The Gauss, the Gomes, the Blass

I love this post, although Shawn Green remains a fresher example in my mind of the Gomes.

Swing Analysis

Jeff Albert has posted guest articles on Andruw Jones and Alex Rodriguez over at Baseball Analysts comparing their swings in different seasons. I understand the appeal of this kind of analysis, but my guess is that it's dubious. Firstly, I'm not even clear on how representative the swing captures are - do all of A-Rod's swings in 2006 look like the pictured? Are we seeing relatively extreme examples? What is the standard season-to-season variance of the appearance of a player's swing? Perhaps these are readily answered questions, or even questions that people who watch more baseball than I could answer without much thought. But I don't know the answers, and I think that those questions are essential before launching into a side by side comparison of the video.

Second, I have a major problem with any analysis that attempts to posit an explanation for statistical variance by looking back at what differences can be seen in how the player played. Now, if intensive scouting data is kept on a player, and you use that scouting data in concert with a) an understanding of the scouting data for all hitters and b) an understanding of the empirical relation of scouting data to performance results, then you can probably learn a lot. But if you ad hoc notice a change in statistics and decide to get all Jake Gittes on a few minutes of video, I doubt your findings can have any degree of confidence.

Looking at their statistical records, it seems entirely reasonable to assume that the performance records in question are the result of variance, pure and simple. A-Rod is on the other side of 27 and is still an excellent hitter; he went from a 4- to 5-win hitter in Texas to a 3-win hitter in 2004 and 2006, with an outlying 2005. His strikeouts have increased a bit in each year in New York, which is to be expected at his age. His $HR came down when he left Texas, as one would expect, and has been the same in 2004 and 2006. The only big deal is that he hit an extra dozen home runs in 2005, right? Since 2005 was his career year, why would anyone expect him to repeat it at 30 in 2006? It's perfectly fair to say his true talent level is 40 HR per 600 AB, which he undershoots by a few in '04 and '06 and overshoots by 8 in 2005. It just seems like his 'slump' only amounts to not getting the extra HR every four weeks that he got last season.

Meanwhile, Jones' 2005 doesn't seem like much of a big deal either. It differed from his established performance levels in that he cut his K's down to a level he hadn't been at since '99-2000 and he hit 15 extra HR. But the difference in HR looks mostly like his doubles turning into homers for one season. From 2002-4, he hit 15.3 extra base hits per 100 batted balls, and has 15.8 in 2006. In 2005, he had 16.5, so we're not talking about a huge difference in power; the only difference is that he had a good year for getting them all the way into the seats. It also didn't hurt the breakout aura that his 2004 season was below his established levels. Now, is it possible that he played differently and his results reflect a different approach? Of course. Indeed, his hits per batted ball were down 25 points last season, so perhaps he was lofting balls more often, resulting in more hits in the stands but also more hits in OF gloves. But the explanation that the differences in numbers is just statistical variance has as much or more merit than the guess that he made mechanical changes to get more air under the ball.

I do not mean to imply that we should just accept 'statistical variance' as an acceptable answer for why player x is struggling/doing well and so forth. But it should be the default explanation.
A player who goes from 3 to 11 HR in successive seasons has not necessarily bulked up, and a player going from 35 to 50 has not necessarily made any changes. Moreover, I am willing to believe that a great deal of statistical variance *can* be explained by looking at changes in a hitter's mechanics (although clearly not all variance can be so explained). Heck, it's of course possible that, due to changes in his swing, A-Rod in 2005 was a true .350/.460/.680 hitter whose observed performance didn't live up to his talent, and that in 2004 and 2006 he's really a .270/.360/.460 hitter whose observed performance is better than his true talent. The point is just that taking the performance data and trying to clarify it by cherry-picking the scouting info that would explain the fluctuation in performance data is foolish. That may or may not be what Jeff Albert (or Don Mattingley, or Alex Rodriguez) does. What I am arguing is that the task should be to first come up with a useful and reliable method for cross-referencing scouting and performance data; since one will never have a perfect sample of either player performance or scouting info, picking and choosing just doesn't seem like a solid method of learning more about a player.

