How important is Nathan?

A STRIB story on Joe Nathan demonstrated once again the danger of statistical analysis. The numbers may be right, the computer spreadsheet almost guarantees it. But they often don't mean what the people using them think they mean.  Here is a snippet from that article:

"

Baseball Prospectus compiled all the data from the past 10 years, and home teams that had a save situation entering the ninth inning -- leading by three runs or fewer -- converted 87 percent of those saves. Road teams converted 86.1 percent of their ninth-inning save situations.

In other words, whether it's Nathan or anybody else standing on the mound, the Twins have less than a 14 percent chance of blowing that save."

It actually doesn't mean that. It means that is the average, but some pitchers are below average - perhaps a long way below average.

The other problem is, as most people know, the real difference is not how well pitchers do with a three run lead. It takes a real meltdown to blow one of those and those are pretty rare for any quality reliever. If you dilute the sample with a lot of sure things - you get a high percentage of success. But it tells you nothing about the relative ability of the closers, it just means you have masked it with a lot of situations where failure is basically random.

A closer who blows three or four more games than the next guy costs his team three or four wins. That is enough to decide a close pennant race.  And if the alternative to Nathan pitches like Dave Stevens, it will be more than three or four games the Twins lose as a result.