Thanks to Tim for giving me this opportunity. To give you a little background, I recently co-authored The Book--Playing The Percentages In Baseball, which you can read more about at http://www.InsideTheBook.com , and I run the sabermetric site http://www.TangoTiger.net . There were a few questions posted already, so let me take this time to answer them. I'll hang around for a few days to answer any other questions you may have. There are three basic forms of analysis: 1 - Using aggregated, seasonal data. There are many out there, and for purposes where more granular data exists, they should be ignored. 2 - Using play-by-play (PBP) data. Mitchel Lichtman's UZR (also my co-author), John Dewan of BIS and David Pinto all lead the field. But, they lead the field not so much because they are smart (they are), but because they have access to data. Their systems all start with the same premise: "what's the chance of this particular ball being caught?". Their premise is exactly the same question any fan would ask. Where it gets interesting is how you account for that context. Where exactly was the ball hit, how many hops did it take, or how long did it take to get there, what was its trajectory, where was the fielder playing, and why was he playing there. How did the park affect the ball? Was the batter a lefty or righty? What kind of pitch and pitcher are we talking about. These are all questions that the fan needs to ask to establish the probability that a ball will be caught. A non-PBP system has to make alot of guesses, while a good PBP system (i.e., how reliable are the scorers), can tell you alot more. 3 - Scouting information, like on my site: http://www.tangotiger.net/scouting/index.html http://www.tangotiger.net/scouting/scoutResults2005_CHN.html Until your sample size reaches a certain point (maybe 200 or 300 BIP), what a fan sees has more reliability than what a stat will tell you. OBP and other stats are results-based performance metrics. It's not enough to say "get a high OBP". It's "how do I approach this at bat to maximize my chances of winning". A guy like Vlad or Soriano know their skillset and figure out how to maximize their talents. If that means hacking or scaring pitchers or whathaveyou, good for them. It's just like in any other sport. Balancing throwing a football 25 yards against 5 yards. Taking a shot from 40 feet from the goalie against trying to stickhandle your way in for a 20 foot shot. There's a whole balancing act that players try to establish as to what's the optimal way that they themselves should play. If your beef is really that players haven't figured out that 4 walks in 4 PA is roughly equal to 1 HR, 1 single and 2 outs, then I agree. But, I think players know the value of the walk, especially as they get older. Their walk rates actually increase, on average, every year until the age of 38 (or later). In our book, we have a 50-page chapter as to when a bunt is appropriate and when it is not. There are many many variables to consider, and in many cases, the old-school had it right. The reason the new thinkers think that the bunt is no good is because they think of only "average" situations. But, the bunt is about leveraging specific situations. To give you a taste as to what we considered, check out Tables 96 through 119: http://www.insidethebook.com/cl.shtml Believe me, it's not as simple as some want to make it. I'm not privy to the inner workings of most teams. Of the teams that I've dealt with, they definitely want to have reliable information. The peak of sabermetrics will be the convergence of performance analysis and scouting observations. It's not an either-or thing. The problem is trying to crunch the numbers to tell you a reliable story. Too often, each side will use selective data from the other side to prop up their own argument. Doesn't make sense. Scouts are also way overworked. To give you one example, I saw the actual scouting reports to Jeff Francoeur and Yuniesky Betancourt. Now, my Fan's Scouting Report was very clear that these guys are great fielders, with Betancourt already one of, if not the, best fielding SS in MLB. The scouts called these guys average to above average. But, then more scouting reports came in, and they change their tunes to give them higher marks. In a matter of months, the exact same scouts moved Francoeur's arm from average to a plus. I would have a bunch of scouts. These guys make peanuts, 30K to 40K I think. Teams only have 30 of them (about 1 million$). I'd double my budget for these guys to 2 million$, at least. As for the "new-age", again, most people don't really appreciate the sample size issue, regression, or the external factors that play a huge role in performance numbers. Teams are right to stick to traditional evaluation, because it's what they know best. The worst thing would be to give someone a tool, and not be taught exactly its use and limitations. I doubt it. Bringing in people accountable for budgets is what would have impact. But, as long as guys like to burn 16 million$ like they did for Cristian Guzman, there's room for improvement. I was born in Montreal, and lived there until a few years ago, and was a huge Expos fan. I also caught alot of Redsox games on the old UHF signal (pre-cable days), and grew up a fan of the Sox as well. I'm not sure why I do what I do. I like it, and others seem to like what I do as well. OBP, absolutely. If you take double OBP and add SLG, you pretty much get a great overall asssessment of performance. The flaws in OBP (weighting everything as "1", meaning underweighting homers and overweighting walks) is nicely balanced by the flaws in SLG (underweighting walks, as 0, and overweighting homers). No other metric will come to achieve similar popularity simply because no other metric can exist. Any new metric will be something convoluted that is not as easily explained as OBP and SLG. The simplest was to evaluate a pitcher is with FIP, which you can read about here: http://www.tangotiger.net/drspectrum.html You really need to understand regression to make sense of pitchers. Each component tells you something different about a pitcher, and it each has its own level of reliability. K/BFP has a high reliability, while non-HR hits per ball in play has a low reliability. For those who are adventurous, I suggest reading: http://www.tangotiger.net/solvingDIPS.pdf And relievers and starters should be evaluated differently. For those who have read The Book, you know that it's alot easier for a pitcher to perform well in the reliever role than as a starter. It's a huge advantage, on the order of 0.80 to 1.00 in ERA. Evaluate things in their entirety. I don't know what people do or don't look at, so I don't know what they should or should not look at. Just look at things in the right light.
