This article was published in the Summer 2010 Baseball Research Journal.
The Bill James Gold Mine 2010
by Bill James
ACTA Sports (2010)
$23.95, paperback. 341 pages
On the surface, The Bill James Gold Mine 2010 looks like a pretty close approximation to the original Bill James Baseball Abstract. The similarities are easily apparent: It’s 300 pages of Bill James talking about baseball; it’s got a section for every team; and it’s got a lot of numbers in it.
But it’s different in one important way: It’s got a lot less hardcore sabermetrics. Not just hardcore as in math and numbers, but hardcore as in scientific. In the days of the original Abstract, Bill would have a bunch of studies in them, scientific studies designed to reveal fundamental truths about baseball. Some of those were in the form of statistics, like Runs Created, where Bill would show that his formula did indeed accurately project how many runs a team would score. Others were the kinds of research that you’d read about in the paper in the context of academic studies in the social sciences. Do strikeout pitchers have longer careers than control pitchers do? Back in the 1980s, Bill did a study where he took pitchers with lots of strikeouts, then found a control group of pitchers with very similar records except fewer strikeouts. It turned out that the strikeout pitchers did last longer—much longer— and now that finding has become part of conventional sabermetric wisdom.
There’s a lot less of that stuff in the new book. Bill’s most complex work over the past decade or so has been Win Shares and Loss Shares, a complex way of breaking players’ contributions down into wins and losses. But this book spares us details of the method and just gives us the results. So, instead of using sabermetrics to explore the structure of baseball, here James uses it mostly as a baseball historian.
In the service of helping us understand baseball history, Bill additionally comes up with a couple of more intuitive measures.
The very first essay in the book is called “Comparing Starting Pitchers Across History.” A traditional sabermetrician might pick up his toolbox and rank pitchers by how many runs they saved, or how many Runs Created they limited the opposition to, or how many Win Shares they had. But the 2010 Bill James says, Let’s come up with a less formal method of ranking outstanding pitcher seasons, the way an ordinary fan would, and proceed that way. So what he does is give points every year to the top six to ten starting pitchers, bonus points for being more outstanding than even the usual best, and just add them up. There’s no attempt at a formal mathematical justification for why this system is the most accurate possible—just an understanding that the system is pretty good, produces reasonable results, and matches our intuition for what makes a good career and what makes a better one.
Coming from anyone else, this might be dangerous stuff. Serious sabermetricians have historically been very critical of what Bill has called amateur attempts at numerical rankings, many of which suffer from the problem that they ignore what’s actually known about how baseball games are won. But Bill understands the difference between a formula that’s supposed to measure something empirical and a formula that’s supposed to come close to measuring something intuitive.
How do you compare Sandy Koufax to Jim Kaat? Koufax was a much better pitcher at his peak, but he had a shorter career. Which achievement counts for more, Koufax’s brilliance or Kaat’s longevity? To most of us, it’s Koufax. But now, try Ron Guidry versus Bert Blyleven. Now, Blyleven’s longevity seems to come out on top.
Bill’s brilliance is his ability to come up with a method that roughly matches our intuitions. In some ways, it’s actually harder than a formula that predicts runs, because, with the empirical formula, you can compare your estimate to actual runs and tweak it until you get it as accurate as you can. But when you’re trying to match an intuition, rather than a number, it’s harder to tweak. The genius of Bill’s method is that he’s kind of figured out what it is we admire in a player’s career and that he’s been able to create an accurate arithmetical framework in which to estimate it.
Is it possible that Bill’s method could be improved on? Sure. Someone could come along and say, If you change the point scoring from this to this, and give a little bonus for this, and add this one other thing, it makes the system a little bit better. But, again, since you can’t compare the results to anything you can measure, how would you know any change is really an improvement?
And, more important, what does it matter? Bill is using the method as a baseline for evaluating careers. It’s not like a runs formula, where accuracy is paramount. A single run, these days, is worth close to $500,000 in player salary. If in his capacity as an analyst for the Red Sox, Bill figures out that one free agent is actually three runs better than another, similar player . . . well, he might have saved the Sox a million dollars right there. But when you’re just trying to get an organized idea of whose careers were the most brilliant, it’s not important to get things right to the last decimal place.
So what do we get out of all this? First, as Bill writes, the system shows that contemporary pitchers compare quite favorably to the greats of the past. He says that this is what he learned from the method, which suggests that he didn’t have a grip on that before. Perhaps we’re too close to the active players and their current reputations to be able to clearly evaluate their legacies, and a structured method like this can help overcome any preconceptions that we have and that the next generation might not.
Second, Bill writes that his system has a fairly clear demarcation between Hall of Famers and non–Hall of Famers. It turns out that his method is so close to our intuitive feel for what constitutes a Hall of Fame career that, as a side effect, it’s extremely accurate in predicting what the voters actually will decide.
Anyway, I’ve written a lot of words on this method of Bill’s, but that’s not really central to what the book is about. Bill’s explanation of his system takes up about the first three pages of the chapter; the other twenty pages or so are a pitcher-by-pitcher discussion of the best starting pitchers in baseball history.