WPA and Overcrediting Relievers

I don't find WPA as interesting as most of the saber-oriented internet community seems to at present. I don't think it's worthless, though. In any event, I've noticed that a very common criticism is that the distribution of credit is heavily skewed toward relievers. While I understand the sentiment and I've read several discussions regarding this point, the argument that I don't recall having seen is that this criticism simply isn't unique to WPA. Run average and ERA already do the same thing, no? It is much easier to pitch in relief (or, rather, it is much easier to be used as major league relievers are used than to be used as major league starters are used; a starter who pitches for one inning would have an easier go than a reliever pitching innings two through eight). Any metric which does not compare relievers to relievers and starters to
starters (and hitters to hitters) will 'favor' relievers.

As far as I can tell, all of the WPA compares relievers, starters, and hitters to the same baseline of R/PA, and since the task for relievers is the easiest of those three, they will come out on top, just as they will in nearly all statistical categories. Or at least, all of the models that use theoretically-derived WE tables seem to apply the same run environment to all players; BP or maybe somebody else might be using empirical WE data, which has its own set of sizable flaws, and BP probably does all of its WE stats against a replacement level, which negates all of the benefits of using the empirical values anyway.

Now, WPA does 'favor' *some* relievers relative to all other players in that a reliever finishing a game needs only succeed in walking away with the game to receive a positive score: a closer who enters with a three run lead in the ninth, walks the first five hitters, and finally induces a triple play groundout will receive a positive WPA value despite pitching very poorly. But so what? Similar quirks exist in any baseball statistic I can think of, and it's not WPA's fault that people treat it as if it were a pure measure of value. There are infinite ways to measure value in baseball, and expecting WPA to be a reasonable stand in for all of them is silly. WPA is Win Probability Added, not Player Goodness. So while WPA's usefulness derives from adding more context to the raw numbers, many of the criticisms of it derive from people's willingness to strip WPA of its context. As with any other metric, it will not tell you anything meaningful about player value across contexts; WPA cannot possibly prove that Jonathan Papelbon has been more valuable than Tim Wakefield. It can only tell you that, comparing the two to the same run environment, Papelbon has added more win probability. Anytime numbers are removed from their context the ensuing comparison will be 'unfair.' So if the problem you have with WPA is that relievers who pitch poorly with a lead in the 9th get more credit than hitters who jack doubles in the fourth - well, why are you even using WPA to compare them in the first place? The only comparison WPA can make is Win Probability Added.

RBI is not a flawed statistic because it credits Joe Carter more than Bobby Abreu; people who use RBI to argue that Carter had more value are flawed. The same holds for WPA.

Height, Weight

David Gassko checks in with a provocatively-titled essay on player size at THT today. I imagine that where David is headed with this is toward how to incorporate player size in regression analysis to improve player projections (if, indeed, player size can improve projections). However, since he's merely presenting rawer data right now, I have to say that there are some different data slices I would like to see other than what he has presented in part 1.

In part 1, Gassko included statistical averages for players in four categories: short and light, short and heavy, tall and light, tall and heavy. While David is very up-front in discussing the extreme nature of sample bias in this discussion and does a good job of explaining the caveats, there are still two problems with the sample that I can think of. First, the samples seem overly small; by limiting the data to 6'3" and above against 5'10" and below, we're not really looking at the question of whether size matters but rather the general tendencies for the exteme groups. Since apparently Kirby Puckett makes up half of the sample for the short and heavy group, I question how much can be learned from this data presentation. Second, I would imagine that there is a much higher proportion of taller and heavier players from the current epoch, which is of course the highpoint in post-WWII run scoring. Since there isn't, as far as I know, any overwhelming evidence that the increase in run scoring (or OBP/SLG, etc.) is do mainly to increasing player size, this should easily skew the results. I may be wrong, and we may be in the golden age of short and light players and it's just slipped by me, but I imagine that my guess as to the trend in player size is correct.

The solution to both of these problems, if I'm thinking straight, is to analyze player size using standard deviations from the league norm, both for size and for the offensive metrics being considered. I think this probably gives us a better snap shot of the relationship of player size to hitting. I'll readily admit I could be wrong.

