MLB HOF - Too Big or Too Small?

[Mike D.]

Two more thoughts:

1. The people who really seem to “get” and use WAR mostly openly admit it’s imperfect and is one of many tools you should use to determine greatness. But having something less subjective to drive a conversation has value. It’s kind of the “scouts vs stats” thing from 20 years ago. The “stats geeks” were saying “and” and the old school crowd was hearing “or”.

2. I hope we’re taking “relative” hall of fame size, not raw size…since every year (or so) more players get in. I’d be interested to see how % of players get elected to the hall…adjusted for things like league size.

[rats60]

Originally Posted by Mike D.

Two more thoughts:

1. The people who really seem to “get” and use WAR mostly openly admit it’s imperfect and is one of many tools you should use to determine greatness. But having something less subjective to drive a conversation has value. It’s kind of the “scouts vs stats” thing from 20 years ago. The “stats geeks” were saying “and” and the old school crowd was hearing “or”.

2. I hope we’re taking “relative” hall of fame size, not raw size…since every year (or so) more players get in. I’d be interested to see how % of players get elected to the hall…adjusted for things like league size.

WAR is totally subjective. BB Reference uses ERA to determine WAR for pitchers. Fangraphs uses FIP. Fangraphs thinks Bob Veale was worth 8 WAR in 1965, better than Juan Marichal's 6.8. Baseball Reference thinks Juan Marichal was worth 10.3 WAR and Bob Veale was only worth 4.0. Which one is correct? How can you trust something that has so much variance? What do the error models look like for these two formulas? Imperfect has to be the understatement of the year.

[Mike D.]

Originally Posted by rats60

WAR is totally subjective. BB Reference uses ERA to determine WAR for pitchers. Fangraphs uses FIP. Fangraphs thinks Bob Veale was worth 8 WAR in 1965, better than Juan Marichal's 6.8. Baseball Reference thinks Juan Marichal was worth 10.3 WAR and Bob Veale was only worth 4.0. Which one is correct? How can you trust something that has so much variance? What do the error models look like for these two formulas? Imperfect has to be the understatement of the year.

To me….that’s not WAR being variable. Within either model, you get pretty good consistency.

But comparing two completely different things that happen to be called the same thing and expecting consistency is nuts. Even then, in most cases I’d guess the two models generally agree, although as you point out there are cases where there is wide discrepancies.

I use BB-R WAR because I’ve used that site since close to the beginning. FanGraphs is a great site I’m sure…but every time I go there it makes my head hurt. Too much advanced math for me!