Pete Palmer interviewby Chris Jaffe
January 14, 2008
I hope Pete Palmer needs no introduction. He's been dabbling in sabermetrics virtually his entire adult life, and he made a name for himself with The Hidden Game of Baseball, and the statistic it debuted: linear weights. Since then he's produced the best damn baseball encyclopedias ever published, Total Baseball and the current ESPN Encyclopedia.
So it's a real treat for me that he was willing to subject himself to the standard grueling THT interview.
Can you tell us a little about yourself (age, background, family, general stuff)?
I will be 70 in January. I was lucky to find a new love after being widowed. Beth is the grandniece of Jigger Statz, probably the best player in the history of the Pacific Coast League and certainly the most popular. We have three children, Emily, 6, Daniel, 4, and Stephen, 2. We have a home in New Hampshire but spend a good deal of time in Florida. Between helping with the kids and doing my books, I am pretty well committed. I also have a son Bob in Chicago, who is 39 and works helping low income people find housing. He never got interested in sports but is very dedicated to his profession.
What's the first memory you have of your childhood?
When I was three, we bought a new house in Wellesley and I remember going over there before we moved in. I have a couple of hazy memories of the previous house as well.
Growing up, what kind of games did you and your friends play? Where did you guys like to hang out?
I remember playing kick the can, also a fair amount of baseball (this would be age 8 to 10). We played on an old tennis court, two boys against three girls.
What was your favorite subject in school as a kid? Least favorite?
I always liked math, did not do well in English. I was not good at speaking before the class, although I was always good at reading.
Growing up, were you always a baseball fan? Who was your favorite player as a kid?
I was unaware of the Red Sox's 1946 World Series but began to follow the Red Sox and Braves on the radio in 1947. Bowman baseball cards came out in 1948, and I was an avid collector for about five years. I sold them all the Goodwin Goldfaden for $10 (about 1,000 cards complete sets for 1948-52), which seems like a lot of money then. Ted Williams was of course my favorite player, although I liked Bobby Doerr, Johnny Pesky and Dom DiMaggio as well. Halberstram's book on them was wonderful
When you go to a game, where do you like to sit? (Upper deck or lower deck? First base side or third base? Stuff like that).
I don't go to many games besides the one at the SABR convention. We are now an hour or more from Fenway in N.H. I went to a couple of games in Florida (Marlins) and the stands were pretty empty. I am more interested in the stats than attending the games.
Your first name is Edgar. You go by Peter. When did you start doing that and why?
Well, my father was Edgar, and they used to call me Sonny, but my mother thought I needed a real name, so she came up with Pete. I have no idea where it came from. I was told that my father knew someone named Ted, who he didn't like, so Ted, which was a possibility, was out.
What's the appeal of baseball statistics to you? Do you remember when you first got hooked on them?
Well it probably goes back to those baseball cards in 1948. I always liked math and numbers. Initially, I compiled lists of like everyone with 100 runs, RBI, 200 hits, etc. I bought a lot of old baseball guides back when you could get them for five or 10 dollars, so I had season stats even before the Macmillan came out in 1969. I expanded to fielding and pitching, and also made lists of the regular players each year from the old Turkin-Thompson encyclopedia. I found out Dick Cramer did the same thing.
How long have you been in SABR? How has membership helped your research and work?
I joined SABR very early. I would have been a founding member if I had realized Cooperstown was only a four-hour drive away. I started corresponding with Bob Davids in 1971, when he put out his Baseball Briefs for sale. Bob started SABR when The Sporting News decided not to accept articles from amateurs like Bob and myself for publication. I would say SABR has helped immensely. I met John Thorn there and many other like-minded researchers like John Schwartz, Walt Wilson, Joe Wayman, Frank Williams, Jim Weigand, Lyle Spatz, Dave Smith and many more. Thanks to my connection with Bill's project score sheet, which developed out of Bill's inability to get situational data from Elias for the NL, I met Gary Gillette and Gary and I have been busy for the past eight or 10 years in various projects like the Barnes and Noble baseball and football encyclopedias.
