Brawndo the Thirst Mutilat

Like fun it was, or I don't have the foggiest clue what a strawman argument is. Your point, if I'm not mistaken, was that one particular player, out of the hundreds of players in MLB, has had wildly fluctuating UZRs from one year to the next. All I did was point out that if that is your standard for the validity of a metric, you'd better throw batting average out the window right quick. Anyway, I'm done trying to defend Fangraphs' WAR or UZR on this site. For half a decade, it's been somewhat of a tragedy that cutting edge valuation measurements weren't freely available on the net. Dave Cameron & co. have done a wonderful thing by making the best of them freely available, and by making their derivation more transparent and peer reviewable than BP or anybody else ever dreamed of doing, and yet citing these numbers still illicits heaps of indignation and scorn even on a site with as relatively intelligent a level of discourse as this one. If you want to talk about a strawman argument, howsabout the idea that anybody stated that WAR was the "end all be all" of anything? I don't think anybody other than its detractors described it in any such fashion anywhere in this thread.

I guess I needed to make the point of my post much more obvious. What I'm trying to say is that volatility is not, in and of itself, a reason to dismiss a statistic as though it isn't measuring anything. Once you account for the fact that in a given season, a position player gets countless more ABs than he gets non-routine fielding opps (the kind that differentiate between good and bad fielders), you see that batting average is at least as volatile as UZR, but you don't hear anybody arguing it's a useless stat as a consequence of that volatility.

If you account the discrepancy in the number of opportunities, this reasoning would also forbid you from ever looking at batting average.

Per McCalvy: Tidbits The Brewers do not intend to recall shortstop J.J. Hardy from Triple-A Nashville before Sept. 1, and as a result, Hardy will not have the Major League service time he needs to qualify for free agency following next season. Instead, he'll have an extra arbitration year before reaching the open market in the winter of 2011.

Kazmir cleared waivers. Anybody could've traded for him.

Just don't see how Bradley and Vick are even remotely analogous. Check the "controversies" section at the bottom of his wikipedia page. There's the thing with the umpire when he was with San Diego (which led to MLB disciplinary action against the umpire) and an attempted confrontation with a Kansas City TV announcer. My reaction was "That's it? What fiendishly heinous and vile acts am I missing that make this guy the devil?" Nothing he's said this year about Cub fans in the bleachers or the Chicago sports media struck me as out of line, at least in the sense that it seems plausible enough that all of it is true. He might be professionally unwise for saying it, but I doubt he's making any of that stuff up. He's got a temper and a chip on his shoulder. So does Prince. Only differences between the two of them are that Bradley makes the media aware of his disdain for them (to his credit, I have no idea how Prince feels about the media) and the chip on Bradley's shoulder is large enough that it might actually qualify as a full-blown persecution complex. If I'm honest, the former difference just makes me a Bradley fan, the latter just makes him a sorta interesting and sad character study. Besides, I think I'm willing to live with the aforementioned quirks for a guy who has a RoS ZiPS projected line of .276 / .394 / .471 and defense at least in the neighborhood of average in RF.

Juan Cruz wound up having to settle for $3 mil per year for 2 years...from the Royals. There may have been others as well. I don't think this year's free agent class is going to be nearly as confident that the free market will treat them more kindly than arbitration would, though, so I agree with you that it would be very unfortunate if Lopez sneaks in as a type A, for both the Brewers and for Lopez. It's pretty sad that the system is sufficiently broken to give Lopez an incentive to hit like crap the rest of the year so that he doesn't become unsignable.

Linear weights values have been derived a number of different ways. You can read about them here. The values I used came from Palmer, the inventor of the concept, and I believe he was using what is described in that link as the "empirical approach". Basically, this method involves calculating the average change in run expectancy that results from every play over the course of, say, an entire season of data. As to whether the coefficients I used undervalue successful steals and unduly punish CS, most versions you see lately are even more unkind. Check out the "run SB" and "run CS" columns in this table. Those are the run values that serve as the basis for all of the advanced offensive stats at fangraphs (they are the "w" in wRC, wRAA, and wOBA). In recent seasons, it looks like I should have used something more like .20 and -.43 for SB and CS, and .30 for the difference between a single and a double. That gives a net value of 4.1 runs for Escobar's SB/CS vs. 9.6 runs for 32 extra doubles. The ratio is about the same, though the run values are lower across the board. For individual players, linear weights are generally considered to be the gold standard of run estimation in the analytical community.

Between Melvin playing dumb at the press conference and Ash saying this today, I think it's fairly clear that they are at least kicking around the idea of delaying Hardy's FA. They have a few weeks to decide. He could help them out by getting injured or by continuing to fail to hit worth a windy spit in AAA. Also, Dave Cameron at fangraphs has a post about this situation. He basically calls shenanigans, and is not pleased.

