February 10, 2005

Interview with Bill Felber -
the author of The Book on The Book: A Landmark Inquiry into Which Strategies in the Modern Game Actually Work

by Steve Lombardi

NetShrine.com was recently privileged to correspond with Bill Felber regarding his new book due out this Spring - The Book on The Book: A Landmark Inquiry into Which Strategies in the Modern Game Actually Work.  The full exchange follows below:

NetShrine:  Thanks for granting NetShrine.com this interview.  How would you best describe "The Book on The Book"?

Bill Felber:  I'd say it's an effort to explore the current state of knowledge of baseball strategies, both on-field and off-field. I've tried to go into some areas that haven't really been tackled previously, and do it in a way that people will find interesting and informed, but not academic. So many of the analyses I read on baseball take one of two forms, neither of which I like. They're either essentially rants whose tone suggests the writer has listened to too many sports talk shows or they are scholarly treatises that are more suitable for an economics lecture hall than a night table. 

I suspect the more controversial aspects will include:

1. A pitch-by-pitch look at the influence of the count on batting average. The cliché is that the most important pitch is strike one. Not so fast. Based on a sample I conducted of five thousand 2004 plate appearances, here's a number that I think will get attention. With a count of no balls and one strikes, batters hit. 341 with a .523 slugging average. That is not a typo. That's 74 percentage points above the overall 2004 batting average (and also 74 points above the overall batting average for my survey of plate appearances). It's 90 points above the MLB slugging average in 2004 (and 94 points above the survey average). At no balls, two strikes, however, batting average vaporizes, all the way down to .148. In fact, every count situation that involves two strikes puts the batter at a huge disadvantage while every situation that does not involve two strikes puts him at a decided advantage. (This is all said without respect to the base on balls, which obviously can only happen in three-ball counts.) Clearly the data suggests that strike one is very important insofar as it reduces the prospect of an eventual walk. But with regard to swinging the bat, getting strike one on a batter is not at all important. Strike two is the decisive pitch.

2. An examination of the validity of the contemporary rules governing use of starting pitchers. Who would you rather have on the mound, Curt Schilling throwing his 115th pitch, or Keith Foulke throwing his first?

3. An exploration of ways to rate managers for their use of certain on-field tactics. Because so much of a field manager's job involves clubhouse-type stuff and other things we can't really measure, I have been hesitant to draw up any type of managerial rating. I do think, however, there are some tactical skills we can do toe-in-the-water ratings of. Yes, LaRussa consistently rates near the top.

4. A study of team-building strategies. Are you better off putting your money into sluggers or pitchers? The answer: Pay for starting pitching. Since 1999, there have been 11 teams that spent an inordinate percentage (more than 1 standard deviation above the norm) of total team payroll on their starting five: Five of those 11 teams (45 percent) played in the post-season. There were 18 teams that fit the same description for both their five-man rotations and their pitching staffs as a whole: 7 of those (39 percent) played in the post-season. By contrast, there were 26 teams that paid their normal 3-4-5 hitters a total salary that equated to more than one standard deviation above the norm for the period: only 2 (7 percent) played in the post-season. And there were 22 teams that paid their highest paid player more than one standard deviation above the norm. Only one of those teams played in post-season.

5. A historical and contemporary look at competitive balance. What's the true current role played by money in determining the outcomes of pennant races?

6. Earned Value: In terms of winning games, how much money is Alex Rodriguez actually worth? Answer: In 2004, the $22 million man generated about $5.5 million in on-field production for the Yankees.

7. A system for rating general managers. In terms of improving his team, who was the best GM in the game in 2004? Was it Theo Epstein? I measure this by applying Pete Palmer's well-known BFW/PW ratings to the actual moves made by GMs during and prior to the season, and then translating the whole thing to an impact in the standings. I calculate both short-term and long-term ratings. For 2004, short-term considers the impact of all moves on the team's 2004 performance that were made since the conclusion of the 2003 season. Long-term considers the impact on a team's 2004 performance of all moves made prior to the conclusion of the 2003 season. Turns out Epstein did in fact have a great year in 2004 (as did the Red Sox). His short-term contribution to Boston's improvement amounted to 9.5 games. Since the Red Sox qualified for the post-season by a margin of seven games, I'll argue that he was one of four GMs in 2004 who could actually claim to have "won" their teams a post-season berth. (The others were Stoneman, DePodesta and Hunsicker.) By contrast, three GMs' short-term 2004 ratings were so negative that they can be said to have "lost" the post-season for their clubs. The three were Sabean, Hart and (yes) Billy Beane. Losing Tejada hurts.

