Saturday, January 24, 2004

Well, unfortunately I've been idle with my writing for over a week, but that doesn't mean that I haven't been working on my analysis. I've been looking at player evaluation methods for almost the past year now, including reading Bill James' magnum opus Win Shares, and I'm convinced that this is the most thorough player evaluation method available for several reasons.

First, it is the only method that breaks down total player contribution to his team's success by dividing "Win Shares" between pitchers, fielders, and batters in direct proportion to the number of games the team has won. The only other method, that I am aware of, that considers fielding is Pete Palmer's Linear Weights System.

Second, Win Shares is the only method that considers a player's ability to be good enough to earn major league playing time as a positive value. The Linear Weights System uses a zero value as its starting point, with each player's contribution measured as either a positive or a negative from this base point. In other words, a player who is good enough to earn considerable major league playing time, but is rated below average by this system, actually earns a negative value. On the other hand, a player who sits on the bench (i.e. doesn't play at all) remains at a zero value. In fact, Steve Garvey rates as -5.2 for his career. This is absolutely ridiculous. Even those of us that suspect, or are convinced, that Garvey is overrated would never even suggest that he was a below average player and that the composite value of his career was a negative. This would suggest that he actually hurt his team more than he helped it, which is a ludicrous notion.

Third, I am convinced that James' system does the most thorough job of building in adjustments for such contextual factors as the era in which a player performed, the contributions of the players on the teams for whom he played, the strength of the competition, and the ballparks in which he played. Linear Weights considers these factors, but I've already pointed out its major shortcoming for which I'll admit to being completely unforgiving. All other player evaluation methods are essentially offensive-oriented systems that express performance on a rate basis, and do not make these important contextual adjustments.

I probably should explain what I mean by "express performance on a rate basis". Batting average is the most basic example of a statistic that measures performance as a rate. There is no consideration given to the player's ability to accumulate statistics. In order to evaluate a player's career, or to compare the careers of several players, we must consider both types of statistics, rate stats and cumulative stats. By doing so, we assign value to both the player's overall ability to perform and his ability to earn playing time, a significant factor of which is his longevity.

There are many rating systems out there that have done an excellent job of expanding on the shortcomings of batting average as a way to evaluate players. OPS (on-base percentage plus slugging percentage) is a simple, but fairly effective, way to measure a player's overall offensive ability. Other much more complex methods have been developed, but they are all offensive rating systems that are only effective at measuring a player's ability in a given season, or across a few seasons, because they do not place any weight on cumulative statistics. They are not designed to do so, and, therefore, are not as relevant to evaluating a player's overall career as the Win Shares system is.

The next part of my analysis will be to compare these four players using the Win Shares system. Despite my endorsement of Win Shares as the ultimate player evaluation method, there still are several ways to use the system to compare players. I'm not just going to state that Tony Perez earned 349 career Win Shares to Keith Hernandez's 311, Steve Garvey's 279, and Don Mattingly's 263, and, therefore, Perez is the best, Hernandez the second best, etc. I'll compare them using a few Win Share rate statistics as well, although their cumulative Win Shares certainly will be taken into consideration as a major factor.

For more information on the Win Shares system, you'll have to locate a copy of Bill James' book, Win Shares. Unfortunately, it's already out of print, so you'll have to look for it at your public library or try to find a used copy on the Web. Pete Palmer's Linear Weights System is explained in detail in Total Baseball: The Official Encyclopedia of Major League Baseball. Other statistics, such as OPS, on-base percentage (OBP), and slugging percentage (SLG), are explained in detail in the Batting Stats Glossary at Baseball-Reference.com.

Thursday, January 15, 2004

Since I ended my last post by bringing up the fielding situation, let's begin today by taking a look at how these four compare in that department. I'll start with a non-statistical discussion of their reputations as fielders.

Keith Hernandez is considered by many to be one of the greatest defensive first-baseman of all-time, having won 11 consecutive National League gold gloves from 1978-88. Don Mattingly's reputation is almost as great. In fact, he took over the throne as the most proficient at the position as Hernandez reached the latter part of his career. Mattingly won nine American League gold gloves in the ten year span from 1985-94, only losing out to Mark McGwire in an injury plaqued 1990 season.

Steve Garvey also was considered to be an excellent first-sacker, winning four consecutive NL gold gloves from 1974-77, just prior to Hernandez's run. Garvey is generally considered to be not quite as good as either Hernandez or Mattingly, particularly because he had a sub-par throwing arm and was less proficient at turning the double play. Hernandez and Mattingly have a distinct advantage in that department over Garvey, both being left-handed throwers, but the fact remains that first base, the least important defensive position on the field, was the only position that Garvey was capable of playing, despite not being limited to what positions he could play by his throwing hand. In my original discussion of Steve Garvey, the debate that prompted me to perform this analysis, I stated that I felt that Garvey was overrated defensively as a first-baseman. I am going to retract that statement and evaluate Garvey's defensive merits at face value, those being that he did win four gold gloves, and was probably only prevented from winning additional gold gloves by one of the greatest defensive first-basemen of all time.