This is quite simply not how it works, at least if your intent is to preserve any modicum of the sorta-useful-in-a-back-of-the-envelope-kinda-way utility of OPS as a run estimator. Those stolen bases aren't like turning singles into doubles, which is what you are doing here. They only advance Escobar an extra base, whereas turning a single into a double advances Escobar and all other baserunners at least one extra base. Using linear weights values, Escobar's 42 SB have been worth 42 x .22 = 9.24 extra runs. His CS have cost 10 x .38 = 3.8 runs. Net gain from attempted steals is 5.44 runs. If you turned 32 singles into doubles (which is what your OPS calculation is doing), that would be worth the .38 x 32 = 12.16 runs. That's more than twice as many extra runs.

Melvin just claimed in his press conference that he doesn't know the answer to this question. He said "Gord might know, but I don't." That has to be a lie, right? It took us stupid fans less than a couple hours to figure out that this might be a possible consequence of, or even the primary motivation behind, this move.

Why would anyone bother? ERA+ is already ballpark adjusted, so I don't really get what you're driving at here. Granted, he's managed to post solid ERAs the bulk of his career, and the last few years as well, but always as a consequence of a fair amount of BABIP luck or LOB% luck. Maybe I'm being SABR dogmatic here, but w/o decent K/9 or BB/9 rates, count me as a sceptic on the question of whether Weathers is still a decent pitcher.

It's his version of a curveball. Here's an interview where he talks about his pitches, as well as his awareness that scouts, GMs, etc. have a hard time taking him seriously.

Some pitching statistics actually do stabilize over very small samples. Swindle faced enough batters in AA and AAA last year for his strikeout rate (which was well over 10 per 9 IP and translated into an MLE just north of 9 K/9) to be considered reliable. His GB/FB distribution should also have been considered reliable based on last year's numbers, though there isn't anything unusually good or bad about it. Add in this year's Nashville line and you've almost got enough TBF to say his walk rate is reliable. That has also been excellent in the high minors, translating to an MLE of 2.34 BB/9. It's basically impossible to be a below average RP striking out that many and walking that few. Swindle was released because of a .455 BABIP and a 39.1% strand rate in the majors. It is also basically impossible to have worse luck than that in these, the most luck-driven of all pitching statistics. The same thing could have just as easily happened to Stetter or DiFelice (in fact, it almost did happen to DiFelice) in their first few big league IP and we would never have seen those guys again either, unless some other smarter team decided to give them a shot, all because of unquestioning worship at the unholy altar of the radar gun. Tampa is as smart an organization as there is in MLB at the moment, and it's likely that they'll wind up with him in their system as bullpen depth, and off the 40 man and out of the way. Good for them. I really hope he gets another shot to pitch in the big leagues someday.

Dave Cameron over at fangraphs weighs in with his take on the Weathers trade. I am inclined to agree with his conclusion: "The Reds were right to get rid of Weathers - he's old, not good, and makes more money than he should." The PTBNL probably isn't anybody of consequence, and I still don't like this deal. If the PTBNL is actually someone vaguely resembling a prospect, I loathe this deal.

An inconsistent pitcher was more valuable in the model Szym ran mostly because the pitcher was bad (ERA was held constant at 5.00), and was giving up runs at a higher rate than the offense in the model was assumed to be scoring them. While the Schizo Samwell distribution will probably produce the best results no matter how good or bad you are (even if my ERA was 90.00, I'd get great results if I could manage to give up 0 runs in 9 CG SOs and then give up 900 runs every 10th game), if Szym had run the model holding the ERA constant such that the fictional pitchers were giving up fewer runs than the offense was scoring, the results would have been very different. Still, I think endaround has it here: don't worry about consistency in scoring or allowing runs -- worry about scoring more runs and allowing fewer. The Brewers have been outscored by enough runs this year that they actually should have a slightly worse record than they do, so it's tough to conclude that the distribution of the runs they have scored has been a problem.

This is just a small sample artifact, though, and in no way is it an indication that the metric is broken. It's the same thing as when a guy has an incredibly hot first couple weeks of the season, and people say, "he's on pace for 108 HR and 264 RBI!" From a simple mathematical perspective (multiplication and long division = ok, linear regression = huh?), the statement is true, but everybody knows that he won't maintain that pace, and that the "projection" is just a silly curiosity. It's a way of expressing how hot the guy has been, rather than forecasting how well you think the rest of his season will go. If you look at the UZR (not UZR/150) leaderboards at fangraphs for any given season, you'll see that only a couple players usually crack 20 runs or so in either direction. Last year, Utley was the only +20 guy, and the only -20 guys were Abreu and Hawpe. Hawpe was the only player in MLB with a result I'd describe as outlandish (-37.2 runs), and honestly, part of understanding statistical distributions is understanding that you have to expect the occasional weird outlier datapoint that is several standard deviations away from the mean. This year is much the same. Nyjer Morgan (if you combine his LF and CF innings) and Adam Dunn are really the only guys with realistic shots at ending up more than 30 runs from average in either direction (give you three guesses which one is positive and which is negative). Out of the 4 or 5 additional guys who could get to +20 or -20, a couple probably will. I suppose my point is that McGehee's huge negative UZR/150 number isn't a common thing for a player with anything resembling a meaningful number innings for UZR to work with. Obviously, it's much more common over short runs of playing time, because the signal-to-noise ratio just gets too low. McGehee's UZR/150 was +59.2 in 41 innings for the Cubs last year, it's -48.3 in 168 innings at 3B this year, and neither of those results should be seen as useful descriptively or predictively, but neither should they be seen as a reason to dismiss useful UZR results out of hand.