David uses Jeff Francoeur (the spelling of which is, to me, immensely more difficult to remember than Saltalamacchia) as an example of a player who should be able to improve because of his hulking size. I don't doubt that that's a reasonable expectation, but it's not as if the primary reason for his struggles is that he's not hitting the ball hard enough. I mean, certainly if Francoeur develops enough strength and bat speed that he's able to crush just consistently crush the ball, his difficulty in strike zone judgement won't matter. But, to look at a very different but nearly as large player, being a big dude hasn't helped Jack Cust become much better than he was at age 22. (That doesn't mean that, were I GM of an AL team out of contention, I wouldn't seriously consider acquiring Cust to determine once and for all whether he can hit major league pitching as half of a DH platoon; I know it's the PCL, but the guy's got a .470 OBP for chrissakes!) I don't know that simply being big is enough to help out a player with strike zone issues as extreme as Francoeur's or, for that matter, Joel Guzman.

And of course, I'm also looking forward to David giving a good picture of how height and weight can help in projecting pitchers.

Calling BS on Reactions to Luna-Belliard Trade

At the time, I didn't really think much of the St. Louis-Cleveland trade of second basemen. Since then, I've seen it panned for ostensibly 'sabermetric' reasons, and I just don't buy it.

The arguments have been that Luna is an equal or superior hitter and is young while Belliard is older, more expensive, and on the verge of free agency. Four years of Luna versus two months of Belliard. Sorry, but I don't think four years of Luna is worth much.

Luna is, of course, having a fine season offensively. Rate-wise, there's been no difference between the two in 2006. In fact, the same could be said of their major league numbers in 2005. There are, however, two obvious problems with only looking at one or two years of data here because Belliard has been a major league regular during that period while Luna has not. First, Luna has logged fewer than 400 PA in that period while Belliard has 959. Second, Luna has had the platoon advantage in 47% of his PA against Belliard's 30%. A wiseass might argue that that's not very relevant (Szymborski kind of did at the Transaction Oracle) because Luna hasn't hit that much better against LHP in 2006 (which isn't particularly true) or because Belliard has had a reverse platoon split in that time. Since those platoon splits are only *samples* of performance, it doesn't matter if the platoon splits don't themselves support the argument, since we know that their *true talent* against LHP is higher than against RHP; if Belliard were to have more of his PA with the platoon advantage, we would expect his performance to improve significantly, and vice versa for Luna. In other words, even if they'd had the same total number of PA, Luna's rate stats wouldn't be nearly as impressive as Belliard since he has been substantially platoon-aided.

Beyond the platoon issue, we actually know quite a bit to indicate Belliard's true talent level. Belliard has clearly been a solid hitter throughout his minor and major league careers. I don't know if there were mitigating circumstances involved in his awful 2002 season, but even if we assume that that was a non-problematic sample of his true talent, all the data available indicates that Belliard's 2005-2006 campaigns are about what we should expect.

Luna, on the other hand, has a lot of the rest of the data going the other way. Luna wasn't much more than average at any of his minor league stops and wasn't young for any level. And at age 25, he hit a terrible .224/.294/.332 in the PCL. He had as many PA in that stint as he's had in 2006; is there a major reason to consider his 2006 a more meaningful sample than that that I'm just not aware of?

I'm not arguing that Luna is, incontrovertibly, an awful or even below average hitter. It is, though, completely reasonable to project that the true offensive difference between the two amounts to 5 to 10 runs over the balance of the season, and I consider it thoroughly unreasonable to think that Luna is the superior hitter. The difference in salary between the two will amount to little more than $1m, so unless Luna is the superior defender - which I doubt - then the Cardinals get a pretty good deal on a marginal win that could make a difference in getting into the playoffs. So in terms of 2006, the talent and dollar tradeoff is just fine for St. Louis.

Beyond 2006, the tradeoff is 4 years of Luna at arbitration prices (he'll be a super-2 this offseason, right?) against the draft picks they'll accrue from offering Belliard arbitration (unless they re-sign him). I don't know how the Elias formula works, but Belliard has nice enough numbers that I'd be very surprised if he isn't above the Type A threshold. That means the Cardinals would get a sandwich pick and a pick from the second half of round 1 or the first half of round 2. Even though one must pay bonuses to draft picks, a team that is as confident in their ability to select players in the amateur draft as the Cardinals appear to be under Luhnow has to attach significant value to additional early round picks. Luna will probably come relatively cheap for his value over the next four years, but even if he is, say, -5 runs per season below average for a second baseman over that time (a projection which is probably generous, I'd say), he'd likely earn in the neighborhood of $7-$8m, which amounts to paying ~$1.5m per win over replacement, which isn't exactly a steal. It's not bad value, but if you were to put a dollar value on having Luna as an asset going forward, it probably shouldn't be much more than $2m. Sure, one could quibble and argue that he'd be a better value for three years and then a non-tender, but the point is that he's not an important or particularly cheap piece. And the flip side is that if he's only used as a backup, he'll likely be overpaid. The value of a high draft pick to a major league franchise is something I haven't researched, but my WAG is that the additional draft picks offset and perhaps transcend Luna's value.