Tony Perez is basically considered to have been an average defensive first-baseman. He never won a gold glove, but he certainly wasn't considered to be weak at fielding his position.

Now let's take a look at the defensive statistics that are available. First, I'm going to compare each player's fielding percentage and range factor to the league average at first base for his career. For a full explanation of the fielding statistics used here, see the Fielding Statistics Glossary at Baseball-Reference.com.

Player                      G    FP    lgFP    RF    lgRF
Steve Garvey         2059  .996  .992   9.65   8.51
Keith Hernandez     2014  .994  .992   9.73   8.51
Don Mattingly         1634  .996  .992   9.33   8.41
Tony Perez            1778  .992  .991   8.67   8.45

What do these limited statistics tell us? Essentially, Garvey and Mattingly rate tremendously in terms of fielding percentage, while Hernandez also rates well, and even Perez is above average. In terms of range, Hernandez rates the highest, but Garvey and Mattingly rate almost as well, and Perez, again, rates just above average. However, I'm not convinced that this method of evaluating a first-baseman's range is very effective, considering that it factors putouts into the numerator of the equation. What do putouts have to do with a first-baseman's range? Very little, in fact, with the possible exception of unassisted putouts, but these are difficult to capture. I'll have to examine this further if I have the time, but right now I'm not sure that it's worth it.

Overall, these statistics show us nothing that helps us distinguish between Garvey, Hernandez, and Mattingly. All three of these first-basemen back up their reputation with outstanding defensive statistics, although the value of one of these statistics is somewhat questionable. They all rate higher than Perez, who rates average to slightly above. I do plan on revisiting this defensive analysis, but right now I'm not convinced that statistics can really be used effectively to evaluate a first-baseman's defensive ability. I need to study Bill James' work on this subject a little further before I can draw any valuable conclusions.

In the meantime, I'm going to stick with the conventional wisdom, and my personal opinion, that Hernandez and Mattingly are two of the greatest defensive first-basemen of all-time, but that the difference between them and Garvey is not very significant considering the relative importance of the position that they play. Obviously, being a brilliant defensive first-baseman falls far short of being a brilliant defensive shortstop. Even the difference between their ability and that of Tony Perez may not be that statistically significant. I hope to shed more light on this subject later in this analysis, but I think it's time to begin a more in-depth look at offensive statistics.

Tuesday, January 13, 2004

I'm going to begin my comparison of these four first-basemen, who essentially represent four of the five best at this position among Hall of Fame inductees and eligible players who played the majority of their careers during my lifetime as a fan. I was born in 1967, and basically became a true baseball fan in 1974, so Tony Perez (1964-1986) just makes the cut, whereas Willie McCovey (1959-1980) doesn't. As I explained in my post of Jan. 11, I'm excluding Eddie Murray because he is a first ballot Hall of Famer and far superior to the other four.

To get started, I'm just going to look at the really basic, or what I like to call mainstream, offensive statistics. These categories are seasons played, games played, at bats, runs, hits, home runs, runs batted in, stolen bases, walks, batting average, on-base percentage, and slugging percentage. Here's how they stack up:

Player                  Yrs    G     AB      R      H    HR   RBI   SB     BB   BA   OBP  SLG
Steve Garvey         19  2332  8835  1143  2599  272  1308   83    479  .294  .329  .446
Keith Hernandez     17  2088  7370  1124  2182  162  1071   98  1070  .296  .384  .436
Don Mattingly          14  1785  7003  1007  2153  222  1099   14   588  .307  .358  .471
Tony Perez             23  2777  9778  1272  2732  379  1652   49   925  .279  .341  .463

So, what do these statistics tell us? Well, most obviously, none of these guys had any speed, so I'm going to disregard stolen bases. Also, in almost every cumulative statistic, these players line up in order consistent with the length of their careers (Perez-Garvey-Hernandez-Mattingly), with the exception being that Mattingly rises above Hernandez in HR and RBI, despite placing last in seasons played, games played, and at bats. Also, Hernandez places first in walks, followed by Perez, Mattingly, and Garvey, but I'll get to that later.

