Stats Geek: Bullpen giving Pirates chances
Share with others:
Were it not for a strong bullpen, this April might have been another disaster for the Pittsburgh Pirates.
That might seem a curious proclamation in the wake of Salomon Torres' three blown saves, but the overall work of the relief corps has kept the team dancing around .500.
I can show that because FanGraphs.com plots the course of every major-league game. Based on thousands of ballgames, these graphs show the odds of winning at any given point. Every action, be it a hit, walk, double play, bunt or steal, is plotted for how it increases or decreases the probability of winning.
Take Thursday's victory against the Astros: Jonah Bayliss takes the mound in the sixth inning with the bases loaded and nobody out, and the Pirates down, 1-0. History tells us the home team wins only 18 percent of the games that reach this point.
But Bayliss induces an infield popup and a double play. No runs score, and the Pirates' chance of winning shoots to 40 percent. The Pirates go on to score five runs and Torres comes out in the ninth to protect a four-run lead. This scenario has a Win Expectancy of 98.4 percent.
Torres gives up a run and is relieved by John Grabow with two out and two on. At that point, the odds of winning have dropped to 96.6 percent. Grabow gives up a single and a walk but gets the third out and the save. Pirates win, 5-3.
On the day, the relievers' collective Win Probability Added is 45.7. Bayliss, with two scoreless innings, accounts for 31.8; Matt Capps, with a scoreless eighth inning, 12.4; Grabow, 3.4; Torres, negative 1.9.
If that makes your head spin, think of it this way: When a game begins, each team has a 50-50 chance of winning, so the winning team always accumulates 50 points and the loser drops 50. In the 16-inning, 4-3 victory against the Astros last week, Pirates batters had a negative 61 points and the pitchers a positive 111.
FanGraphs also includes a Leverage Index to indicate the precariousness or ease of any given situation, which tells you a lot about your Pirates. With an assist from FanGraphs founder David Appelman, I discovered:
No NL team other than the Diamondbacks has put its bullpen in more precarious situations than the Pirates. No pitcher in baseball has been on the mound in more crucial situations than Torres.
Capps, pitching in the setup role, has done more to help his team win than all but three relief pitchers in the league. Shawn Chacon, Bayliss and Grabow have been positive forces, too.
The best WPA on the Pirates belongs to starter Ian Snell at 119. That's fourth best among NL starters.
Going into the game last night, Pirates starting pitchers collectively had a nearly neutral 8 WPA, the relief pitchers 185 and the hitters -- we'd better make that "batters" -- a negative 243.
The WPA for the bullpen is sixth best in the NL. The Dodgers, Padres, D-backs, Brewers and Cardinals have been better. The Rockies, Phillies, Marlins and Reds are the league's worst.
More than a dozen major-league closers have negative WPAs. Torres, with a minus 59, is currently in the company of the two greatest closers of our time, Mariano Rivera, minus 128, and Trevor Hoffman, minus 54. Bob Wickman, minus 43, and Tom Gordon, minus 76, are other prominent names in negative territory
One way to score well in WPA is to stink or excel at the right times. Chacon gave up three runs in three innings on Sunday, but the Pirates were already down, 6-4, when he entered, so his negative WPA was only 13. When he kept the Astros scoreless from the 11th through the 14th innings in the 16-inning victory, he scored 57.
The WPA offers one answer to that unanswerable question: Who is "clutch?" Was Chacon clutch to shut down a team when it counted and falter when it didn't need him as much? Or were his performances more about the batters he faced and the way his arm felt?
The WPA can turn on a guy. Last season, when Torres tied the Pirates' record for games pitched at 94 and finished the season as the closer, he finished fifth among NL relievers in WPA, just ahead of the man he succeeded, Mike Gonzalez.
In years past, Mike Williams went from a very good relief WPA in 2002 to the worst in baseball in 2003. Jose Mesa went from a very good WPA in 2004 to the second worst relief WPA in 2005.
A closer who saves only eight of every 11, as Torres has done to start the year, isn't doing his job. The average team wins more than 80 percent of the games in which it holds a one-run lead in the ninth, and well over 90 percent of the games when it has a two- or three-run lead.
The ninth isn't always the most crucial inning. If the team blows the lead beforehand, the closer never even gets in the game. A setup man, like Capps this year or Torres in years past, can help his team more than the closer. Flipping the roles of Capps and Torres won't do much if the old guy doesn't pitch better. Period.
Why there must be a single pitcher assigned to small leads in the ninth inning is another question. The point here is the bullpen is a collective operation, and this one has done the Pirates good despite Torres' struggles. Here are the WPA totals through Sunday: Capps, 113; Chacon, 68; Bayliss, 56; Grabow, 26; John Wasdin 3; Damaso Marte, minus 10; Juan Perez, minus 11; Torres, minus 59.
A FanGraphs chart plots how the probability of the Pirates winning increased and decreased in five separate situations in an April 26 game against the Houston Astros.
Click image for larger version.
Chris Park, Associated PressGood news, Pirates fans. San Diego Padres closer Trevor Hoffman's Win Probability Added number is almost as low as closer Salomon Torres'.
Click photo for larger image.
First Published April 30, 2007 10:27 pm
