Monday, August 21
If it weren't for bad luck ...
Special to ESPN.com
Editor's note: The team of writers from the Baseball Prospectus (tm) will be writing twice a week for ESPN.com. You can check out more of their work at their website at baseballprospectus.com.Between now and opening day 2001, Mike Mussina is going to sign a big multi-year contract, with the Orioles or some other team. And you can bet when he does, more than one newspaper columnist or radio talk show host will focus attention on Mussina's poor 2000 win-loss record, which currently stands at 7-13. Is it really worth $60 or $70 million, they'll ask, to get a sub-.500 starter? I always have the same response to questions like this: Pitchers don't win games, teams do. Sure, a starter exerts a lot of influence over his W-L record by preventing opponents from scoring (or not). But that influence still constitutes less than 50 percent of what determines a win or loss. The other 50-plus percent -- offense, bullpen support, and defense -- is just luck from the starter's perspective. For some starters, the luck will even out over the course of a season; but for many, it won't. But what if we could take a lot of the luck out of a starter's W-L record? That is, what if we could come up with a "fair" W-L record for a pitcher -- one that depends only on how he pitched, and not on the support he received from his offense and bullpen? A statistic we use at the Baseball Prospectus, called the "Support-Neutral W-L Record," attempts to do just that. This stat looks at a starter's season game-by-game and computes an Expected Win (EW) and Expected Loss (EL) total for him, given the number of innings he pitched, the number of runs he allowed, the number and position of runners he left on base for relievers to handle, and the pitcher- or hitter-friendliness of the parks he pitched in. It assumes that the starter has a league average offense and league average bullpen supporting him. I won't go into any more detail here, but you can find more by going to the Starting Pitcher Report at the Baseball Prospectus website. The stat is designed to be a good measure of overall starter effectiveness, but one of the fun uses of it is to see which starters have had the best and worst luck in the league. We measure luck by looking at the difference between his expected record and his actual record: (W - EW) + (EL - L). If that number is positive, he's been lucky; if it's negative, he's been unlucky. Here are the unluckiest starters this season, through Wednesday's games. The numbers reflect only their starts; any relief appearances are ignored.
Pitcher Team W L Pct EW EL EPct Luck Mike Mussina BAL 7 13 .350 11.7 8.3 .585 -9.4 Masato Yoshii COL 5 12 .263 8.5 6.7 .559 -8.8 Brad Radke MIN 8 13 .381 10.7 7.9 .575 -7.8 Steve Parris CIN 6 14 .300 7.8 8.1 .491 -7.7 Kris Benson PIT 8 10 .444 12.2 6.9 .639 -7.4 Pete Schourek BOS 2 9 .182 5.8 5.9 .496 -7.0 Joe Mays MIN 6 14 .300 7.5 9.2 .449 -6.3 Hideo Nomo DET 4 10 .286 8.2 8.0 .506 -6.2 Dave Mlicki DET 4 10 .286 5.4 6.3 .462 -5.1 Jason Johnson BAL 0 8 .000 3.2 6.1 .344 -5.1There's a wealth of underreported stories in this short table -- Kris Benson's emergence as an elite starter; the excellent seasons by Mussina, Brad Radke, and Masato Yoshii; and the continuation of Hideo Nomo's surprising 1999 comeback. They don't get the ink (or, in our case,the electrons) because the baseball world at large gives too much credence to assigning pitchers sole responsibility for W's and L's. A lot of baseball fans would argue that what I'm calling "luck" isn't luck at all. The guys in that table haven't won, the argument goes, because they don't "know how to win," or "pitch just well enough to lose," or can't "pitch to the score." I don't buy it for a second. Exhibit A for my side of the story is Mussina himself. Throughout the entire decade of the '90s he was one of the league's luckiest pitchers, with 17 more wins and 6 fewer losses than expected. Do you think that he suddenly forgot "how to win" this year? Or, do you think that his record this year has more to do with the geriatric Orioles' offense giving him the worst run support in the AL? I vote for the second explanation. And Mussina's far from the only example. Brad Radke and Pete Schourek are two other bad luck starters in 2000 who were among the league's luckiest in recent years. So how do this year's bad luck seasons compare to the all-time bad luck seasons? We don't have the game-by-game data needed to compute Support-Neutral records prior to the '80s, but we can generate an expected W-L record another way: by taking the pitcher's park-adjusted runs allowed totals and league-average scoring levels, and combining the two using Bill James's Pythagorean formula. Here are the unluckiest seasons of the post-war era, among pitchers who started at least half their games:
