It was thirty years ago today

March 17, 2017

Thirty years ago today, give or take a month or two, Bill James’s 1987 Baseball Abstract came out, and I’d like to take a moment to appreciate what that book meant to me. I’d been well aware of Bill’s vast talents for quite a few years by that time (I’m a proud owner of one of Bill’s pre-Ballantine Abstracts, tattered and falling-apart as it and I both are at this point, and my ramblings about outfield conversions to 3B had even filled a few pages of the 1986 Abstract) but the 1987 book just about blew my mind apart, an explosion I want to pause to remember here.

I consider it far and away the best of the Abstracts, the best book Bill has ever written, and quite possibly my favorite baseball book of all time. In addition to covering the 1986 season (one of my all-time favorite years) thoroughly, this yellow-covered Abstract was the one that featured a revolutionary study of rookies throughout baseball history, which opened my eyes to the still-incredible observation that a 20-year-old rookie is far more likely than a 21-year old with the same exact stat line to become a star, as a 21-year-old rookie is likelier than a 22-year old, and so on. Up until this point, I had considered players’ ages to be a junk stat, far too many of which cluttered up my brain, taking up space that I needed badly for such functions as remembering the sequence of events needed to get out of bed in the morning and get dressed and fed.

That same study (it went on for forty-one pages) was also remarkable for the bravery required to publish its conclusion about race. In examining rookie seasons, Bill concluded that black rookies went on to have different career paths from their identical white counterparts, and that overwhelmingly their careers were not only different but better, often far better. In case after case, Bill provided evidence that pairs of rookies who differed only in the color of their skin showed vastly greater development for the darker-hued players. Henry Aaron’s stat line in his rookie year, for example, looked an awful lot like that of Gus Bell, but over the course of their careers Aaron outplayed Bell (as he did with everybody else who played the game, of course), and so did black rookie after black rookie. There were a few exceptions (44 out of 54 comparisons concluded that the black player went on to have a better career than the white one), and Bill tested his conclusions in every way imaginable to see if he had committed some sort of methodological error, but the conclusion held its ground.

This information was shocking, not only from a baseball point of view, but from a sociological perspective as well. Not setting out to comment at all about American society, Bill confronted a basic truth underlying our entire culture, one that flies in the face of those bigots clinging to the belief that black people are lazy, genetically underprivileged perpetual under-achievers. On the contrary, Bill’s study showed that a black person (more specifically, a black major leaguer) will be able to advance his own career far more capably than his closest white counterpart.

This conclusion came about, I will remind you, not because Bill set out to prove it, but because Bill studied the subject of rookies so systematically that the conclusion was inescapable. By including 666 rookies (and not once alluding to the diabolical connotation of that number), Bill was able to show that his every conclusion, including the racial one, was far from a flukish outcome of a few skewed stats.

Speaking of stats, the 1987 Abstract also displayed how statistics are invented, how they evolve, and how they are used.  In the 1980 Abstract, Bill had introduced the concept of the Value Approximation Method, which sums up players’ entire seasons (and ultimately careers) in two digits. The V.A.M. was ultimately abandoned, mainly I think because it didn’t actually measure anything. It was just a number between zero and seventeen that, as the name says, measures the approximate overall value of a baseball player’s season. Oddly, though it has been superseded by WAR, which does claim to measure something tangible (Wins), approximately, Bill doesn’t seem to care for WAR on the grounds of imprecision. Personally, I apply to WAR the same caveat that Bill applies to his own VAM:  it is the Value APPROXIMATION Method, not the Value Precision Method. "It isn’t necessary that we agree with it in every case," Bill wrote on p. 46, "that it never evaluates the players differently than we might evaluate them. It is necessary only that the system be reasonable, consistent and fair."   Its purpose is to approximate value on a gross but rational level—ranking one integer higher on the VAM is not a guarantee that a player is better, or had a better year, or anything really. But ranking five integers or ten integers higher gives that sort of guarantee, with some degree of reliability.

