March 4, 2011
The Impact of World Series Starts (or How Much Was Jack Morris Really Worth?)
Believe it or not, most of our writers didn't enter the world sporting an @baseballprospectus.com address; with a few exceptions, they started out somewhere else. In an effort to up your reading pleasure while tipping our caps to some of the most illuminating work being done elsewhere on the internet, we'll be yielding the stage once a week to the best and brightest baseball writers, researchers and thinkers from outside of the BP umbrella. If you'd like to nominate a guest contributor (including yourself), please drop us a line.
The date is October 27, 1991. Brian Hunter stands in against the Twins' Jack Morris. David Justice leads off second. It's the second inning of a tie game, with one out. But this is no ordinary tie game—this is Game Seven of the World Series. Hunter swings and misses, striking out. Greg Olson lifts a popup to the second baseman, ending the inning. The Braves remain scoreless.
There are some people who think that a shutout in the World Series is no more meaningful than a shutout thrown in the middle of May. Such people are a minority of baseball fans, though a larger share of the analytical community. Others seek to quantify the impact of playing one's best in the most important situations. A statistics such as Leverage Index, developed by Tom Tango, is an appropriate tool for such purposes. A home run in the ninth inning to tie a ballgame means more than a home run with a 12-to-1 lead. Pitching a scoreless ninth to protect a 4-3 lead, as Mariano Rivera does so often, has more impact than a scoreless fourth inning when your team is down 7-0.
Win Probability Added (WPA) measures how each event in a baseball game increases or decreases a given team's chance of winning the game. This statistic, available in the game logs at Baseball-Reference.com, looks at the probability of each event without regard to what will happen in the future. As an example, a pitcher enters the top of the eighth inning of a tie game with the bases loaded and nobody out. He is pitching in an extremely high-leverage situation; if he allows even one run to score, his team is more likely to lose than win, and most teams in that situation allow multiple runs. He strikes out the side, and at the end of the inning his team is more likely to win than lose. He has advanced his team's win probability by a significant amount. If his team then scores 15 runs in the bottom of the eighth for an easy win, then in retrospect his three strikeouts were not very important. WPA, however, does not know the future, and credits him for what his performance looked like at the moment it happened.
Back in October '91, a single, a passed ball, and a walk put runners on first and second with one out in the third inning. Morris gets Terry Pendleton to fly out to left, and Ron Gant grounds out to shortstop to end the inning. The shutout is intact.
WPA and Leverage Index help to quantify the impact of events on the individual game level—in other words, what each play means to the team's chances of winning that game alone. To go a step beyond that, we can attempt to quantify the game's impact on the series. In a simple model, I assume that each team has a 50 percent chance of winning each game that it enters. At the start of game one, each team has a 50 percent chance of winning the World Series. After that game, the winning team needs to win only three more games, and its chance of doing so is 66 percent. The losing team needs to win four to recover, and the odds of that are only 34 percent. Winning game four of a series where you already lead 3-0 increases your chance of winning the series from 94 to 100 percent. Winning game seven takes your chances from 50 all the way to 100 percent. Winning game seven has eight times the impact on winning the series as finishing out game four of a sweep—the team that fails to win game seven has lost the series, but the team that fails to win game four has three more chances to emerge victorious. By combining the WPA with the series advancement percentage, we can measure each pitcher's contribution to winning the championship.
In the fifth inning, the Braves give themselves an even better chance to score. After a leadoff single, Mark Lemke moves to second on a sacrifice bunt and then to third on a bunt single. Morris strands the runners again, with Pendleton popping up to short and Gant striking out.