I should mention that I was working as a consultant at Sports Information Center, which did the AL stats from 1976 through 1987, and I readily supplied Bill with whatever he needed for the AL stats. I introduced OBP as an official stat for the AL in 1979, but the baseball guide did not pick it up until 1984, when Elias started doing it for the NL. That was when calling a sac fly an out for OBP purposes was put in, which I disagreed with. We don't have sac fly data for Cobb, Ruth or Gehrig because they were combined with sac bunts in the stats when the rule was in effect from 1908-30 and 1939, and separated out only in the 1954-date variety.
In the book The Hidden Game of Baseball that you co-authored with John Thorn, you introduced the linear weights statistic. Can you briefly say what the idea behind that stat is? How did you come up with the idea for it?
I was working on relating team stats to wins and losses and developed the potential run model both from the world series play by play data and from simulation. Then I just took each event and measured how much the scoring potential went up or down in each situation. My model and raw data also showed the probability of each base-out situation occurring, so the average value was the sum of the difference in each situation times the probability of that situation. I did this for all event and then used the numbers to calculate the number of runs a team should score. I think I got it within a standard deviation of 24 runs or so, which is not bad.
What, in your opinion, is the main advantage of linear weights? What is its biggest drawback?
Linear weights assume each event over the course of a season has an average value based regardless of game situation. That is all you can do from basic stats. But now, thanks to Dave Smith at Retrosheet and Gary Gillette who picked up the pieces at Project Scoresheet after Bill James and John Dewan left, we have play-by-play back to 1957. So you can look at this data and calculate actual changes in run potential from each play. If a hitter did better in key situations, then his play-by-play linear weight run number could be higher than his season average figure. You can further use win probability differences as well. That is what Harlan and Eldon Mills did back in 1969 and 1970, but it didn't catch on. This will give you a number of wins which should be closely related to one-tenth the number of runs from the season stats. I am currently analyzing this data to see if there are significant differences in the three measures, which could indicate that clutch hitting exists. I believe so-called clutch hitting is just normal random variations, but until I finish the study, I won't be sure.
You currently help produce the ESPN Encyclopedia with Gary Gillette. How did you come in contact with Gillette? How did the two of you end up collaborating on an encyclopedia together?
Gary and I met through Project Scoresheet. As I mentioned, when Bill James and John Dewan went to join Dick Cramer at stats, which I think was in the late '80s, Gary ended up taking over Project Scoresheet, which had started in 1984. Gary also joined Total Sports around 1998. Since then we have kept the entity going under several names. I think it is now 24-7 Baseball. We worked initially on collecting the play-by-play data. Gary found various customers and finally we got the Barnes and Noble deal. I do most of the stats and Gary does the managing and editing. We work pretty well together.
Do you have a subscription to Baseball-Reference's Play Index?
I do not. What it is? I have been quite busy with keeping up the play-by-play data, doing the baseball and football encyclopedias and some other projects. I have been working with Steve Gietschier at the Sporting News on their record book for some years now and also on Who's Who in Baseball for about 15 years. Diamondmind Baseball, founded by Tom Tippett, uses my data for their historical editions and lately I have had several collaborations with Jim Walsh of Maple Street Press on team annuals and encyclopedias. I have been a member of the New England Patriots stat crew, doing stats at home games, since 1975. I also keep an NBA data base which has not seen much professional use except for some minimal assistance on Total Basketball. I have not done much with web sites.
In both the Historic Abstract and Win Shares James has criticized linear weights saying (among other things), "linear weights cannot possibly evaluate offense for the simplest reason: Offense is not linear," and "the creation of runs is not a linear activity." (page 451, Historic Abstract). How do you respond to his reservations?
Well Bill has a great way with words. I owe him a great deal because of his development of the field of statistical analysis, which helped some of the rest of us get recognized. Actually he is right. Baseball is not linear. The linear weight values do vary slightly with the overall level of batting. A single is worth slightly more in a good hitting year and an out is more costly. However, within the range that is actually found in major league play, these variations are small. You could develop a model with a different set of weights for each season, but it wouldn't change the results much. What I do is keep constant weights for every event except an out, and then I set the out value to whatever it takes to make the whole equation come out zero. This figure varies from around -.260 to -.290.
Another stat has gained attention for itself (both here at THT and at other places) in the last few years: Win Probability Added, which is essentially a souped-up version of the probability tables you worked up those many years ago. What are your thoughts/opinions about WPA?