Gaining Tejada helps. This came as a surprise to me, and will, I'm sure, to those who read this. But the GM in 2004 with most positive impact was not Epstein or Cashman or Jocketty. It was the team of Flanagan-Beattie in Baltimore, who helped the Orioles by 14.5 games. (Think how bad they'd have been before Flanagan-Beattie went to work.) If that sounds odd, consider what they did. They landed Tejada (+6.6), re-signed Mora (+5.1) for three years) and got Javy Lopez (+1.3). Then they had the foresight to eliminate (and inflict on other teams) the following players: Hentgen (-2.0), Fordyce (-1.7), J. Johnson (-1.7), Ligtenberg (-1.0), Moss (-1.0). GMs help their teams both by who they get and who they get rid of; that's what Flanagan-Beattie did. They turned what was going to be a truly wretched Orioles team into a middle of the pack one.

NetShrine:  This sounds very exciting!  What was the reason why you wrote this book?  Was there one particular driving force?  Or, was it more of an evolution of sorts?

Bill Felber:  That's a question with a two-part answer.

Here's part one: I started on this book, in the middle of 1999, out of a sense (first) that interest in baseball was recovering from the 1994 strike and (second) that information availability and computerization was reaching a stage where it was actually possible to answer some questions I thought were intriguing. Can we quantify the effect of the decisions that cause teams to win or lose? Is there some way to determine the degree to which payroll actually influences performance? (You'll recall that in 1999 that was a major issue.) Does it matter whether you emphasize pitching, hitting or neither? How much of a $10 million contract can actually be justified in terms of on-field contributions...in other words, if the goal is winning a pennant, who's overpaid and who's underpaid? What's the objective evidence behind all of the assumptions that govern the modern strategies involving the use of starting pitchers: five-man rotations, pitch limits, arm injuries, wearing down as the season progresses? It seemed to me that where previously most of these questions had been answered with a lot of subjective rhetoric, in fact we were at a point that the arguments could be deconstructed and examined objectively on their merits.

Here's part two. I've been a Cub fan all my life, born and raised in Chicago. When you are a lifetime Cub fan, you are predisposed to seek the answer to the question, "What went wrong this time?" Now I'm closer to knowing. Of course the thing you have to recognize is the limits of objective knowledge. I can attach a number to Sammy Sosa's decline, but explaining it in real-world terms probably has more to do with clubhouse aspects, inter-personal relationships, and the sorts of things it's not possible for me to get a handle on. That's why one of the points I try to make early on is the limitations of even objective knowledge.

NetShrine: The Cubs still may have their year - after all, the Red Sox finally won one.  Do you expect that your findings will change the way people manage their teams?

Bill Felber:  With respect to on-field conduct of a game, my assessment of the potential for change and the likelihood is as follows:

1. The batting count. As a general proposition, I was struck by the degree to which the advantage is with the hitter until the count reaches two strikes, and how dramatically it then switches to the pitcher. (I wonder whether this is linked to the degree to which batters strike out in the modern game rather than shortening up to try for contact.) The problem with drawing any broad strategy-based conclusions from this is that the same rules don't apply to Ichiro and Barry Bonds, and they certainly don't apply equally to Ichiro and Willie Bloomquist. I haven't done the studies for individuals, but I assume it still makes a huge difference who's at the plate. If I were a manager looking at this data, I'd have to consider which of my hitters -- it would be the ones with good strike zone judgment -- I could encourage to be more aggressive on the first and second pitch. Certainly against a control pitcher, this data re-enforces the notion that you want to go up there swinging. But I don't think that's a message I want to send to, for example, Sammy Sosa.

Will managers actually do that? What I expect them to do is take a harder look at the batting average breakdowns for each hitter at each stage in the count, and draw lessons appropriately. The thing that really fascinates me is that for all the rhetorical analysis that takes place in baseball season after season, I've never heard this subject broached. I've heard what a guy hits with runners on base, with two out and men in scoring position, against lefties, against particular lefties, and on balls thrown on the low outside corner. But I've never seen anybody do a breakdown of batting average at one strike, two balls, or 3-1.