That’s how I think WAR should be used, not as a yardstick of superiority, but as a rough yet reliable measurement of quality.  Would I trade every single player with a WAR of 7.0 for a player with a WAR of 8.0? Not necessarily. But do I feel comfortable saying that a player with a WAR of 9.0 had a better year than a player whose WAR was 4.0? I am at complete and utter peace with that concept.  That is what the VAM does, and that’s what it was designed to do, to make broad general statements about groups of players’ seasons (or groups of a player’s seasons) that are more accurate the larger the groups of seasons are.

As far as a single season goes, sometimes people will understandably use these figures as tie-breakers, as in, "You think X had a better year than Y, I don’t, and we’ve keep going around and around on that, so I’m just going to go with ‘My guy had a higher WAR than yours.’" Next thing you know, you’re using it to settle every dispute under the night sky. That’s an error, but as I say an understandable error.

Eventually, Win Shares replaced (or Win Shares and Loss Shares is going to replace) the VAM, in a never-ending quest for precision and tangibility in approximating players’ values. Until Win Shares and Loss Shares are presented to us, WAR will have to do, performing the valuable function of allowing us to generalize about great groups of seasons in a meaningful way. Some analysts (including Bill) find WAR presumptuous or inaccurate or unreliable, but I maintain that WAR (and WS & LS, and obviously VAM) was not intended to be humble, pinpoint-accurate nor completely reliable in the first place, just a method of reducing a page full of numbers down to two digits, which it did well. The 1987 Abstract used VAM widely, and in large groups, where it made the most sense. It was an instructive shortcut, letting us think about pages full of numbers with maximum efficiency.

This efficiency proved mind-blowingly useful in the study of rookies, for example. Bill could say (and did say) that Henry Aaron had essentially the same value as a rookie that Gus Bell did, according to VAM, but a much larger VAM over the course of his career. He didn’t need to cite how often they got on base, or scored runs, or accumulated total bases, or a zillion other incremental integers that showed their abilities, just a couple of digits worth of VAM, permitting him to make his points in only 41 pages instead of the several hundred pages he would otherwise have needed to make the same points less convincingly.

VAM was a great statistic, now superseded by other broad measuring stats, which it inevitably led to. Bill’s other attempt at a synoptic stat was his Runs Created stat, which also, I think, led directly to WAR, once folks figured out how many runs it took to make a win. The appendix at the back of the 1987 Abstract also usefully defined these and the other tools Bill invented or refined up to that point: Brock2, Trade Value (which was a refinement of Approximate Value), and Speed Scores, which were fun. It let you quantify how fast players were, again settling a lot of dumb arguments and thoughts about who was faster than who, and how do you know when a player is losing speed or just having a bad year on the basepaths.

Sim Scores got introduced around here too (I’m not vouching for the actual first-use date, just noting that by 1987, Bill was referring to them), a stat that’s so common now I think people have forgotten where it came from or when. But as epochal as these advancements may be in baseball history, I was mostly charmed by Bill’s writing, which ranged from the eloquent to the profound. Thumbing through it, which I do every other year or two, so much seems fresh out of the oven: this was the Abstract where some of my favorite bits appear, the hilarious "Will the McMeeting Come to Order?" parody of a San Diego Padres board meeting, the close analyses of the 1978 and 1986 AL MVP races, with Guidry, Clemens, Mattingly and Rice having their candidacies considered and compared and rejected and promoted, objectively and coolly,  "The Greatest Outfields Ever" essay, "The Ken Phelps All-Star Team," the cruel but dead-on "Chuck Tanner’s Funeral Home"—this was back when Bill wrote mean, accurate critiques of players and managers, sometimes so mean and so accurate that I still gasp a little bit, even though I’ve read what’s coming two dozen times before.