This is Player Win Averages, developed by Eldon and Harlan Mills in 1969. I think it is a good method. The main problem is that it over emphasizes a few events. You can get more points in one crucial at-bat than in the rest of the season combined. The question would be, does a player with a higher than expected PWA one year generally also do better the next year? Dick Cramer's original clutch hitting study based on Mills data from 1969 and 1970 said no, and I think that is still the answer. When you consider each at-bat equally, then you have a larger and therefore more reliable sample.
How closely do you follow current work in sabermetrics?
Not as well as I should. I really have been kept busy with my publications.
What baseball websites do you frequent?
Recently, you wrote the forward to The Book by Tom M. Tango, Mitchel Lichtman and Andy Dolphin. How did you come in contact with them and find out about their work?
I have never met Tom, but I have corresponded with him via e-mail. Mitchel has purchased some reports from Gary and me, and I don't know Andy. I thought their book was very good. What they did, as opposed to the Prospectus book that came out at the same time, was analyze in great detail a few situations, rather than cover a lot of different things. Another very important thing they did was show how much difference it actually makes to use the optimal strategy rather than the normal one. Unfortunately it turns out the answer is not much. I have been touting for years the idea that the third base coach should send the runner in from third almost any time there are two outs because it takes only a 30 percent chance of success to break even, since the next batter will probably go out anyway. All it takes is for the throw to be a few feet off or take a bad hop to make the runner safe. The actual success rate is about 98 percent. The coach and manager are afraid if the runner gets thrown out it will look bad, so they don't do it. It turns out the actual difference in using the superior strategy is only a few runs a year.
Now for some mostly random questions I like to call "stupid stuff." Do you have any non-sports related hobbies?
I like to read, science fiction and history. I completed Heinlein, Clarke and Asimov. I am working on Orson Scott Card and Piers Anthony. I am reading a book by Evans on Joe McCarthy (the senator, not the manager). It turns out everything you know about McCarthy is not only wrong, but it was what was done to him, not what he did to others. It's amazing how Roosevelt's commie-filled staff changed the course of World War II and Korea and aided the communist takeover in China.
What's your dream car? What's the dream car you can reasonably afford?
I just need reliable transportation and prefer Toyotas.
How many words-per-minute can you type?
I do a modified hunt and peck with a little touch in there. Washington, Adams, Jefferson, Madison, Monroe, Quincy Adams, Jackson, Van Buren, Harrison, Tyler, Polk, Taylor, Fillmore, Pierce, Buchanan, Lincoln, Johnson, Grant, Hayes, Garfield, Arthur, Cleveland, Harrison, Cleveland, McKinley, Roosevelt. That's about 30 words in a minute, admittedly longer ones.
As a child, what was the best Christmas present you ever received?
Probably an electric train, a Lionel. I used it for several years and then donated it to my nephew.
Of all the people currently vying for the Oval Office, whom would you most like to see elected? Least like to see win?
I really don't like any of them. The Democrats are useless and the Republicans not much better and most of the Republican candidates are pretty liberal. I think government is way too big and way too expensive and doesn't accomplish much. They never do anything until it is a crisis and usually it is too late. They have been letting immigration go for decades. They can't even admit there is a Social Security problem, much less Medicare. The SSA reserve fund is full of IOUs that will have to be paid by the taxpayer when they come due. And the unfunded pension liability at all levels of government is monstrous. Most of the economic gain in the market is due to the devaluation of the dollar. Even the Canadian dollar is worth more than ours. China is holding trillions of dollars in US currency and if they decide they need more interest to buy our bonds, the Feds will have to spend the whole budget just on interest on what they have already overspent.
And last, but certainly not least, what's your favorite ballpark food?
I'm too cheap to buy anything at the ballpark.
History instructor by day, statnerd by night, Chris Jaffe leads one of the most exciting double lives imaginable; with the exception of every other double life possible to imagine. Despite his lack of comic-book-hero-worthiness, Chris enjoys farting around with this stuff. His new book, Evaluating Baseball's Managers is available for order. Chris welcomes responses to his articles via e-mail. Oh, and now he's on twitter.