The other thing that I might think about if I were a manager is loosening up on the take sign somewhat. I think some of us have the image that in the big leagues everybody's a free swinger, and the manager never puts the take sign on. One of the things that jumped out at me was the infrequency with which a 3-0 pitch is actually put into play. I'll give you the answer to this, but first I'm going to offer it as a question and let you take your best shot at it. Out of 5,000 plate appearances that were surveyed, how many involved some kind of "action" (defined as the at bat coming to an end) with a count of three balls and no strikes? Now, of whatever number you guess, how often did that "action" involve a ball actually being hit into play?

Here are the answers. Of the 5,000 plate appearances I surveyed, only 108 (that's 2.1 percent) ended on a 3-0 pitch. And of those 108, only 13 involved a ball being hit into play. (The rest were all walks or hit batsmen.) That doesn't literally mean only 13 hitters swung at 3-0 pitches, of course...some swung and missed. (No, I didn't keep tabs of how many...but I doubt it was very many.) But it seems safe to say that even major league hitters are very, very bashful about taking a swing at any 3-0 pitch. 

2. Starters. To really get into this subject would require an essay as long as the chapter in the book, so I'm just going to touch on the highlights and put in a plug for the book for more details. But the data strikes me as pretty plain on this subject. 

A. If we consider a starter who is pitching "well" (a term that is defined in the book), the point at which his performance deteriorates sufficiently to justify bringing in a closer is about pitch No. 115. The point at which his performance deteriorates sufficiently to justify bringing in a setup man is about pitch 120-125. I think we can agree that all those numbers are higher than most managers would usually think of letting a starter go in a close game. But the thing you have to keep in mind is that the question of removing a pitcher is never abstract; it is always comparative. As I said in response to an earlier question, the proper question can be framed as follows: Would you rather have Curt Schilling throwing his 115th pitch or Keith Foulke throwing his first? Often we assume that any fresh pitcher is better than any tiring pitcher, but the data strongly indicates there's no reason to make this assumption as a broad statement.

B. Due to lighter use patterns, it is becoming more difficult to accurately gauge the impact of heavy use on pitchers as the season wears along; that is, to tell whether they wear down due to accumulated work. After all, you can't study the impact of heavy use if nobody is heavily used. I generally define heavy use as 250 innings, and in the modern game that's one to three pitchers a season. But the modern data appears to be pretty consistent with historical data on this score. The performance of heavily used pitchers improves with work, it does not decline. Look at the April-July ERAs of all pitchers since 1998 who wound up with 250 innings of work; then compare those ERAs with the end-of-season ERAs. You'll find that as a general proposition they declined. Since 1998 there have been 13 pitchers who worked 250 innings. Their collective ERA on Aug. 1 was 3.02. Their collective August-September ERA was 2.86. The pitchers got better, not worse, with more work. In 2004, this was true of Hernandez; and it was true of Halladay in 2003. 

C. As to injuries caused by overwork, the question is unanswerable. These days, so many pitchers are being diagnosed with arm or elbow injuries that cautious use of pitchers is endemic, and yet injuries still occur. Here's one possible reason: Most of these guys have been pitching since Little League, and literally nobody has any idea how many pitches they've thrown. Most youth leagues have inning counts, but they don't have pitch counts. Instances of arthroscopic elbow and shoulder surgeries among mid-teens are way up. Kerry Wood's arm broke down in 1998 and everybody blamed the Cubs' use of him, but the reality was Wood's average number of batters faced was just average...about 27 per game. The more pertinent question is probably: What did he do at Grand Prairie High, and during summer ball, etc.? I think we tend to blame major league use because that's what we see and record, but the reality is unknown.

If I were a pitching coach I'd treat young (under age 27) arms gingerly because I don't think they're fully developed yet. But beyond age 27, I'd say it's rented mule time. But managers and pitching coaches absolutely will not adapt on this because whenever a pitcher comes up sore they're sure to get blamed by the press and fans...and lord help that manager if the injured pitcher threw 120 pitches the previous outing. He'd be lucky to ever work again.