But as much fun as Mean Bill was to read, it was Accurate Bill, Obsessive-About-Accuracy Bill, who made me think about this stuff like I’d never thought about it before.  He wrote a kind of preface, "Meaningful and Meaningless Statistics," that ranked stats on a very thorough grid, in three main categories (Category Importance, Reliability, and Intelligibility)  that considered  each stat’s virtues and vices, and ranked each one on a scale of 1-10, and then wrote a conclusion "Putting the Elements Together" that distinguished the wheat from the chaff.  ("ERA" was the most meaningful stat, followed by OBP, and at the bottom of the list was "Sacrifice Hits.") I was sorry to find Game-Winning RBIs ranked as low as it was, partly because the All-Time GW-RBI leader was one of my heroes, Keith Hernandez, but that was another important lesson Bill taught me: it’s not about players you like or dislike, Bub, it’s about what’s true and what’s accurate. If you want to rave about your faves, well, that’s a few doors down from here.


COMMENTS (22 Comments, most recent shown first)

My favorite authors, in no particular order, are Bill James, Stephen King and Stephen Jay Gould. Opening any of their books is, to me, like sitting down and talking to an old friend.
9:42 PM Mar 31st
My first introduction to Bill James was the 1979 Baseball Abstract. I went to my pal Phil's home and we sat down to play our favorite table top baseball game Strat-0-Matic, but before we started our game, Phil showed me the 1979 Baseball Abstract. I was blown away. Account form box scores, a comparison of the 1978 seasons of Jim Rice and Ron Guidry, a comparison of the 1970's Reds with the 1950's Brooklyn Dodgers: Campy and Bench. The Value Approximation method. I wrote the address down and ordered my own copy. Bill would send out a letter that the next Abstract was ready. The 1980 and 1981 editions are wonderful. The 1982 edition took a bit of a step back for the new readers to catch up. I also ordered the 1977 and 1978 editions from Bill. The 1977 edition is mostly all stats with players monthly batting records and an article about seeing lots of games for a team and not noticing if one player hit .275 and another hits .300. The only way to notice quality is in the stats. I have been reading Bill's work ever since. Take Care, Tom Nahigian
10:00 PM Mar 26th
Brock Hanke
I remember Don Malcolm telling me that he'd been getting a self-published thing from somebody named Bill James, and James had gotten a book deal, and that I should make sure to get a copy of this new book. I'd been doing what amounts to prehistoric sabermetrics ever since I'd bought a copy of MacMillan in 1969, but this Bill James Baseball Abstract was DIFFERENT.

And the main thing that made it different was Runs Created, which I still consider to be the basic foundation of sabermetrics - the idea that you can take a hitter's component stats and accurately estimate the number of his team's runs that he has created. Before then, all anyone could do was argue. And, as part of Runs Created, but in its way even more important over time, was the method Bill used to validate it - taking the component stats of teams and figuring out how many Runs Created that would amount to, and checking that against the KNOWN and RECORDED numbers for the teams. That allowed Bill to justify his method - to fact check it. That was why you could only argue before. There is still no real way to record which player is responsible for which parts of which runs. But you can do that for teams.

When Pete Palmer introduced Markov Chain analysis in The Hidden Game, that was the second biggest breakthrough that I know of in early sabermetrics.

I complain about WAR systems a lot, but I don't really complain about WAR. I complain about the systems that are (not) built around it. The New Historical system, with career, peak, prime, and rate, is something that no one involved with WAR has ever attempted; the closest is probably JAWS, which does not consider enough time frames, and the one it uses is 7 years, which is too many. I have no idea why no one developing a WAR system does not just do this with their WAR. Maybe they're afraid to confront the timeline issue. I don't know. But I do know that the New Historical is closer to right than anything else I've ever seen. That it uses Win Shares instead of some version of WAR is almost incidental.