I think that what happened to Abreu and Anderson was largely a function of a glut of good bat - no glove corner OFs on the market at a time when the economy was really depressing the market for all FAs. Just supply and demand. Edmonds didn't get a new gig because he seemed pretty insistent on being the primary CF going into the season, and enough teams have wisened up some about evaluating defense that it doesn't matter anymore that he used to win gold gloves. On the original topic, I like Morgan's defense but I'm wary of any hitter that needs to hit over .300 to be a useful offensive player. His base-stealing would be nice if he didn't get caught so often, but so far in his MLB career he has gotten caught often enough that his teams would be better off if he never ran at all.

Now I'm really sorry I missed it. And all because somebody evidently went to the Jim Plaschke school of one-sentence paragraphs. Because, as everybody knows, the mark of truly great writing is its capacity to fail to alienate readers with non-existent attention spans.

That is a terrible idea. If he goes to arbitration, the arbitrator picks between just two choices -- the salary the team is offering and the salary the player proposes. If the team lowballs an arbitration offer, that's a great way to wind up paying the player whatever he wants to make. If I'm Kendall and they offer me $1 mil, I'm jumping for joy, because that's a guaranteed win for me in my arbitration case even if I ask for something pretty unreasonable like $7 mil (maybe more?). The only question becomes, "just how high can I push this before my request looks even crazier than their offer?" This is going to be a very strange offseason. A number of veteran FAs took pay cuts or crappy one year deals last offseason, and it's not like the economy has turned around in the interim. I think that if teams are as loosy-goosy with arbitration offers as they have been in the past, a huge majority of the marginal veteran FAs are going to accept arbitration, because the free market is no longer likely to produce a better result for them. But I think the teams sorta know this and will probably anticipate it, and they'll get ahead of the game by simply not offering arbitration to anybody unless they're certain he wouldn't come out of his arbitration hearing massively underpaid. I am virtually certain that it will not go down like it has in the past, i.e. everybody is offered arbitration, and only half a dozen or so guys accept because everybody figures they can do at least as well (and probably a lot better) on the open market as they can with the arbitrator. Edit: I owe FTJ a coke. Curse my long-windedness.

Well, the same thing would have been true if they had kept LaRoche. I don't get this from the Braves' perspective -- there isn't much if any difference between these two guys and Kotchman is still under team control at least through the end of next season (might be two more -- fangraphs says two, cot's says year two of arbitration was this year) while LaRoche is a free agent at the end of the year who would be wise to accept arbitration.

I bet the difference between Hardy and Hart salary-wise next year isn't very much. Based on what they signed for this year to avoid arbitration, and the fact that neither one of them is going to have the kind of year statistically that earns an unusually large raise, you're probably looking at about $6-6.5 mil for Hardy and $5-5.5 mil for Hart. Escobar does make Hardy easier to move, but Hardy's defense at a difficult position makes him much more likely than Hart to substantially outperform his cost.

Sanchez has an $8 mil club option for next year. Hard to see the Giants parting with Alderson unless they plan on picking that up. As for the type A / type B question, mlbtraderumors estimates that he is a type B right now, but he's close to the top of the type B range, and as far as anybody knows Elias only uses the last 2 seasons of stats for their rankings, so it's very difficult to guess where he'll fall after next season.

And I guess all I'm saying is that while it probably isn't actually true that "individual IP are independent events with no carry-over or causal relationships", there isn't any evidence that we can gain anything by using the kinds of data you're talking about (batter vs. pitcher, recent performance, etc.) to help make decisions about who should play and who should sit. As a consequence, it is my belief that unless a manager is aware of one of those things you mention (mechanical problem, confidence issue, non-hidden injury, etc.), the best thing he can do is completely ignore all the small sample size detritus that most of the other managers are using to justify dumb player usage decisions, and thereby gain a small competetive advantage. As fans, it is highly unlikely any of us would ever be in a better position than the manager to assess whether a short-term performance spike upwards or downwards has a specific identifiable cause, so we are much better off just assuming that such fluctuations are just dumb variance. The majority of the time (perhaps the vast majority of the time), that is all they are.

Also, check out this article at THT that uses PitchFX data to investigate the accuracy of the claims Bannister makes in the interview.