3. Relievers. Relief pitching is the great crapshoot because most of these guys never pitch enough innings to establish a reliable pattern. That being the case, almost no reliever can be trusted with a big dollar contract from year to year. The evidence is all over the place, and I cite it in the book. For teams leading in the late innings of close games today, there is no increase in winning percentage over two, three, four or five decades ago. Let me put that another way. In 1948, teams that had no bullpen hierarchy whatever did every bit as well at holding leads in the late innings as teams with a very rigid bullpen hierarchy do today. How much are you paying Mariano Rivera?

The relief crapshoot is easily recognized when you think about what pollsters call "sample size." In a given season, a reliever might pitch 50 to 70 innings. That's not enough to establish the pitcher's true merit because it is not a sufficient body of work to reduce the luck factor to a manageable proportion as happens, for example, when a starter pitches 200 innings. It's much more reasonable to forecast that in 2005 Johan Santana will have a good year than it is to forecast that Brad Lidge will. The reason is that Santana accomplished his 2.07 ERA in 228 innings of work, making it much more stable as a predictor of future performance than was Lidge's 1.90 ERA in 95 innings. 

The poster boy for this is Jose Mesa. His is the classic reliever's career. He became known in 1995 for saving 46 games and compiling a 1.13 ERA. But he did it in only 64 innings. In fact, Mesa has never pitched more than 85 innings in any season since he became a reliever in the mid 1990s. And since that marvelous 1995, here's what's happened to Mesa's seasonal ERA: In 1996, it tripled to 3.73. Then it fell all the way to 2.40. Then it nearly doubled again to 4.57. Then it jumped to 4.98 and finally to 5.36 in 2000. A reasonable speculation would have been that Mesa was through. But all he had really established was the working of random chance on a small number of innings. In 2001, he reduced his ERA more than in half, to 2.34. In 2002, it was just 2.97. One might have deduced that a new Mesa had emerged. Again, what baseball was really seeing was the working of random chance in a short-inning reliever. In 2003, his ERA more than doubled for the second time in his career, to 6.52. In 2004, he cut it in half for the second time, this time to 3.25.

What would I make of Jose Mesa? I'd say he's a reliever. What would I expect of him in 2005? The same thing I'd expect of any reliever: something between great and awful.

Managers will not adjust to this reality because they and the game as a whole have become convinced that there is something metaphysical about relieving. I recall Kerry Ligtenberg coming out of nowhere to save 30 games for Atlanta in 1998, after which Bobby Cox pronounced him "a born closer...he's got that something indefinable." That born closer lasted part of one more season in the role for Cox and since 2000 has saved five games. What he did in 1998 wasn't metaphysical at all; it was very simple. He got 14 opportunities to protect (on average) a two run lead for one inning and he succeeded 12 of those 14 times. What's the big deal?

NetShrine:  Bobby Cox certainly is one of those "old school" guys.  Related, why do you believe "The Book" has persisted despite analytical evidence to the contrary?  In other fields--medicine, car design, agriculture -- new ideas, if backed up by research, take hold quickly.  Granted, men who have been in the game as long as Cox will have those deep roots/ties that are hard to remove/change.  But, there's also some younger blood in dugouts and front offices these days - and, yet, there is not a huge wave towards rewriting the book. Any thoughts on this?

Bill Felber:  I'm glad you mentioned Cox. I agree with your assessment of him. Schuerholz as well. The whole Atlanta philosophy is very solid and as much in line with what I find to be the logical approaches as any team. Generally, and with some exceptions, Atlanta has done two things that most teams do not do, both of which make tremendous sense.

1. They really emphasize starting pitching. They scout for it, pay for it, and lean on it. In 2000 and again in 2001, the Braves were what I classify as a "rotation" team. That means they spent more than one standard deviation outside the norm (in the percentage of their total payroll) on their five rotation starters. In 1995, 97, 97, 98 and again in 2004, they were a "rotation plus pitching" team. That means they spent one standard deviation outside the norm on their rotation and also on their pitching staff as a whole. Of the eight payroll classifications I outline in the book, those are the two most successful approaches in terms of making the playoffs. One could see a chicken and egg aspect to this: Did the Braves succeed because they followed the best road to team construction, or are those the two best roads to team construction because they're the two the Braves commonly chose? I tend to think the former, but who can say definitively?