Jack Morris does not compare well, by most of the stats used today, to a group of pitchers who have received little support for induction into Cooperstown. Here's how he compares to David Cone, Orel Hershiser, Bret Saberhagen, and Kevin Brown by WAR (from Baseballprojection.com), WARP (from Baseballprospectus.com), ERA+ (from Baseball-Reference.com), and overall postseason performance:
Morris ranks first by pitcher wins, but his victories were more dependent on his teammates than his own performance. That's difficult to argue against, although many have tried. Morris, pitching for American League teams before interleague play, had one regular-season at-bat in his career, which means that he had nothing to do with the 50 percent of the game that comes down to offense. If fielding counts for even one percent of the game (and most people think it's a lot more than that), then Morris was responsible for less than 50 percent of the outcomes of his games. By ERA+, which mostly assesses pitching but includes runs saved by fielders as well, Morris was the least effective of these five pitchers, although he did pitch more innings. WAR and WARP balance the rate (ERA+) and quantity of his pitching, and he again ranks last among the five. His postseason numbers are better than Brown's and Saberhagen's, but well short of what Hershiser did and a bit behind Cone, considering the inning advantage. Yet while all of the other pitchers received minimal support for the Hall of Fame, Morris has climbed over the 50 percent mark and still has a shot at induction in the next few seasons. If this makes any sense at all (and I'm sure many of you think it doesn't), it can only stem from the timing of his accomplishments.
The score is still tied as we head to the eighth, when Morris will face his greatest test. Lonnie Smith leads off with a single. Terry Pendleton doubles to left center. In a play that elicits curses from Braves fans to this day, Smith is unsure if the ball will be caught and fails to score, even though he's running on the play. He's held at third, but with nobody out retains an excellent opportunity to score. Morris does not surrender. He gets Ron Gant out on a grounder to first, and Smith holds at third base. After intentionally walking David Justice, Morris gets the speedy Sid Bream (I'm just going on what Barry Bonds told me) to ground to first. Kent Hrbek throws home for the force out, back to first for a double play, and the Braves do not threaten again as Morris retires them in order in each of the next two innings. Minnesota finally wins the game in the bottom of the 10th.
I've combined the WPA and the series advancement index for all World Series starting pitchers. Jack Morris recorded the perfect storm, a great pitching performance in the tightest of situations in the most important game of the year. Had he allowed a single run over the ten innings, the Twins would have lost the 1991 World Series. How does his performance stack up to other pitchers throughout baseball history? Have there been others who pitched equally important games? How about his failures one year later? Morris pitched poorly and lost two games in the 1992 World Series. Although his team won anyway, he certainly decreased its chances of doing so. How does he rank when his failures are considered along with his successes?
Here are the starts with the greatest game impact in World Series history, ranked by WPA:
Something must be wrong with my code. This isn't a greatest offensive player list... Oh, right, Babe Ruth was a pretty good pitcher, too, before he started hitting home runs full-time. His 1916 start goes down as the greatest game impact pitching performance in World Series history. He allowed a first-inning, two-out, inside-the-park homer to Hi Myers, but that would be it for the day. In the third inning, Ruth's groundout scored Everett Scott to tie the game (he was hitless, not one of Ruth's better hitting days), and then he held out until his team scored the winning run in the 14th.
Art Nehf is the real find on the list, the Jack Morris of the 1920s. He makes the list twice, once at number two and once just behind Morris's fifth place. Nehf pitched 15 seasons with a 184-120 record and matches Morris with a 105 ERA+. In 1921 he beat the Yankees with a 1-0 complete game victory in Game Eight (yes, they used to play best-of-nine series back then). Two years later, in Game Three against the Yankees, he beat them 1-0 all over again.
Considering series impact combined with WPA, the World Series starts with the greatest impact are as follows:
All but two are Game Sevens, with Nehf's Game Eight (there has never been a Game Nine) and Clem Labine's dominant Game Six making the list.
Two years earlier Ralph Terry was the goat in Game Seven, coming on in relief in time to serve up perhaps the most famous home run of all time, Bill Mazeroski's game-winner. In 1962 he redeemed himself, pitching a 1-0 shutout against the Giants. Terry made it interesting, though: in the ninth inning, a single by Matty Alou and a double by Willie Mays put the tying and winning runs on with Willie McCovey at the plate. McCovey lined to second for the final out.