Pitcher Year Team W L Pct EW EL EPct Luck Turk Farrell 1962 HOU-N 10 20 .333 18.3 11.7 .609 -16.5 Nolan Ryan 1987 HOU-N 8 16 .333 15.4 8.6 .642 -14.8 Jim Abbott 1992 CAL-A 7 15 .318 14.4 7.6 .654 -14.8 Roger Craig 1963 NY-N 5 22 .185 12.4 14.6 .457 -14.7 Don Larsen 1954 BAL-A 3 21 .125 10.3 13.7 .427 -14.5 Lou Brissie 1950 PHI-A 7 19 .269 14.1 11.9 .544 -14.3 Jack Fisher 1965 NY-N 8 24 .250 15.1 16.9 .470 -14.1 Ned Garver 1950 STL-A 13 18 .419 19.8 11.2 .638 -13.6 Galen Cisco 1964 NY-N 6 19 .240 12.7 12.3 .508 -13.4 Steve Rogers 1976 MON-N 7 17 .292 13.6 10.4 .567 -13.2Nolan Ryan's infamous 1987, in which he finished with an 8-16 record despite winning the NL ERA and strikeout titles, is the season most frequently brought up during discussions of W-L bad luck. And this method confirms it was indeed among the unluckiest seasons of the last 50+ years. But Turk Farrell had an even more miserable summer in Houston 25 years earlier, and not just because he had to pitch outdoors. Thanks to the dismal offense of the first-year Houston Colt 45's, Farrell lost 20 games despite a 3.02 ERA. The Colt 45's were not the only '60s expansion team to treat starters badly. The mid-60s pre-Miracle Mets had no fewer than three of the 10 unluckiest post-war seasons. So a starter's luck during a given season can get pretty bad. But surely that luck is going to even out over the course of a long career, right? Not always. Using the same method that produced the last table, here are the unluckiest career W-L records since 1900:
Pitcher Years W L Pct EW EL EPct Luck Walter Johnson 1907-1927 417 279 .599 462 234 .663 -89 Bob Friend 1951-1966 197 230 .461 227 200 .531 -59 Milt Gaston 1924-1934 97 164 .372 126 135 .481 -57 Nap Rucker 1907-1916 134 134 .500 162 106 .606 -57 Raffensberger 1940-1954 119 154 .436 147 126 .540 -57 Ned Garver 1948-1961 129 157 .451 155 131 .542 -52 Bert Blyleven 1970-1992 287 250 .534 312 225 .582 -51 Buster Brown 1905-1913 51 103 .331 76 78 .491 -49 Eddie Smith 1936-1947 73 113 .392 97 89 .521 -48 Jack Powell 1900-1912 184 210 .467 208 186 .527 -47At first glance, it's hard to believe the second-winningest pitcher in history could have actually won 45 more games if he'd just gotten average support from his teammates. Hard to believe, that is, until you take a closer look at some of those Senators teams Walter Johnson played for. Johnson holds the all-time record for shutout losses, 1-0 losses, and 1-0 wins (by large margins in all three cases), as well as the AL record for shutout losses in a season (in 1909 he was shut out a whopping 10 times en route to a 13-25 record). Before you get any ideas, though, I'll mention that Cy Young also rates on the unlucky side of the ledger by this method, so there's no argument that the Big Train somehow deserves to be the all-time wins leader. The other name that really jumps out of this list is Bert Blyleven. If he had gotten league average support throughout his career, Blyleven could have cleared the 300-win barrier with lots of room to spare, and that surely would have made him a mortal lock for the Hall of Fame. Instead, he's left to wait for a phone call every January, hoping that somehow the BBWAA has managed to look past the W-L record this year (not that a 287-250 record is anything to sneeze at). Another starter who deserves to be in the 300-win club, according to this method, is Robin Roberts. His luck rating of -36 isn't enough to put him in the top 10, but it gives him an expected W-L record of 304-227, much better than his actual 286-245. At least with Roberts, his failure to reach 300 didn't prevent him from reaching the Hall. What the list above really shows how much a pitcher's legacy in baseball can hinge on the team he happens to hook up with. Most of the guys above were good or even great pitchers. We don't talk about them today because they happened to spend big chunks of their careers with poor offensive teams -- e.g., the Brooklyn Dodgers of the early 1900s (Rucker), the Cincinnati Reds of the late '40s and early '50s (Ken Raffensberger), and the St. Louis Browns of that same period (Garver). If Bob Friend had hooked up with the Yankees, we'd mention him today in the same breath as Whitey Ford. Instead, he hooked up with the weak-hitting Pirates, so all most of us can say is "Who?" Michael Wolverton writes for the Baseball Prospectus, the annual book by the same name, covering over 1500 players with in-depth statistical analysis and hard-hitting commentary. Michael may be reached at email@example.com.