2. Usually, Schuerholz and Cox manage to resist the urge to fall in love with high-dollar closers, who as a rule are a drain on resources that would be better applied toward a rotation starter or a middle of the order hitter. With really only two exceptions, they've followed this approach for the past decade. In 1994, their closer was McMichael, a second-year pitcher. In 1995, they gave the job to Wohlers, a low-priced former bullpen knockaround. When he had a big year they strayed from the mold, although not too bad at first, going to $1.425 million with him in 1996. Their first real deviation from the wise practice was to give Wohlers a multiple-year deal going in to 1997. Out of character for the Braves dealing with a closer. They got one good year out of Wohlers, then he went south for the last two years. To the Braves' credit, when he did they brought in Ligtenberg, low-priced, to take his job in 1998 and then Rocker, a second-year pitcher, in 1999. When they dumped Rocker in mid 2001 many thought it was due to his mouth, and while I don't discount that factor he was due to be dumped merely because his price tag was about to exceed his possible value. Then in 2002 circumstances caught up with them. Smoltz came back from an arm injury but could only do limited work...so they made him a saver. A very high-priced one, but I think it was something the Braves felt they were forced to do rather than something they wanted to do. Best evidence for that? Smoltz is back in the rotation this year. They got Kolb for $3.5 million, which is good (but not great) money for a closer and sort of breaks the pattern again. The Braves' record on this score is not pristine, but it's better than most teams.

Sorry about the diversion...back to your original question concerning why the book is so hard to change. I think fans, media, sports talk shows have a lot to do with it. GMs and managers risk a real pounding when they go against the grain of conventional wisdom. Yes, the younger more thoughtful GMs, Epstein and DePodesta and the like, could defend themselves using logic...but only if their audience is listening, which it often isn't. Two examples from recent history serve to illustrate.

1. Bill James came to Boston with Epstein in 2003 and decided the strategy of using a hierarchical bullpen with a dominant, highly paid closer made no sense. As a matter of cold logic, he was absolutely right on that. Then the Sox' pen went out and blew a couple of games early in the season, and everybody rushed to roast the team's pen strategy. It was just an easy shot to take. And I think the players buy into it as well, believing on fairly short order that there is something mystical about the closer's job despite the abundant evidence that guys move into and out of the job all the time. If Epstein and James had done the logical thing -- which would have involved pointing out how small a sample size of games people were reacting to, the degree to which luck operates on any bullpen, and how in most cases any competent major league pitcher could do what a $5 million closer does -- they'd have been roasted outside Fenway and might have faced a clubhouse rebellion as well. Instead they bowed to the pressure and did something they knew was only necessary from a public relations and maybe from a team psychology standpoint, they traded for Urbina. Then in the playoffs their boy Grady Little stayed with Pedro in the 8th, statistically a perfectly defensible move, and it happened not to work. In Vegas, the odds favor "7," but sometimes you roll boxcars. So, they got roasted again.