Even if you decided that the Cardinals got no added value from having those draft picks, I think I'd still be on board for St. Louis making this trade. If we say they get .8 marginal wins for a $3.2m price tag, they're paying $4m per marginal win, which is reasonable given the race they're in, and they also improve the postseason roster. Additionally, a player like Luna isn't exactly the kind of player that St. Louis needs to have around; this is a team that's trying to make a World Series run each season and has a pretty high payroll, and you'd hope that they would only be accepting below average production at one or two positions on the field going forward; though they haven't done a particularly good job at it this season, you'd think in the future they'd be able to find those below-average pieces at value better than what Luna represents.

The final piece that baffles me is that Cleveland has caught flak on this deal because it's a reminder they let Luna go in the first place. While I could understand arguing that Cleveland should have picked up something with more upside than Luna or simply should have waited to collect on the draft picks, I find this particular critiqur quite ridiculous. Christina Karl argued at BP (subscription required) that:

Luna was a player the Indians developed in the first place, and lost to the Rule 5 draft not once but twice, in consecutive years (2003 and 2004). That Chuck LaMar didn't have the sense to keep him on the D-Rays the first time around is no defense, and Shapiro needs to be taken to task for protecting guys like Corey Smith and Ivan Ochoa on the 40-man in his place, and then finally making a deal to undo a past mistake, instead of dealing Belliard for somebody he shouldn't have let go in the first place. It's sort of like the 2004 deal in which Dave Littlefield patted himself on the back for getting Jose Bautista, instead of expressing sufficient regret for blowing it by exposing him to the Rule 5 draft the previous winter. Shapiro made the same mistake, and now, instead of having Luna and something for Belliard, he's got just one of those things.

This is a pretty big head-scratcher. I don't know all about what Cleveland protected on their 40-rosters back then, but who, exactly, would have been arguing in favor of protecting Luna over Smith or Ochoa? Smith and Luna played together for Akron in 2003 and had equal value as hitters in that season; Smith was two years younger and a former 1st-round pick, and he'd been as good in his previous seasons. Though Cleveland subsequently traded him for Gautreau, he's actually having a good year for Birmingham and is still only 24, so it's not like he was some obviously flawed player kept around just because he was a former 1st round pick. Ochoa is less obvious, but was at the time a 20-year-old shortstop coming off a bad but not terrible season in high-A, and almost certainly represented a higher celing than Luna. The notion that Luna, coming off an uspectacular age-23 season in AA and without any impressive track record, was someone who needed to be protected is pretty out there. Given how poorly he played for St. Louis in 2004 and, if you factor in his Memphis performance, 2005, I don't see why anyone would assume that Luna was likely to stick on a 25-man roster or why anyone should be crying about not donating a 40-man roster slot to someone who wouldn't contribute until August 2005. You have to be unrealistically high on Luna to take him to task for that.

Even more disappointing is the gratuitous LaMar put down. A book may eventually be written about how poorly LaMar did as a GM, but if that book has an editor any references to Luna will be left on the cutting-room floor. LaMar didn't have the sense to keep around a 23-year-old who had yet to hit AA and managed just .276/.334/.361 in the Carolina League? That criticism is thoroughly unwarranted, and I can imagine that a list of players who LaMar could have acquired in the Rule 5 draft over the years would not include Luna as one of the more valuable. Luna was sent back because the Rays deferred to Rey Ordonez, Marlon Anderson, Brent Abernathy, and Damion Easley. Hector Luna on a major league roster for Hector Luna back in 2003 would actually make that group look good by comparison. And had Luna survived the first cut, should LaMar have held onto him in May when he otherwise would have acquired Julio Lugo?