Baseball is different from any other sport. Very different. ..In most sports the ball, or object, is put in play by the offensive team; in baseball the defensive team puts the ball in play, and only the defense is allowed to touch the ball. ----George Carlin
The unusual thing about defense is that the results--runs saved--show up in the pitcher's line. In recent years, defense has been described as an undervalued asset in the baseball market. That means acquiring defensive skill often has less cost per run saved than the cost per run of acquiring offense. However, defense is packaged together with the player's offense; and frequently players with exceptional defensive skill do not have an exceptional ranking on offense. That makes baseball roster construction into kind of a Rubik's Cube. Where can you hide the slugging offensive player where his defensive shortcomings are minimized? Where on the field do you need the best defensive player, even if it's at the expense of offense? What effect does the profile of the team's pitching staff have on constructing a defensive team?
The 2011 Astros were a below average defensive team. But they were far from the worst defensive team, if you follow the advanced fielding metrics. DRS ranked the Astros defense at No. 17 with -9 runs saved and UZR pegged the Astros defense at No. 21, with -16 runs saved. This indicates that the Astros defense cost the team 1 - 2 wins, compared to an average fielding team. This wasn't too bad considering that a team like the Mets lost 6 wins due to defense, based on either metric. Interestingly, the Fans Scouting Report rates the Astros defense worse than the advanced metrics. FSR ranks the Astros at No. 27, with -25 runs saved. Did the Astros' fielders just look worse than they really were (the Carlos Lee effect)? The Astros lost Clint Barmes to free agency, and he was the biggest Astros' fielding contributor (10 - 14 runs saved). That is the biggest fielding deficit that will have to be overcome this year.
As you can tell, I favor advanced fielding metrics (UZR, DRS, and to a lessor extent, TZ) to evaluate fielding. The Fans Scouting Report can be a useful adjunct. Unfortunately, advanced fielding metrics require a lengthier period (as long as three seasons) for the results to regress to a player's norm. Advanced defensive metrics have much less reliability for shorter periods. This is a problem for evaluating the 2012 Astros, because most of the position players have too little playing time in the majors to provide a reliable sample for accurate advanced fielding results. Even the long time veteran, Carlos Lee, has inadequate playing time at his new position, 1st base, to fully trust his UZR and DRS results at the position. The only way to address this is to factor in some subjective judgements and, if necessary, regress the advanced fielding results heavily.
I project the Astros with a team Runs Saved of -1.4, which would be an improvement over 2011. The results by position: C +1; 1b +0.5; 2b -1.9; SS 0; 3b -8; LF +1; CF -3; RF +9. I probably don't need to tell you this---a lot of uncertainty around all of these numbers.
A position by position assessment follows after the jump.
Catcher defense is sort of a wasteland, in terms of accurate and comprehensive stats. However, we are seeing some cutting edge concepts for new metrics, but they are not captured by the more widely used systems. Jason Castro will be the starter, and he has a good defensive reputation in the minors. We know he has a good accurate throwing arm to control the running game, but the data for the remaining catcher skills is sparse. FSR gives Castro +5 runs and DRS tallies +2 runs, all in a small ML sample. I think +1 runs is reasonable. Chris Snyder is the back up catcher and he has considerably more playing time in the majors. DRS shows +13 over 8 seasons and the fans show +5 over 3 seasons. Given that Snyder is coming back from back surgery, I will assume that he is average (0). Snyder is above average at pitch framing, according the studies of that characteristic. The data for Castros' pitch framing is too small a sample to make a conclusion. Overall, I assume that the catcher position will be +1 run saved.
Carlos Lee began playing first base in 2011, but he only played 614 innings at the position. His advanced metrics at 1st base are quite good-- + 5 to +10. However, given the sample size, the results probably will regress toward average. I would expect something like 0 to 2 runs saved. Carlos is an iron man and isn't likely to require much back up at 1st base. Presumably either Chris Johnson or Matt Downs will be the reserves at 1st base. We don't have any real data to evaluate them at 1st base, but I would expect that either player would be below average at 1st base, since they have little experience at the position. Overall, I assume the first base position is +0.5 runs saved.
