Wednesday, August 21, 2013

Pedro Martinez's WAR In 2000 vs. Bob Gibson's In 1968

Something does not make sense to me and maybe I am doing it wrong or looking at something the wrong way.

Pedro Martinez has an 11.7 pitching WAR in the year 2000 while Bob Gibson has 11.2 in 1968. Martinez has a WaaWL% of .834 and Gibson has .754. It looks like that is the winning pct they would each have in the games they pitch if they were added to an otherwise average team. Gibson's lead in IP was 304-217.

If we divide their IP by 9, here is how many games they each had followed by their wins using the above percentages

Martinez 24.11, 20.11 (20.11 = .834*24.11)
Gibson 33.89, 25.55 (25.55 = .754*33.89)

Now if we subtract their WAR from these wins, here is how many games the replacement pitcher would win

Martinez 8.41
Gibson 14.35

But Gibson had 33.89 games and Martinez had 24.11. The Martinez replacement has a winning pct of .349 (8.41/24.11). The Gibson replacement has .424 (14.35/33.89). So it seems like Martinez is getting compared to a replacement with a much lower winning percentage. Am I doing something wrong or not understanding WAR correctly? Or is it supposed to work out like this?

Gibson's page at Baseball Reference

Martinez's page at Baseball Reference


Unknown said...

While it's less than perfect, if we use the bbref converter, this is what we get projecting first Martinez pitching in 1968 in Boston and then in StL.
Then I do the same thing, transposing Gibsons record to 2000 with both Bos and StL. This is what it looks like.

28 21 3 .875 1.17 224 107 33 29 14 27 293 12 0.598
28 22 3 .880 1.08 225 104 31 27 14 26 296 11 0.578
32 26 4 .867 2.05 272 251 68 62 14 79 239 9 1.213
32 26 4 .867 1.85 277 241 62 57 14 75 243 8 1.141

Cyril Morong said...

Thanks for your input. So this converter changes the IP? Maybe that somehow accounts for what I see in the WAR numbers, but I am still not sure

Vinnie said...

Without taking anything away from Bob Gibson, that 2000 season that Martinez had to me was the most dominant performance I've ever seen. He didn't battle hitters, he toyed with them in ways that numbers can't begin to express.
While I have no way of expressing it mathematically, I can't imagine that even with the additional innings that Gibson could have been close to the value Pedro had.

Cyril Morong said...


Thanks for dropping by and commenting.

But what I am trying to understand is how the WAR numbers were computed, not so much who had the better or more valuable season.

But here is a way you can see how Gibson might have been more valuable.

Gibson had a 258 ERA+. So in his year, a park neutral ERA for him would be 1.16. Avg was 2.99. That translates into a Pythagorean pct of .869. In 33.89 games that is 29.46 wins. If a replacement pitcher had a pct of .400, he wins 13.56 games and so Gibson is 15.9 above that.

If I do the same kind of thing for Martinez, he gets an era of 1.69 in a league that has 4.91 for average (he had an ERA+ of 291). His pyth pct is .894 or 21.56 wins in 24.11 games. A .400 pitcher would win 9.64 games so Pedro is 11.92 abov that

So Gibson beats him in WAR 15.9-11.92. This is because he has so many more IP, almost 90 more or almost 10 more games worth

Suppos a guy pitched 5 perfect games in a year and that was it. The most he could be above replacement would be 5. Quantity is a big factor in WAR, or I thought it was. But like I said in the post, it looks like a different winning pct is used for the replacement in each year

Vinnie said...

But context is everything. Had Gibson been pitching in 2000, he wouldn't have thrown those extra innings to begin with.
It may be impossible for that reason to compare who was more valuable based on that.
Why not try running the numbers based on what Pedro would have pitched in 68 and Gibson in 2000? I think you'll have a completely new view.

Cyril Morong said...

Again, I am not trying to have a view on who had a better season. I want to understand how their WAR was computed.

How can we decidec how many innings each guy would have pitched in the other guy's year?

Gibson gets a 2.05 ERA pitching for Bost in 2000 in 30.22 games. I get him with a WAR of 13.64. If I put Pedro on the 1968 Cards he gets12.11

Why does Gibson fall to 272 in 2000 and why does Pedro rise to 225 in 1968? I don't know how BB Ref does that