2. Listen to virtually any telecast of a major league game involving a pitcher who exceeds the 100-pitch mark. The tone of the announcers becomes "he must be getting tired" and from there fairly quickly escalates to "he's almost done (110 pitches) to "I hope he doesn't get hurt (115 pitches) to "I can't believe they're doing this" (120 pitches) to "will he ever be able to pitch again?" (125 pitches). In that context, almost any time a pitcher comes up with an elbow or shoulder injury the blame somehow gets back to the manager, and that in turn causes managers to get extremely cautious, really way overly so. Pitchers, too. One of the vignettes I note in the book was the night when Randy Johnson tied the single-game strikeout record in May of 2001. He struck out 20 Reds. But after nine innings the game was only tied. And after the ninth, Johnson told his manager he wanted out. It wasn't that he was all that tired, he said afterward, but he felt the Diamondbacks had a better chance of winning the game by bringing in the pen. Think about that. Randy Johnson in May of 2001 is a 37-year pitcher with a fully matured arm at the top of his game on the best night of his best season who says he could have continued...and he thinks a guy out of the pen gives his team a better shot to win the game than he does? The irony was that the Arizona pen gave up two runs, but the D-Backs got three and won anyway. But the point is that players, especially pitchers, buy into this notion as well, and once you buy into it mentally it becomes very difficult to change. The first thing a manager has to do is change the mindset of the players, especially the good ones. That probably has to happen organizationally. With respect to the best pitchers, there needs to be an understanding that the hierarchy begins with the rotation...that those guys are pitching the most innings because they are the best pitchers, and the manager's going to use them in that context. Another aspect of this is the five-man rotation. And I'm not talking here about using an ace pitcher every fifth day...I'm talking about the common practice in the big leagues today of using all five starters in a strict rotation. There are some teams who need to go to whoever their No. 3, 4 or 5 starters are and say, "look, Jason Schmidt's our top guy, and whenever he's had four days of rest, he's taking the mound and you're getting pushed back, and the reason is because he's better than you are. In 2004, Schmidt made 32 starts as the Giants ace; he worked in a very strict five-man rotation. San Francisco lost a post-season spot by one game. The Cubs missed out by three games. Maddux led the Cubs in starts with 33; Zambrano had 31. In a five-day rotation, the top starters would pick up about three starts per season and the bottom starters would get about three less. The book discusses the fallacies behind some of the arguments against this, but the bottom line is I'd rather have 36 starts out of Jason Schmidt and 30 out of whoever the Giants think their five guy is than 32 out of each of them. Put Schmidt on the mound for three more games in 2004 -- which merely means pitch him on four days of rest -- and I like the Giants' chances.

So in part I blame the media, which sets a "frame" for strategical discussions that ratify assumptions they have tested weakly if at all, in part I blame the fans for not demanding more rational analysis, and in part I blame the teams themselves for their willingness to follow a herd mentality.

NetShrine:  These are very interesting points - especially on the situation with the Red Sox pen. Personally, I still think the acquisition of Foulke paid for itself this past post-season - at least in the eyes of Red Sox Nation - but, that is entirely subjective on my part and I very well could be wrong.

Back to "The Book on The Book" - what were the major issues/challenges that you had to deal with in writing it and/or getting it published?

Bill Felber:  The "getting it published" part is pretty insiderish, so I'll try to get it out of the way quickly. The book began in 1999 as a project of mine in conjunction with Total Sports. Unfortunately, Total Sports ran into financial problems, which caused its assets (including the contract for this book) to be purchased by Rogers Media. That tied up rights to the book for about 18 months, and when I got them untied of course that meant finding a new publisher. I feel very fortunate that St. Martins was interested and willing to take it on. Landing at St. Martins is really landing on your feet. But the reality is we're now going on six years from the start of the work until publication, and that created one other completely practical problem. As the clock ticked, I kept having to weed out "old" data and conclusions that had been valid when written two or three years before, but which had been superseded by events. For example, the entire relationship between team payroll and performance shifted between 1998 and 2003...then sort of began to shift back in 2004.

One of the questions I set out to answer was: Can we quantify the extent to which team payroll dictates team performance? Pete and a few other math-minded friends suggested using regression theory to try to answer the question. I'm never going to pass myself off as a mathematician, and I'm not sure I could give you a cogent two-sentence explanation of regression theory right now. But we created some templates to apply to the problem and discovered that in recent seasons (this was done in 1999) the relationship was very strong indeed. I expressed it as a percentage, and between 1995 and 1999 the relationship ranged around 70 percent, and up to a maximum of 82 percent in 1998. (That use of a percentage may be a little misleading; I think the reader is better served to think of 50 percent as the break-even point between a casual and meaningful relationship.) 

At any rate, no sooner did I get comfortable with the conclusion that (based on recent data) payroll really was a key element in team success than the data shifted. In 2000, the relationship number fell from 72 percent (in 1999) all the way to 51 percent. It fell again in 2001 to 31 percent, a very casual relationship. In 2002 regression theory put it under 50 percent for the third straight year...so that meant it was time to rewrite my conclusions. The same held in 2003. 

Then of course as soon as I submitted the final manuscript toward the end of the 2004 season I calculated the relationship for 2004, and wouldn't you know it, the figure bounced back to 54 percent, a borderline high figure and the strongest relationship since 1999. So once again we made some late adjustments to the chapter, although in this case with only one year's worth of data I'm hesitant to say that the picture is changing again and the team success is strongly becoming a function of payroll, as was the case in the late 1990s. Too early to tell whether 2004 was a new trend or an anomaly in an ongoing trend.

