Plenty O'Links

This won't be as good as Aaron's Link-O-Rama posts but I have run across a few interesting things related to performance analysis and other topics recently.

An Interview with Tom Tango, one of the authors of The Book over at Baseball Digest Daily. I found it interesting that Tom is more interested in hockey and is currently doing more analysis on that side. I've read the book and plan to write a review either here or on Baseball Prospectus but suffice it to say that if you're serious about analysis you need to have this book in your library. I've referenced their work on platoon splits in several of my recent columns and their work is more sophisticated than what I presented.

My latest column is also on the topic of the wisdom of crowds and platoon splits where I attempt to answer some of the common questions I received. The long and short of it is that platoon splits appear to differ when looking at players with smaller and larger isolated power (ISO) numbers. I also show that the splits are fairly constant over the course of careers and left-handed hitters don't really seem to learn to hit lefties better (at least relative to how they hit right-handers). A reader noted that his theory has always been that a pitch coming from the same side as the batter will necessarily decrease reaction time and since the margins are so small already, the differences are probably more a function of physics than anything else. Makes sense to me.

A new issue of SABR's By The Numbers has been published. In addition to reviews of both The Book and Baseball Between the Numbers, this issue includes a comparison of catcher evaluation statistics, an article on the relationship between runs scored and runs allowed, another derivation of Bill James' pythagorean method, and an article by Jim Albert that breaks down a hitter's outcomes into strikeouts, walks, homeruns, and hits.

Albert shows that when using these four outcomes as rate statistics, he can get an R-squared of .896, which is right inline with Runs Created and OPS. However, when he adds a differntiation between all hits and doubles and triples the correlation doesn't improve dramatically and only rises to .909. Although he doesn't show it, I would assume that strikeout rate doesn't add much to the picture and that perhaps using doubles and triples instead of strikeouts may in fact yield a correlation greater than .896. He then goes on to use a model to estimate the distribution of talent across his four measures and calculates the regression to the mean that would be used when estimating a player's true skill levels. Of course he shows that hit probablity is far more influenced by luck than any other factor and that strikeout probabilty is the most dispersed of the measures. In any case, some very interesting work here as usual.

Rany Jazayerli has a great article today on rain outs and suspended games that's so chock full of common sense you have to read it. It's one of those pieces that makes you say, why didn't I write that?

Retrosheet has posted some new event files! Yes, the 1911 NL, 1922 NL, and 1957-1998 data is now all there along with 2005. The 1999 data should be coming "in the not too distant future." And I've heard that John McGraw's strict platooning of Casey Stengel is borne out by the 1922 data. It gives me goose bumps just thinking about digging into that data :)

And Tom Tango has an article on The Hardball Times today discussing win probability and its use in leverage.

Maury Brown, whom I'll be presenting with at SABR36 in Seattle had another excellent piece on the upcoming labor negotiations. He even has the amounts paid in a taken out by each team in the revenue sharing setup. The Marlins this year received $31M in revenue sharing and are fielding a team with a payroll of $15M. I sure hope they're using that money on player development.