Even You, Dorn
by Alex
I've finally found a way to combine two of my great loves, sabermetrics and Major League. (Surprisingly, this is only the third Major League-related post in this blog's history.) Using a script I found online and Christopher Shea's Win Expectancy Finder, I set out to determine who deserves the credit and who the blame in the Indians' win in their memorable one-game playoff against the Yankees, way back in 1989. In the end, I wanted to create something resembling the graphics available for each game on Fan Graphs (you may be familiar with these same charts from wrap-ups at Let's Go Tribe).
The main problem, I discovered, is that the movie only depicts snippets from that fateful contest. I was able to reconstruct the seventh and ninth innings (almost) completely. And what I couldn't deduce I simply removed from the analysis. For example, in the top of the ninth, Harry Doyle says that Eddie Harris has given up eight hits, though we've only been shown one of them. I won't worry about when those other hits occurred, since they didn't affect the score of the game. Since the online Win Expectancy Finder is based actual results from real games, I set the time period for it to check to the ten-year window centered on the release of the film, 1984–1994, for really no good reason at all. I thought it would be both a decently-sized sample and close enough to the run environment of the late eighties.
The film breezes through the first six innings, without any real highlights. But we do know that it's still 0–0 in the seventh. Therefore, I can calculate Harris's win probability added (WPA) for each of those innings, since they each started the same way they ended. I simply subtract the Indians' win expectancy at the beginning from their win expectancy at the end of that half-inning.
Top of the 7th
Bottom 7th
At this point, we skip past the eighth inning—where no runs are scored—and head straight for the ninth. I'll include the eighth in my analysis as I did with the first six innings, to evaluate Eddie Harris.
Top of the 9th
Bottom of the 9th
You know what?
Getting back to business, Taylor's bunt was the clutchiest of the Indians' many clutch plays that night, netting .407 wins, and besting Cerrano's home run by four-hundredths of a win. Jumping backwards for a moment, you many have noticed something fishy with the numbers in this inning. The Indians' win expectancy goes down after Hayes's single. Why would this happen? You'd expect that a runner on first with two outs is better than bases empty with two outs. Remember that these numbers aren't based off a model, but real-life. So, from 1984 to 1994, home teams that had the bases empty with two outs in the ninth of a tie game went on to win 56.8% of the time; home teams that had a man on first with two outs in the ninth of a tie game went on to win 55.9%. Given a larger sample size of games, I'm sure the numbers would move closer to what we expect.
I'm not so sure I want to assign all the credit from the final play to Taylor, though. It wasn't necessarily a sure thing that Hayes would make it from second base on a bunt. Frankly, I don't think I've ever seen it happen anyplace else. If Hayes had gotten a bad jump, he could've been held at third. Then the Indians would've had men on the corners with two outs—better than before, but not a sure win. So, I reran the numbers.
Alternative Analysis of the Last Play
You knew I couldn't resist this:
Based on all my number-crunching, the true heroes of this game were (in no particular order): Jake Taylor, Pedro Cerrano, Willie Mays Hayes, and Rick Vaughn. "What about Eddie Harris," you may be asking. "He gave up two earned runs in 8 2/3 innings!" The problem is that his WPA comes out to a measly .093. And why's that? Because he left a tie game in the ninth with the bases loaded! And WPA doesn't even account for him loading the bases in front of the best hitter in the league, the Triple Crown winner, no less.
I know this has all been in good fun, but if you'll allow me to be philosophical for a moment: fans will remember Taylor's courageous bunt and Vaughn's electric fastball and maybe even Hayes's timely base-running, but will they pay proper respects to Cerrano's power? I can't answer that for sure. If you look at the numbers though, you see that Cerrano's home run was (arguably) the most consequential play of the day for the Tribe. Once again, we see stats glorify sluggers who only sit back and swing for the fences and condemn the little guys, the run manufacturers, who seize every opportunity for gaminess and hustle to win. The numbers may be right and they may not, but it'd be impossible to deny that this team won—not just this one game, but throughout the entire season—as a result of their determination, grit, heart, desire, and clubhouse chemistry. Score one for the purists.
The main problem, I discovered, is that the movie only depicts snippets from that fateful contest. I was able to reconstruct the seventh and ninth innings (almost) completely. And what I couldn't deduce I simply removed from the analysis. For example, in the top of the ninth, Harry Doyle says that Eddie Harris has given up eight hits, though we've only been shown one of them. I won't worry about when those other hits occurred, since they didn't affect the score of the game. Since the online Win Expectancy Finder is based actual results from real games, I set the time period for it to check to the ten-year window centered on the release of the film, 1984–1994, for really no good reason at all. I thought it would be both a decently-sized sample and close enough to the run environment of the late eighties.
The film breezes through the first six innings, without any real highlights. But we do know that it's still 0–0 in the seventh. Therefore, I can calculate Harris's win probability added (WPA) for each of those innings, since they each started the same way they ended. I simply subtract the Indians' win expectancy at the beginning from their win expectancy at the end of that half-inning.
