance) and Standard Distance (normalized to remove the influence of wind, temperature, and altitude). Alas, this is a bit more slippery, since the data for previous years that's currently available at the site differs from what I reported in April. Hit Tracker site domo Greg Rybarczyk told me that after analyzing data provided by Sportsvision (the PITCHf/x and HITf/x folks), he made subtle revisions to his aerodynamic model, incorporating spin modeling and lift coefficients, then went back and reanalyzed his old observations, resulting in slightly different figures. Here are the revised ones:
Year Avg True Dis Change Avg St Dis Change 2006 399.1 N/A 395.2 N/A 2007 397.3 -0.45% 394.9 -0.08% 2008 397.1 -0.03% 393.8 -0.28% 2009 Apr 399.5 N/A 396.5 N/A 2009 Full 398.8 0.43% 395.7 0.48%
As with overall home-run rates, True and Standard Distances showed a year-to-year increase for the first time during the period of Hit Tracker coverage, doing so by nearly two feet in both categories. Even so, those final numbers fell off considerably from April, when the increases over the previous year were closer to 2
What's also at least somewhat surprising is that we didn't see an increase in what Rybarczyk calls "Just Enough" homers, balls which clear the fence by less than 10 vertical feet or landed less than one fence height beyond the fence. Instead, we saw a considerable increase in "No Doubt" homers, balls that cleared by at least 20 vertical feet and landed at least 50 feet beyond the fence.
Type 2009 2008 2007 2006 Just Enough 31.2% 31.4% 33.0% 30.7% Plenty 48.2% 50.4% 46.4% 48.6% No Doubt 20.3% 16.8% 16.5% 19.6% Inside Park 0.2% 0.2% 0.3% 0.2% Unknown 0.1% 1.2% 3.8% 0.8%
In other words, it appears that the increase in homers is from an uptick in balls that were crushed, not ones that barely cleared-consistent with the finding that the new parks really didn't have all that much to do with the annual changes.
Which, if we're trying to pin the increase on something, basically leaves us in the realm of speculation about juiced players (increasingly unlikely in this age not only of testing, but expanded powers of investigation; see Ramirez, Manny) and juiced balls. While the 'No Doubt' spike does pique my curiosity a bit, I'll avoid rehashing the ball-based explanations I've offered in the past and note that it's significantly more fun to see advances in the science of measuring what's happening than it is to speculate about steroids or MLB conspiracies. Perhaps a year from now or even sooner, we'll be able to look back on HITf/x data and match the varying fly-ball distances in each park with the changes in home run rates, expanding our understanding of the game by, oh, another 400 feet or so.
Special thanks to Greg Rybarczyk of Hit Tracker for his generous data assistance.