August 28, 2005
Believeing Win Shares
Looking at the Win Shares on the Hardball Times, I'm surprised to see Gary Sheffield ahead of Alex Rodriguez. Alex is having a much better year with the bat. The players are on the same team and both are right-handed so the park factors are the same. Yet Sheffield has two more offensive win shares than Rodriguez. How is that possible?
The answer lies in the clutch adjustments to win shares. When Bill James worked on making the runs created formula more accurate for players, he found he need to make adjustments for batting average with runners in scoring position and home runs with men on base. He makes these adjustments based on comparing the situation to the player's overall averages.
When these adjustments are made for Alex Rodriguez, he loses 9.3 runs from the runs created formuala (I'm doing the calculation through yesterday). He loses 6.9 for his hitting with runners in scoring position and 2.4 for his home runs with men on.
Meanwhile, Sheffield gains from his superior battting in these situations. He gains 11.5 runs from his batting with runners in scoring position and 5.9 for his home runs with men on base (19 of his 27 have come with men on base). That's 17.4 runs in the plus column for Gary. Overall, Sheffield picks up 24.3 runs in the adjustment! It's the big reason why Rodriguez and Sheffield are almost even in RBI, despite Alex having a superior batting average and power numbers.
Bill James believes these adjustments give him the best estimates for runs created. I don't know how much of this is random luck rather than the ability of the players. But right now, Win Shares gives the AL MVP to Gary Sheffield.
Posted by David Pinto at
02:45 PM
|
Statistics
|
TrackBack (0)
Obviously, instinct and perception are of limited value, but it's really amazing how confident you can be as a fan when Sheff is at the plate in a big spot. It's not even a knock on A-Rod -- nobody on the Yanks since perhaps 1990s Bernie (or even occasionally, Straw) has brought that feel to the plate every single time there's somebody on base.
"I don't know how much of this is random luck rather than the ability of the players."
I don't know either, but I think James would say it doesn't matter. His system is designed to measure what a guy "did do," as opposed to what he can do. By that reckoning, I think it's perfectly defensible to add clutch performance as an ingredient.
I agree with Brian. It's not saying that ARod "isn't clutch", just that, this season, he hasn't done well in "clutch" situations. "Clutch" isn't a skill, but that doesn't mean you can't measure what someone has done in a "clutch" situation, particularly because you define what the word "clutch" means.
Agreed. Whether Sheff is getting lucky or not is irrelevant. Actual runs and wins are resulting from the timing of his performance. If you don't credit Sheff with that timing, you have to create a "timing" bin, and credit those runs and wins to a non-player.
It's like a monkey throwing a dart 15 years ago and hitting "MSFT". Today, he'd have 10 million$ and Eva Longoria.
What Brian, sabernar and tangotiger said - clutch hitting may be luck, but pennants are won and lost every year on its presence or absence.