- Top of the 1st: .592 - .541 = .051
- Top of the 2nd: .581 - .531 = .050
- Top of the 3rd: .581 - .531 = .050
- Top of the 4th: .587 - .519 = .068
- Top of the 5th: .590 - .521 = .069
- Top of the 6th: .592 - .518 = .074
Top of the 7th
- start of inning: .509
- [not shown]
- 2 outs, runner on 2nd: .527
- Harris gives up homer: .207
- assuming Harris gets the next man out, end of inning: .218
Bottom 7th
- start of inning: .218
- Willie Mays Hayes pops out: .187
- Jake Taylor grounds out: .148
- Roger Dorn singles to left: .182
- Pedro Cerrano homers: .551
- assuming the next guy gets out, end of inning: .517
At this point, we skip past the eighth inning—where no runs are scored—and head straight for the ninth. I'll include the eighth in my analysis as I did with the first six innings, to evaluate Eddie Harris.
- Top of the 8th: .632 - .517 = .115
Top of the 9th
- start of inning: .526
- [not shown]
- 2 outs, bases empty: .631
- Harris gives up single: .609
- Harris gives up double, runners on 2nd and 3rd: .489
- Harris walks batter, bases loaded: .434
- Vaughn replaces Harris, strikes out Haywood, end of inning: .654
Bottom of the 9th
- start of inning: .654
- [not shown]
- 2 outs, bases empty: .568
- Hayes hits infield single: .559
- Hayes steals second: .593
- Taylor bunts, Hayes scores: 1
You know what?
Getting back to business, Taylor's bunt was the clutchiest of the Indians' many clutch plays that night, netting .407 wins, and besting Cerrano's home run by four-hundredths of a win. Jumping backwards for a moment, you many have noticed something fishy with the numbers in this inning. The Indians' win expectancy goes down after Hayes's single. Why would this happen? You'd expect that a runner on first with two outs is better than bases empty with two outs. Remember that these numbers aren't based off a model, but real-life. So, from 1984 to 1994, home teams that had the bases empty with two outs in the ninth of a tie game went on to win 56.8% of the time; home teams that had a man on first with two outs in the ninth of a tie game went on to win 55.9%. Given a larger sample size of games, I'm sure the numbers would move closer to what we expect.
I'm not so sure I want to assign all the credit from the final play to Taylor, though. It wasn't necessarily a sure thing that Hayes would make it from second base on a bunt. Frankly, I don't think I've ever seen it happen anyplace else. If Hayes had gotten a bad jump, he could've been held at third. Then the Indians would've had men on the corners with two outs—better than before, but not a sure win. So, I reran the numbers.
Alternative Analysis of the Last Play
- Taylor bunt singles, Hayes to 3rd: .686
- Hayes "steals" home: 1
You knew I couldn't resist this:
Based on all my number-crunching, the true heroes of this game were (in no particular order): Jake Taylor, Pedro Cerrano, Willie Mays Hayes, and Rick Vaughn. "What about Eddie Harris," you may be asking. "He gave up two earned runs in 8 2/3 innings!" The problem is that his WPA comes out to a measly .093. And why's that? Because he left a tie game in the ninth with the bases loaded! And WPA doesn't even account for him loading the bases in front of the best hitter in the league, the Triple Crown winner, no less.
I know this has all been in good fun, but if you'll allow me to be philosophical for a moment: fans will remember Taylor's courageous bunt and Vaughn's electric fastball and maybe even Hayes's timely base-running, but will they pay proper respects to Cerrano's power? I can't answer that for sure. If you look at the numbers though, you see that Cerrano's home run was (arguably) the most consequential play of the day for the Tribe. Once again, we see stats glorify sluggers who only sit back and swing for the fences and condemn the little guys, the run manufacturers, who seize every opportunity for gaminess and hustle to win. The numbers may be right and they may not, but it'd be impossible to deny that this team won—not just this one game, but throughout the entire season—as a result of their determination, grit, heart, desire, and clubhouse chemistry. Score one for the purists.
19 Comments:
Any plans to tackle the sequels?
http://www.hulu.com/watch/4180/saturday-night-live-ed-glosser-trivial-psychic
Hayes leads off the 7th and bats in the 9th with 2 out & nobody on base. So that would mean that 8 guys came to the plate between those two events. However, we know that 4 guys came to the plate after Hayes in the 7th inning, 3 came to the plate in the 8th and 2 in the 9th, which adds up to 9, meaning that Hayes is the 10th person to bat after his previous AB.
On the other extreme, even if the Indians left the bases loaded in the 7th & 8th innings, the lineup still wouldn't have worked back to Hayes.
http://www.baseball-reference.com/boxes/NYN/NYN200505210.shtml
This is a brilliant post. Apologies for being so late to the party; just found it through Aaron Gleeman.
With no runners on base the final out would have to be made at first base. If there is a runner on first base then the final out could be made at first base or second base. Therefore having a runner on first base makes it easier to get the final out.
From the analysis, in the ninth inning with two outs and no base runners the win expectancy is .568. With a man on first the win expectancy goes down to .559 because there is now a force-out at two bases. When Hayes steals second the win expectancy goes up to .593. I think this is for two reasons. One, the runner is now in scoring position. Two, the force out for the base runner has been eliminated. You would say that the increase is entirely due to the runner moving into scoring position.
What if, at this point, Taylor walks to put a man on first and second? Does the additional base runner make the win expectancy go up or down or stay unchanged? I would contend that it goes down because there are now three possible force-outs. You would say that it stays the same because any force out at second or third could just as easily be made at first base. Do you have this number? If you do then we would know how much of the increase to .593 is attributed to the runner being in scoring position and how much is attributed to the elimination of a force-out possibility.
If the win expectancy does go down as I suspect it would, then the Yankees should have walked Taylor as soon as Hayes stole second base. Of course the two-out intentional walk is sure to upset some observers.
Post a Comment
« Home