Steve Blass is more famous now for his complete loss of the ability to throw strikes, but he had no such trouble in 1971. He did not allow a run until the eighth, when he had a 2-0 lead. He didn't allow a second run and pitched the Pirates to the trophy. Johnny Podres pitched the Dodgers to a 2-0 win over the Yankees in 1955; after seven losses, this was the first and only World Series win for the Brooklyn Dodgers. Of course, the Dodgers won a few more after they moved out west—In 1965 they beat the Twins 2-0 in Game Seven, and Sandy Koufax had perhaps the most dominant performance of anyone on this list, striking out ten and pitching a three-hitter.
The next table shows the career impact:
In addition to his three starts in 1991, Jack Morris made two excellent starts in the 1984 series. His positive accomplishments outweigh his 1992 losses, and only Art Nehf had a bigger impact on winning World Series titles. The WAR equivalent column shows how many wins above replacement a pitcher would have to contribute, over the course of his career, to have the same impact on winning the World Series. To get this number I came up with a formula to estimate how many additional playoff appearances a team should expect given a career of a certain level of WAR. Then, I assumed that a playoff appearance gives a team a one-in-eight chance of winning the series. I applied the same formula to everyone for comparative purposes, although prior to the current eight-team playoff system the calculations would have been a bit different.
To estimate a pennants added formula in the eight-team playoff era, I looked at how often a regular player (500+ at-bats) with a given level of WAR ended up in the playoffs. For example, there were 279 players with less than 0.5 WAR, and these players made it to the postseason in 17.2 percent of the seasons. There were 285 players between 3.5 and 4.5 WAR (average of 4.0), and they made the playoffs in 37.5 percent of the seasons. The data appear to be linear, implying that a great player at 8.0 WAR helps you just as much as two good players at 4.0 WAR each. (This is consistent with David Gassko's study a few years ago at The Hardball Times). The linear fit for pennants added is a little different than what Bill James found in The Politics of Glory or what Michael Wolverton did nine years ago here at Baseball Prospectus. That's okay—the expanded playoffs of post-1994 baseball have been a game-changer, and the earlier non-linear models used by James and Wolverton worked for the seasons they were looking at. The formula that fits the equation is simply playoffs added = WAR * .051, or World Series added = WAR * .0064.
For Morris, Game Seven of 1991 alone is the equivalent of 65 WAR—roughly the career of Kevin Brown. Morris's career World Series impact is worth almost 90 WAR—roughly the equivalent of Bert Blyleven. This does not mean Morris is better, or even as good as Blyleven or Brown. What it means is this—if you had to choose between the entire regular-season career of Kevin Brown and October 27th, 1991 from Morris, the two choices would give you an equal number of expected World Series rings. Brown may or may not have his best years when your supporting cast is good enough to compete for a title, and even when you get into the playoffs you've got only a one-in-eight chance of coming out on top.
While Morris certainly rose to the occasion, the opportunity presented to him was beyond his control. Not every pitcher, no matter how great, gets to pitch a seventh game in the World Series. Even of those who did, few have the opportunity to put up a +.845 in WPA as Morris did. He can thank John Smoltz (18th-greatest in series impact for that game) for that. Bret Saberhagen threw a shutout in Game Seven of the 1985 World Series, but his WPA was only +.203 since the Royals blew out the Cardinals. Unlike Morris's Twins, the Royals would have won with any ordinary, non-disastrous pitching effort. Morris combined the luck of finding himself in the perfect situation with a great performance and a little bit of good fortune to pull it off (such as the Lonnie Smith play).
Here's the question—is the pursuit of a ring the only thing that matters? Or does the regular season have value in itself? If you treat the regular season and earlier playoffs as having value only in the sense that they qualify you to play in the World Series and take the position that every season has one successful team and 29 failures, then Morris accomplished as much in a single game as a borderline great pitcher can in his entire regular-season career. I find that position to be a bit too extreme. For one thing, the regular season has value in itself. Even if they did nothing in the playoffs, the 2001 Seattle Mariners were a great team that their fans should revere. I think the Braves of the 1990s accomplished much more than the Marlins of the 1990s. Of course, that's just one man's opinion. I've outlined some tools to measure the impact of great pitching in the toughest situations of great games. How that should be weighted alongside the regular season and other playoff series is a decision every fan has to make for his or herself.