Of course, longevity is a factor, and Perez deserves credit for ranking first among these players in R, H, HR, and RBI. But, I'm interested in far more than determining who played the longest and, as a result, being fooled into rating a player higher simply due to the length of his career. In fact, it's interesting to note that, despite the fact that Perez has clearly superior HR totals to Mattingly (379 to 222 total, 25.8 to 31.5 in AB/HR), Mattingly actually has a higher slugging percentage. This is in part due to Mattingly's higher batting average (.307 to .279), but this still suggests that maybe Perez's power is slightly overrated. However, it's difficult to argue with his career RBI total, which ranks 21st all-time, and places him behind only 17 other Hall of Famers and three players not yet eligible (Barry Bonds, Cal Ripken, Rafael Palmeiro).

Along those same lines, Garvey out-homers Hernandez 272 to 162 in career totals, and 32.5 to 45.5 in AB/HR, yet only edges him slightly in slugging (.446 to .436). This is despite the fact that they are virtually even in career batting average (.296 for Hernandez to Garvey's .294). Does this fact negate Garvey's perceived power advantage? We'll have to look closer before making that determination.

Anyway, I haven't even begun to look at any statistics in depth, and I haven't brought in any of my favorite non-mainstream stats yet. Oh yeah, and the subject of fielding has yet to come up, although I think we'll find that, despite the 24 gold gloves that Hernandez, Mattingly and Garvey can claim between them, the relative importance of the position they play will render this part of the analysis less meaningful than we might think.

Sunday, January 11, 2004

With this past week's Baseball Hall of Fame elections announced, I found myself reviewing the voting results and noticing what I felt to be some serious discrepancies with respect to the level of support that some players received compared to others who are much less worthy. I put more than a little bit of thought into this, but fell way short of conducting a complete analysis of all eligible players, and came up with the following top 5 list of players underrated by the voting (with comments in parentheses):

1. Alan Trammell (I think this guy is a HOFer, but he scores a ridiculously low 13.8%)
2. Dale Murphy (as good as, if not better than Jim Rice and Andre Dawson, but Rice and Dawson get 50+% to Murphy's 8.5%)
3. Don Mattingly (borderline HOFer and only gets 12.8%, un-fucking-believable)
4. Keith Hernandez (almost as good as Mattingly and deserves way better than 4.3%)
5. Rich Gossage (better than Bruce Sutter, but receives 19% fewer votes)

While I'm not going to go to great lengths to analyze and defend all of my picks, I will say that I shared this list with a few friends and was questioned by one for not including Steve Garvey. My response was essentially that I rated Mattingly and Hernandez higher, and that since Garvey received greater support (24.3%), he didn't make the list. He countered by stating that Hernandez once said that you had to vote for Garvey before himself...certainly a strong argument. How could I possibly rate Hernandez higher than Garvey, when Hernandez himself rates Garvey higher? Well, my answer is, that I certainly value Hernandez's opinion with respect to how he rates compared to one of his peers, but no more than I value the opinion of the baseball writers who gave Garvey almost six times as many votes as Hernandez. Therefore, if I'm willing to question the opinions of the voting members of the Baseball Writers Association of America, then I'm certainly willing to question Keith Hernandez's view.

Anyway, over the course of the next couple of weeks, I'm going to do some serious analysis, comparing the careers of four contemporary first-baseman: Steve Garvey and Keith Hernandez, for obvious reasons; Don Mattingly, because I'm also curious as to why he only gets half the support that Garvey does; and Tony Perez, because he is a contemporary of these players (well, at least of Garvey and Hernandez), and because he is a Hall of Famer who wasn't an obvious choice and, therefore, provides an excellent point of reference. I'm not including Eddie Murray in the comparison because he is a first ballot Hall of Famer and easily one of the top ten (if not top five) first-basemen of all time. None of these players would come close to stacking up to Murray.

Before I venture into this undertaking, I have two admissions to make. First, Don Mattingly is not exactly a contemporary of these players. He is, in fact, 19 years younger than Perez, and made his major league debut in 1982, while Perez retired in 1986. However, there is some overlap in his career with all of these players, and my analysis is going to take into consideration such contextual factors as seasons, teams, leagues, and parks.

My second admission is one related to my personal bias. Obviously, I have already tipped my hand by stating that I feel Mattingly and Hernandez are better than Garvey. This would seem to create a bias in their favor, at least in comparison to Garvey. Additionally, considering that I am a Yankees fan, a bias could exist in comparison of Mattingly to Perez, since I'd like to make a case for Mattingly's Hall of Fame candidacy. However, I will say that I am perfectly willing to prove myself wrong, since this would prove to be a learning experience for me, and because this would at least ensure that I would feel certain of my opinions should I enter into a similar debate in the future. Furthermore, since my analysis will be very statistically oriented, with reverence to the work of Bill James, I will certainly make a strong case for whatever my findings turn out to be.