As for Bill himself, my favorite thing about him is that he's not afraid to turn out to have been wrong, or less right than someone else. He'd publish someone else's WAR or Runs Created stat in his own Abstracts if that new method tested out a little bit better. I can still remember Bill paying good money out of his own pocket to purchase stats that someone else had hand compiled showing that Rogers Hornsby did not get a big home run benefit from Sportsman's Park (which was an asymmetrical bandbox that favored lefty hitters). Bill had publicly contended that Hornsby's homer stats were inflated by the park. Anyone who will pay his own money to prove that he himself was wrong, about a player he does not like, has all the intellectual integrity that I will ever need in an author.
1:41 PM Mar 23rd
My introduction to Bill was the article in Sports Illustrated about him (not sure what year). It blew my mind. I sent Bill a letter asking how to get a copy of the Abstract, and if he had back issues for sale; I remember in his response he wondered why anyone would want back issues, but I did!
I especially love the two Historical Abstracts. All my life (I was born the year Jackie was a rookie with the Dodgers) baseball fans have opined on the best players in baseball history, and Bill dealt with that in a way no one else ever had. I have almost memorized his mini-essays on Oscar Charleston, Arky Vaughn and Jackie, among others. Bill is always challenging, provocative, and fun to read. He changed my life!
12:17 PM Mar 22nd
I have all the published abstracts too. I think 1987 is my favorite. The 1982, 1983, and 1984 rank high. I think that the 1985 and 1986 were a bit too technical for me.

Is it possible to buy any of the old self-published abstracts. I'd love to see the one for 1981.
6:37 PM Mar 21st
I have all of the Abstracts from 1982-88 on my shelf, and I still take them out and read them occasionally. I know the older ones can be worth decent money, but mine are so well worn from decades of leafing through them at the dinner table that I doubt they'd be worth anything. I wouldn't want to to part with them either.
4:23 PM Mar 21st
I was late to the party, and fortunate to join it at all. In a rare trip to the public library, I came across the 2001 big, Historical Abstract book and life has never been the same.
9:01 PM Mar 18th
Many stones built this temple of knowledge. I discovered this stone, the Abstracts, in 1985, my first year of college. I went to Albany Business College in Albany, NY, at a bookstore in the Egg. It was the 1985 Abstract and it was incredible and changed the way I watched and thought about baseball. Strat-O-Matic Baseball in 1981 was the first to change the way I viewed baseball. Thank you, Bill.
7:21 AM Mar 18th
Not enough credit is ever given to Pete Palmer. His book "The Hidden Game of Baseball," co-authored with John Thorn (1984, I think) was the pioneering work of advanced saberemetics.
7:58 PM Mar 17th
Couldn't agree more. The 1987 Abstract displayed more original thinking about more different aspects of the game than anything I had read. The only downside was going from being a member of a tiny elite (readers of the privately published Abstracts) to one of a multitude. Well worth it, too.

My explanation for the stronger careers of black rookies between 1947 and 1987 was that a young black player had to be better than his white counterpart to get an equal chance. I don't remember if that was also Bill's explanation.
7:35 PM Mar 17th
Thanks for the essay. I haven't looked at that Abstract in several years, yet remembered just what you described in it.

A friend turned me on to the Abstracts a bit earlier; I bought my first one in '84. Like everybody here, they profoundly affected the way I looked at baseball numbers and thus baseball players and executives, and how I watched a game.

The point about WAR being an approximation and not an exact measurement is indeed a welcome reminder.
1:35 PM Mar 17th
The 1987 Abstract was the one I found on a shelf in the bookstore of the Valley Mall in Harrisonburg, Virginia during a trip to Grandma's house at the age of 16. The moment I bought that book (over that year's Elias annual, just by dumb luck or maybe divine influence) was one of the half-dozen biggest inflection points in my life. It's probably the most influential book I've ever read. The rookie study warped my mind, and I still reference it frequently.

As I look across my home office I see the 1987 and '88 Abstracts sitting there. Which one had the Revolution article about freeing the minors? I'll have to look, but that article also had more influence over my thinking than maybe it should have. Ever since then I've dreamed of minor league emancipation.
12:20 PM Mar 17th
The '87 Abstract was the first one that I ever bought, too, Steven. I found it at a "Chapter 11" bookstore in Atlanta, and they had both the '86 and '87 versions there. I bought the '87 and started reading a little over lunch at a nearby burger joint, then went right back and bought a copy of the '86 Abstract as well.

I bought all Bill's following publications...the annual Baseball Books and the Player Ratings Books. I also found some new copies of the earlier Abstracts (the Ballantine editions) for two bucks apiece at a booksellers "warehouse" kind of place. (Wish I had known that the 1982 Abstract would eventually bring in the hundreds of dollars...I could've bought a dozen new ones.)