Jose Altuve's fielding reminds me of David Eckstein. Both are small guys, and Altuve puts his all into the throws, like Eckstein. Altuve's height may be a slight disadvantage in the field, particularly on high choppers toward the second baseman. However, Altuve's speed and energy will make up for any disadvantage. Again, the sample size is small for Altuve, and we don't have a clear picture of his defense. DRS has Altuve at +2, UZR/150 at -2.5, and TZ/Year at -5. Averaging the three arrives at -1.8. Over his career, I think Altuve can be average to slightly above average, but given his age and relative inexperience, my instinct tells me -1.5 is reasonable for next season. Matt Downs is the back up for second base, and his defense will be worse than Altuve. Based on 3 years of data for his play at 2d base, the annual average of TZ, UZR, and DRS is -3.5 per year. Overall, second base is -1.9 runs saved.
Jed Lowrie is hard to predict, given that he has played a number of positions in the majors. The annualized average of Lowrie's DRS, TZ, and UZR/150 at shortstop over the last four years is +1.7. I'm not sure that it means anything, but the Fans Scouting Report is -2 per year over the same period. Lowrie doesn't appear to be a "wow!" defensive player, and that may affect the fans' report. But Lowrie's spring games suggest that he is a steady fielder. I think slightly above average (+1) is a good prediction. Marwin Gonzalez is the back up at shortstop. The Astros have given good reports to themedia about his defense this spring. We don't have any ML metrics to review for Gonzalez, but his cumulative minor league TZ is negative for the years that are available. I think average (zero runs saved) is a good call for the overall shortstop position.
The odds are that this will be the worse defensive position for the Astros in 2012. Chris Johnson is the starter at 3d, and the advanced metrics have not been kind to him during his ML career. The advanced metrics range from -10 to -20 on an annual basis. There are some positive signs of hope, if you are inclined to look. As I've pointed out before, Johnson is a decent athlete who seemingly has the tools to play 3d base; and his minor league total zone results point to a slightly below average fielder, rather than a horrible third baseman. Matt Downs, the back up, has comparable advanced metrics at third base (-25, UZR/150), but an even smaller sample size. I'll assume Downs is approximately the same in the field as Johnson, and heavily regress Johnson's UZR results by 60%. This results in a -8 runs saved for third base, and recognizes the small sample size and expected regression. In addition, the result is similar to the Fans Scouting Report (-7) for Johnson.
J.D. Martinez is the starter in LF. The advanced metrics liked J.D. Martinez last year (+2, +2, +8, for TZ, DRS, UZR in 436 innings) but it was a really small sample size. The minor league scouting reports on Martinez usually characterized him as a bad defensive player. However, the minor league total zone results point to a slightly above average fielder (+1 per year in LF). Martinez got a boost throwing out runners at MMP, earning him the TCB nickname "arm gun." The assists probably are partially due to the short LF, and may decline as teams become less likely to test Martinez. Considering these conflicting factors, I will project +1 runs saved over the course of a season, which is the same as the fans scouting report, by the way.
Jordan Schafer is the starter in CF, and he has always had the tools to be a good center fielder. But his advanced metrics playing CF for Atlanta aren't so hot (-3, -9, -7 per year for TZ, DRS, UZR). Given how his age, injuries, and (naturally) small sample size may have affected those results, I think there is a chance that Schafer could prove to be an above average fielder in CF. But at this point, the available data doesn't support that kind of projection. Including Schafer's minor league yearly TZ in CF (-3), the average of advanced metrics, annualized, is -5, which I will regress by 60%, resulting in -2. Any of the potential back ups in CF are likely to be worse CF defenders than Schafer, since all of them are primarily suited for corner outfeld duty. Therefore, I will rate the overall position as -3.
Brian Bogusevic is the presumed starter in RF. Like Martinez in LF, Bogusevic had impressive advanced metrics in a small sample last year. Unlike Martinez, Bogusevic had the profile of a plus defender in the minor leagues. Bogusevic, a former pitcher, has an arm that is a major weapon from the outfield. The TZ, DRS, and UZR metrics for Bogusevic support an average -30 annual runs saved. Bogusevic may be a good fielder, but these numbers appear to be inflated by the small sample. Regressing the runs saved by 60% results in +12. Travis Buck and Justin Ruggiano are the likely back ups, with Jack Shuck potentially figuring into the mix. Ruggiano is an above average defender, and his small sample advanced metrics are similar to Bogusevic. Buck is also an above average fielder, according to the advanced metrics. Shuck probably is a bit worse fielder than Ruggiano and Buck. Combining the fielders, and assuming 60% playing time for Bogusevic, I will rate the overall position at +9.