clock menu more-arrow no yes mobile

Filed under:

Red Reposter: Sabermetric Saturday

Can we use pitchf/x data to make a better projection of what to expect from Mike Leake in the immediate future?  Nick Steiner can.
Can we use pitchf/x data to make a better projection of what to expect from Mike Leake in the immediate future? Nick Steiner can.


Projecting Hanson | The Hardball Times
Over the past few years, I've become increasingly interested in projections. Not so much to predict the future, per se, but as a way of most objectively assessing a player's true talent. The problem with traditional projection systems is that they are generally limited to only using only official statistics: hits, homers, innings pitched, strikeouts, etc. Folks have done a great job with those data, but we're at the point where little progress can be made without additional data.

This is why I think Nick Steiner's article on Wednesday is quite possibly the most important sabermetric article published so far this year. It's the best and most comprehensive attempt I've seen to explicitly use pitchf/x data to improve a projection system. Here, the approach is to calculate similarity scores based on velocity, movement, pitch types, release points, and location and use similar pitchers to establish a baseline for his case study, Tommy Hansen. Then, Nick regressed his traditional pitching numbers to this baseline. Simple in theory, very complicated in practice, but it's very exciting work. Best part is, now that he has a methodology in place, it should be easy(er) to do it on other pitchers. This is the next frontier in projection systems, folks.



Baseball-Reference Blog » Blog Archive » 10 for 10: #7 Player Wins Above Replacement
B-Ref is now hosting Rally's WAR (rWAR) data. This is terrific--makes it much easier to access, and much easier to compare to "normal" stats. I'm thrilled. It does include 2010 data for hitters, and soon for pitchers. Should make for a great comparison reference to FanGraphs WAR (fWAR). In some ways it is superior to fWAR--it uses an appropriate league adjustment that differs between AL and NL. rWAR also, however, uses an inferior fielding stat: Total Zone doesn't have all the info that UZR does, and so it's probably less reliable.




Do resigning teams know more? | THE BOOK--Playing The Percentages In Baseball
Tango continues his investigations of Matt Swartz's study from a few weeks ago that showed that players who re-sign with teams tend to hold up better than players that leave as free agents. He finds that, in fact, the effect seems particularly strong with pitchers. Teams will re-sign pitchers that they feel confident will continue to pitch well, and let walk those who are more likely to break down. This means that teams, probably with scouting, can predict durability of pitchers--at least, they can when they know them intimately.


Caught Quantifying | Baseball Prospectus

Eric Seidman has started working on catching metrics, including, last week, interviews with some actual major league catchers about what they think should be used to evaluate their position. This week, he gives us a nice overview of some of the different attempts at catcher fielding metrics. This has been a pet topic of mine for a few years, and the guy actually quotes me at one point (though, fwiw, I'm no longer convinced that my approach on errors that he cites is the right one). Anyway, I'm interested to see what he comes up with. Already there are some promising improvements mentioned: he's apparently going to account for pitcher handedness in SB prevention, for example. I'm sure that's worth doing. The pitch framing thing is also exciting, though as we've shown in this space before, effect sizes in studies thus far are far too large to be believable--lots of confounds on that front.



2010 PrOPS Over- and Under-Performers



JC releases the top 30 over- and under-performing players according to his "old" stat PrOPS. This is a nice diagnostic statistic that can be helpful in determining how a player has "really" hit compared to his performance. It uses batted ball data, walks, strikeouts, etc, but no information about actual singles, doubles, hr's, etc. No Reds made either list--which, I think, is a good thing.




Reliever WXRL and Pythagorean Over Achievement | Capitol Avenue Club
Neat study from last October looking at the relationship between bullpen strength and performance above pythagorean winning percentage. I'm not entirely sure I like the use of WXRL here, as there are some potential bases that have more to do with how often a team gets a 1-run lead to the bullpen than how well the bullpen performed. But still, it's a strong effect and supports the idea that a good back-end of the bullpen will help you beat your pythagorean record.



Baseball Prospectus | Manufactured Runs: Everything You Wanted to Know About Run Prevention But Were Afraid to Ask, Part 1
Colin Wyers gives us a nice, readable overview of the basics of run creation.




Interleague begins today: AL teams look to pad their records! - Beyond the Box Score
Shameless plug for something I wrote at BtB summarizing what's been happening with interleague play: in other words, the AL completely spanking the NL teams year after year since 2005 (but NOT in the eight years prior to that). I summarize what's happened at the league and team levels, and then discuss several of the explanations (good and flawed) for why the AL seems to be so significantly better than the NL. This, to me, is one of the great, poorly-understood questions in baseball right now. It's SUCH a huge disparity that it boggles the mind.