I joined BJOL too late to get copies of Bill's earlier "self-published" ones, and I envy those of you who have complete sets.
12:15 PM Mar 17th
Thanks for bringing up VAM, Steven. I truly wish with all my heart that the baseball world would use WAR the same way that VAM was designed and intended to be used.

Before there was money there was the barter system. If you needed a chicken and the other fellow only had a chain saw, how can you judge if it's a good deal? Money is like VAM or WAR, converting the chicken and the saw to numbers along a common scale.

So, what's it gonna be? What's more valuable? The chicken or the chain saw? It depends on how hungry you are right now.
11:52 AM Mar 17th
I too have the 87 Abstract (the only one that survived), and it is also my personal favorite. I always thought that the Star Value used in the study of rookies was a pretty simple yet elegant way of measuring the talent of ballplayers, despite its arbitrary scoring. Years ago, I remember calculating the Star Value of every ballplayer who played 10 years in the majors. I found that there was a high correlation between Star Value and Hall of Fame election. The cut-off score seemed to be 45 points. Anything under that, and your chances of making the Hall of Fame was low. I wanted to use Star Value to find out what was the greatest real All-Star team of all time (probably the 1937 AL and NL teams), but never got around to it.
11:36 AM Mar 17th
I remember the article(s) that you are talking about. It was not a surprise to me that a player good enough to play in MLB at a younger age would have a better career, on average, than a player who was not good enough to be in MLB at that age. That is what one would expect.

The difference incareer paths between white and black was a surprise to me. After seeing the results, I thought that Bill's explanation was right on target.

I think that win shares and win losses are great statistics, especially if one wants to present material in such a way that it's understandable for people with a limited background in statistics. I think that using a variety of less comprehensive statistics often yields more accurate results, however. I work with data a good part of each day, and my statistical models with a variety of simiple measures perform much better than models that use a small number of comprehensive measures. That could probably be proven mathematically, but I am not into proofs. The reason is that the comprehensive measures lose a lot of the differences that are captured by a variety of simpler measures. A friend of mine has a saying that "my head is in the oven and my feet are in the freezer, but I'm alright on average", and his saying captures what I am tryiing to say here.
11:26 AM Mar 17th
Steven Goldleaf
Both articles are unsigned, so I assume that they're Bill's work. The article immediately following the McMeeting one, entitled "Some Random Thoughts on Baseball and Ballplayers," IS signed by Mike Kopf, and that's the only one in the 1987 book that is.
10:56 AM Mar 17th
Are you sure the McMeeting and Tanner's Funeral Home articles were written by BJ? They read like Mike Kopf pieces.
10:47 AM Mar 17th
Steven Goldleaf
astros34--couldn't agree more. VAM was mindblowing, not because it was the best possible system, but because it opened the door to more accurate systems.
7:55 AM Mar 17th
I will check if they are available on kindle
7:47 AM Mar 17th
I should mention that Bill himself basically disavowed VAM in his Historical Abstract (pg. 338), citing basically the same reasons I did.
7:32 AM Mar 17th
VAM ( was better than anything that existed at the time, even revolutionary, but has been pretty much replaced. Several flaws existed in its calculation, to wit:

1) it used arbitrary cutoffs. A point for hitting .250, another for hitting .275, another fir hitting .300, .325, .30 etc. If you hit .273 or .299 you got nothing above the lower rank. This occurs at almost every step in the calculation. (see #5 below)

2) It used batting average in the first place. It did factor in walks as one step but again, arbitrary cutoffs.

3) SB cutoffs were 20, 50, and 80, with no consideration of caught stealing.

4) it used fielding average.

5) slugging average cutoffs were every .100 points. No differentiation between a SLG of .475 and .401.

6) It used pitchers' decisions not once, but twice, and also awarded an additional point for 18 wins.

I know, it's just an approximation. That's why Bill looked for something better and came up with Win Shares. WAR is also better.
7:30 AM Mar 17th
©2021 Be Jolly, Inc. All Rights Reserved.|Powered by Sports Info Solutions|Terms & Conditions|Privacy Policy