The interesting thing is that it appears payroll waxes and wanes as a force. Although it was very strong in the late 1990s, it was very weak in the late 1980s...as low as a 10 or 15 percent relationship in 1988. Given that 1988 marked the apex of the collusion era, one could speculate in all sorts of directions on whether there is a cause-and-effect in there. In the late 1970s, however, the relationship was pretty strong. It was also very strong -- very strong -- in the 1950s, the one pre-1970s era for which we have full team payroll data. If you graph it all out year by line, what develops is sort of a wavy pattern.

There were two or three areas where I basically had to figure out how to answer a question for which nobody had developed a formula previously. That was one of them. Figuring out how to rate general managers was another. Another was deconstructing park effects. I wanted to know whether it could be used as a predictive tool: In other words, could a GM look at a player's performance in one home park in one season and reliably extrapolate from that the player's likely performance in another home park in a subsequent season. For various reasons, I don't think that's easily done. But the big problem turns out to be what I might call the denominator effect.

I was intrigued by this after looking at some fairly obscure park effect-related data regarding the St. Louis Browns and St. Louis Cardinals in the 1920s to 1940s. If you know those two teams, you know that between 1920 and 1953 they played in the same ballpark, Sportsmans Park. You know some other things as well: In this period, there were very few new parks constructed. With the exception of the coming of night baseball, there were few changes in the atmospherics of the game. In short, it seemed to me that two teams playing in the same park at the same time ought to have pretty much the same park factors. 

Yet they didn't -- not even close. Sportsmans Park was pretty consistently a lot more of a hitters park for the Browns than for the Cardinals. As one example, the park effect for the Cardinals in 1945 was 95; for the Browns it was 128. If you know park effects, you know it can't be because the American League had better hitters: since park effect is a ratio, that fact would account for itself. My math experts and I puzzled over that one for 18 months, and then the solution finally occurred to one of them. The problem wasn't Sportsmans Park (the numerator in the park effects ratio), and it wasn't that the hitters in one league were better than the hitters in another league. The problem was that the National League parks as a whole (the denominator in the park effects ratio) were pretty stridently more hitter-friendly than the American League parks. That meant, in turn, that because Sportsmans was (by National League standards) a pitcher's park its park effect would consistently be low. And yet since (by American League standards) it was a hitters park it's park effect would be consistently high. The same park, same conditions, two vastly different ratings. 

Seen in that light, the problem with park effects as a predictive tool becomes more obvious: the readings depend not only on what the player does and where he plays, but on what architects in other cities do. The past several years have been hectic ones in park building, making park effects even less reliable as a predictive tool. I know roto geeks love it, but GMs have to proceed cautiously.

NetShrine:  Well, you certainly have touched on something with park factors that I bet many have never considered - or, at least, that no one has successfully published findings on (before you). 

Before we get to the last question, I want to thank you for this interview and your time spent engaged on all the questions. And, of course, I sincerely hope the book is a major success - it should be one (because I would bet that it will be on the wish list for many baseball fans this Spring).

In closing, the last question is an open one for you. Is there anything you would like to go on record with regarding the book, or perhaps you have a message for anyone interested in the book, that you would love to state but have not been given the opportunity? If yes, please consider this an "open mike" of sorts. Given this chance, what would you like to say?

Bill Felber:  The thing that keeps coming back to me is very simply how much fun baseball is. In any analysis that is done, that should not be lost sight of. Many issues currently confront the game, from stadium leases to drug issues to contraction to questions related to competitiveness. I note that the current Basic Agreement -- the one negotiated without a work stoppage in 2002 -- expires after the 2006 season. I hope that sentiment can be mustered during negotiations concerning the 2007 Basic Agreement to confront these larger issues in a joint fashion -- and I specifically include the players, owners and fans -- for the betterment of the entire game. In any sport that is also a big business, there is the potential for losing sight of the societal role the game plays. It's important that not be permitted to happen. That's not really a point of strategy, but it's at the root of interest notwithstanding, and therefore must be part and parcel of the premise of any discussion.

That's it.  Once again, our thanks to Bill Felber for granting NetShrine.com this interview!

Return to Interviews Main Menu