This is our old blog. It hasn't been active since 2011. Please see the link above for our current blog or click the logo above to see all of the great data and content on this site.

## Bloops: Patrick Flood’s Mets Blog – Exile on 126th Street: WAR problems and the Mets Crazy Horse.

Posted by Sean Forman on July 20, 2010

If you are a serious stathead and are curious about the differences between our WAR numbers as produced by Sean Smith and those on Fangraphs.com check out this two part (very extensive) set of posts by Patrick Flood. Very worth your while if you are into this sort of thing and written in an entertaining way as well.

### 5 Responses to “Bloops: Patrick Flood’s Mets Blog – Exile on 126th Street: WAR problems and the Mets Crazy Horse.”

1. DavidRF Says:

Thanks! I've been waiting for this question to be addressed.

2. Michael E Sullivan Says:

His decimal hatred is a bit weird from a guy who cares about sabermetrics. Seems to be assuming a pretty astounding level of innumeracy from the casual fan to want to avoid decimals altogether.

3. dukeofflatbush Says:

Its a lot of fun to read an article that marries our nerdy number systems to witty and insightful writing. I think a casual baseball fan could appreciate that article's style. The author obviously knows baseball, knows how to write, knows how to make fun of himself and us. But I think the overall point, and a great one at that, is there is no sure fire way to measure player value. I was skeptical of WAR at first, but it became standard in every discussion very quickly. It became our yardstick before many of us were even sure how to tabulate it ourselves. I'm all for the idea of creating one number to attribute value, but there is way to much subjectivity with WAR. Anyway, Patrick Flood's writing is 7.8 WAR... give or take.
Outstanding.

4. Johnny Twisto Says:

From the article:
Also, at the end of the season, B-R’s offensive number is adjusted so that the number of runs a team is credited with creating matches up with the ACTUAL number of runs the team scored; Fangraphs doesn’t do the same thing. B-R’s version of WAR is rooted a bit more in what actually happened on the field.

I didn't know Sean Smith/BR did that. I like that. When I calculate runs created that's how I do it too.

Thing I have thought about WAR is that a "replacement level" defender is actaully probably better than an average MLB fielder because players make it to the bigs with their bats, but it seems like WAR treats a replacement level players as being well below MLB average at everything. The same thing may also be true of baserunning.

I don't think that is true. That is what Baseball Prospectus's WARP did -- set a replacement batting level AND a replacement defensive level, and compare players to both. The accepted wisdom now seems to be that "replacement level" players are mostly bad with the bat, and average, on average, with the glove. WAR just estimates replacement to be worth about 2 wins less than average, and those 2 wins are not broken down into batting/fielding/whatever.

Michael:
His decimal hatred is a bit weird from a guy who cares about sabermetrics.

He didn't say this (I don't think), but he may have a problem with these numbers implying a level of precision that they don't have. There is a lot of research and solid theory as the basis of these calculations, but they are all still estimates, and WAR is a bunch of estimates added together.

5. Michael E Sullivan Says:

That didn't seem to be his complaint. And even if it was, I don't really like the idea of randomly rounding, as it just gets rid of even more precision, when you start doing calculations with the results (like tabulating career WAR). It might be worth doing a study, so at least stat heads who spout the number have an idea of what the confidence intervals are, given the general distribution of probability around events.

But the real lack of precision comes from the fact that you simply can't consider what isn't on the scorecard (or visible and interpretable on the video record for UZR, or whatever), and the other fact that no measure of ability is fully predictive. Even if we found the perfect measurement, players get better and worse.

It does seem to me that, whatever their shortcomings, modern sabermetric statistics give us a *much* better sense of player's value than the statistics which were most common 20-30 years ago.