Bill,

 
In your Historical Baseball Abstract of 2001 you listed Frank Thomas as the best major league player of 1997 (p.312). I felt that it was Mike Piazza who truly had a historic year, the best year of any catcher in the history of baseball.  His average (.362), OBP, SLG, OPS was the highest ever for a catcher.  His 201 hits is the only time a catcher ever had over 200 hits in a season. He scored over 100 runs, had 124 RBIs and his 40 HRs was only one shy of the catcher's season record at the time.  All of this playing catcher full time in the toughest park to hit .300, and Thomas added little defensively to his team as he was DH much of the time.  You had both of them equal with 39 Win Shares.  I realize that you had not considered Loss Shares when you wrote this.  Would that consideration now change your mind?  By the way, Mike should have been MVP that year also, not Larry Walker.

Jim Ammon

            Your analysis appears to be correct on all points, or at least it coincides with Win Shares and Loss Shares.   I have Thomas with a Won-Lost record for 1997, as a hitter, of 25+4 (meaning 25 Win Shares in the space alloted for 21 decisions), but 1-4 as a fielder, thus 26-0 overall, whereas I have Piazza at 26+5 as a hitter, 4-2 as a catcher (his best defensive year, other than his rookie year in 1993), making a total of 30+3.    Walker did not have a good defensive year, and he comes in at 24+2 as a hitter, 2-4 as a fielder, 26-2 overall, so they would appear to rank 1. Piazza, 2. Thomas, 3. Walker.

            Piazza’s team, the Dodgers, finished two games out in 1997 and did not make the playoffs, but still finished with a better record (88-74) than the White Sox or the Rockies.  This poses the question of why Piazza was undervalued by the MVP voters, and obviously I have to include myself in that issue, since I also apparently undervalued his contribution. 

            For one thing, although Park Effects are by now well known and well documented, I think there is still some tendency to under-compensate or under-adjust for them, and also sometimes to adjust for them in incorrect ways.  I think a well-designed study could show that this was true.   I did a study a few years ago asking this question:   how much has the tendency of teams to build “opposite” their park diminished over time? In other words, it is easy to show that teams that play in good hitters’ parks tend, on average, to have poor offenses but good pitching staffs, while teams that play in pitchers’ parks tend, on average, to have good offenses but poor pitching.   This happens because teams evaluate their pitching and hitting by comparing their pitching and hitting stats to those of other teams in the league, without park adjustments or with inadequate park adjustments.  Teams that play in pitching parks tend to score relatively few runs but allow relatively few runs, thus tend to focus on improving the offense, while tolerating relatively mediocre pitchers because those pitchers have good ERAs because they play in a pitcher’s park.    The opposite in hitters’ parks. . .teams in hitters’ parks tend to think that they have good offenses, but always need pitching.

            This has always been true and it is easy to show that it is true, but the question I asked in this study, about 2002, was whether this had changed over time.   As we now have better data on park effects, have teams learned to filter out park illusions in building their teams?

            My conclusion was that it had changed over time—it had gotten worse.   The tendency to build opposite the needs of your team was stronger in the 1990s. . .this is just my memory of an old study that was never published, so I could be wrong. . .but I believe that the tendency to build opposite the actual needs of your team was stronger in the 1990s than it had ever been.  

            I think there are other ways you could show that, while we are more aware of Park Effects now, we still don’t do a very good job of mentally adjusting for them.   We still tend to give the MVP Award to someone who played in a hitters’ park, and the Cy Young Award to someone who played in a pitchers’ park.

            In 1997, when he won the MVP Award, Larry Walker played in Coors Field, but hit 29 home runs on the road, as opposed to 20 at home.  This made it easy for Walker’s advocates to dismiss the park as having any impact on his numbers. . ..”look at the facts,” they could say.   “Coors Field didn’t inflate Walker’s numbers at all.   He hit .346 on the road, with 29 home runs.   His OPS was higher on the road (1.176) than it was at home (1.169).”

            That’s true—but it has really nothing to do with Walker’s value.   We don’t adjust Walker’s value because Walker “would have hit” more or less in some other park.   We adjust it because the runs that he creates have less impact in a context where runs are more plentiful.   The Rockies scored 545 runs at home, 378 on the road, and allowed 501 at home, 407 on the road.   Obviously, that’s a run-inflated context which makes each run contributed by Walker less valuable—regardless of how Walker, as an individual, was effected by the park.   The park factor for Coors Field in 1997 was 1.33, which was way down from the previous two seasons (1.64 and 1.72), whereas the Park Factor for Dodger Stadium was .823, which was up from the previous three years.   With Park Factors of 1.33 vs .82, obviously the runs created by Piazza had significantly more impact.  The voters perhaps failed to adjust adequately for this—and I perhaps failed to adjust adequately for this.

            Second, since we lack (or lacked, in 1997). . .since we lack really good, day-to-day measures of fielding performance, we tend to think about fielding in simplistic terms.   If a player who usually hits .270 with 18 homers has a season in which he hits .300 with 30 home runs, we recognize this fully, and we evaluate his hitting the same for that season as if this was his normal performance.   In fielding, because we have relatively crude measures of defensive performance, we lack the ability to make the same adjustments.  Fielding is evaluated much more by reputation and much less by performance data. 

            Piazza in 1996 allowed 155 stolen bases, with only 34 caught stealing, a horrific record which helped to cement his reputation as a poor defensive catcher.   He had a better year in 1997, throwing out 43 runners while allowing “only” 112 stolen bases.   This is not a good performance, but it was not the worst in the league, either; Chris Widger of Montreal allowed 116 stolen bases with only 23 caught stealing in only 85 games at catcher.   Still, because of his poor defensive reputation, his defense was held against him by many MVP voters, perhaps to an unjustified extent.  

            Walker was the opposite.   Walker had a reputation as a very good fielder, which in fact he had been.   He didn’t really have a good defensive season, recording only 232 putouts in 1,248 innings in the outfield, but. . voters still thought of him as an outstanding fielder.  

            Third, there is the issue of how much weight should be placed on defense.   On this issue, it is my opinion that a compelling logic has yet to emerge.   In other words, I don’t think that we have this one figured out yet; I don’t think that those of us who study these issues really have any solid basis to say how much weight should be placed on defense relative to offense in evaluating a fielder at any position.   John Dewan has a method he is working on that relates to this issue, and perhaps that will move the ball forward a little bit.  

            In Win Shares and Loss Shares, a player’s defensive responsibility is based on

            1)  His outs as a hitter, and

            2)  His innings played as a fielder.

            In other words, Frank Thomas or David Ortiz doesn’t get a pass, as a fielder, because he doesn’t play the field.   If he’s using up outs for the team, he is incurring a defensive responsibility, which he is unable to discharge, typically resulting in a defensive won-lost record for the season of 0-3.

            Players have a defensive responsibility, and they have defensive credits.   Those defensive credits pile up more rapidly for a catcher or a shorstop than they do for a first baseman or a left fielder.   The result is that “up the middle” defensive players tend to have defensive winning percentages over .500, while “corner” defensive players tend to have defensive winning percentages under .500.   For his career Piazza—a poor defensive catcher—winds up with a defensive won-lost record almost the same as Walker, a very good defensive right fielder (36-37 for Piazza, 36-40 for Walker.) 

            That was the best system I could come up with, but it’s not perfect, and not all the corners of it are grounded in actual knowledge.   Thus, it could be that, in another generation, we will look back on this debate, and realize that we didn’t quite get it right.