If you’re familiar with Bill’s other books, you know what that’s like. He’ll list a pitcher, then talk about whatever comes to mind about him and his career. Sometimes it’s just a sentence or two, sometimes it’s a couple of pages when he gets going on some way this pitcher relates to the issue of ranking careers. Reading the pitcher chapter here was a lot like reading the pitcher chapter in his Historical Abstract. It’s perhaps a tiny bit more technical, but not so mathematicky that it should turn off non-sabermetricians.
It’s history as much as it is sabermetrics.
There are a few such history chapters in the book, mostly essays that originally appeared on Bill’s website during the year. He’s got one breaking down the 1959 Go-Go White Sox (a chapter that would be of very little interest to me if Bill weren’t such a good writer— if he did the 1981 Blue Jays, now, that would be a different story). He’s got a chapter breaking down the Hall of Fame qualifications of four Chicagos—Minnie Minoso, Andre Dawson, Bruce Sutter, and Lee Smith. He investigates whether players in the free-agent era switch teams more often than they used to (they do). And he presents a method for estimating what percentage of a player’s potential he actually reached in his career (Luis Aparicio and Willie Mays, 90; Jackie Robinson and Ralph Kiner, 43).
And one of my favorite chapters is where Bill lists his observations as he watches a rebroadcast of a 1974 World Series game. Among them:
Runners on first and second, Cey grounds to shortstop for a 6-4-3 Double Play.
But neither runner is anywhere near being out. On the play at second Dick Green is 3 to 4 feet off of second base, and hasn’t been on second base anytime recently. I would have thought the in the neighborhood call at second was getting worse, rather than better, but . . . I haven’t seen anything like that in years. There is no question that if you did that now, the umpire would not give you the call.
And then Cey beats the throw to first, and they call him out as well.
After more than half a lifetime of reading Bill James, my instinctive reaction is to think about how to study the issue. What kind of statistical evidence would there be if the phantom tag of second were indeed getting less prevalent? Can we pinpoint the timeline when it happened, figure out whether it was a gradual or a sudden change?
Perhaps you could check the percentage of groundballs that turned into double plays . . . if that percentage has been going down, while other measures of fielding skill have been going up, that would qualify as evidence. Something to think about.
The book’s title, The Bill James Gold Mine, refers to the several hundred little statistical facts that form most of the text. They’re the kinds of unusual statistical anomalies that, back in the old days, Bill would quote in his player ratings.
Most of them pertain to specific, nontraditional stats that Bill (and his staff) kept track of for 2009. For instance:
The secret to [Gary Sheffield’s] 2009 success . . . was that his ground balls got through the infield. Sheffield hit .349 on grounders, almost 100 points higher than a normal batting average with the ball on the ground.
The Cardinals’ opening day starters accounted for only 57% of their regular-season starting lineup, a remarkably low figure for a playoff team. All other teams that made the postseason were at 62% or higher.
When Placido Polanco swung the bat in 2009 he put the ball in play 57% of the time, the highest rate among American League hitters. This is typical for Polanco. His 57% career rate leads all active batters who have seen a minimum of 5,000 pitches.
Brett Anderson in 2009 threw 237 changeups— none of them to a left-handed hitter.
On pitches in the strike zone, Josh Hamilton swung 82.5% of the time, the highest percentage in baseball. He took only 103 called strikes, which works out to about one for every 3.5 plate appearances.
In 2009, Hank Blalock hit 51 fly balls to left. Fifty of them were caught, giving him a .020 average on them. None left the park. He also hit 56 fly balls to right. Twenty-nine of them fell safely, including 19 that left the park, giving him a .518 average on those.
Russell Branyan swung at the first pitch he saw more frequently than any other player in the American League. He took a rip at 37. 3% of the pitches he saw [for a .415 average in those atbats]. Franklin Gutierrez took the other approach, swinging at just 7. 7% of the first pitches he saw [and hit .348 on those]. . . . The interesting thing is that Branyan is not a swing-at-everything guy. He actually takes a lot of pitches. He just likes to swing at the first one.
These little nuggets, I guess you could call them, are usually presented without comment (but with lots of white space between them, and the same little photo of stacks of coins, over and over again—gold mine, get it?). Sometimes I kind of wished that Bill would tell us what he thinks they mean. Take the Hank Blalock factoid, for instance. Should Hank stop trying to hit to the opposite field? Or would that screw him up to the point where he’d be worse off than when he started? Or maybe he makes up for all those fly-ball outs by hitting lots of line drives to left (batting averages on line drives are in the .700 range)?
It’s something to research, I guess, or at least think about. In any case, serious fans of baseball in 2010 will find these the most thought-provoking part of the book. The kinds of granular data that Bill and his staff calculated, the breakdowns of pitches, at-bats, innings, games—these were new enough to me that they got me thinking about issues I’d never considered before.
And if the pictures of gold coins were line drawings instead of photos, I could color them in while I’m thinking.
Bill has access to 2009 zone-rating stats, from John Dewan, I believe . . . and, on page 162, he’s got a chart rating every position in the American League.
In 2009, Seattle Mariners fielders were 110 runs above average. The Kansas City Royals were 62 runs below average.
That means the Mariners were 1.06 runs per game better than the Royals just in fielding. Since the Royals gave up only 150 runs more than the Mariners did, that suggests that Kansas City actually had better pitching than Seattle did, even though the Mariners led the league with a 3.87 ERA and the Royals were third-last at 4.83.
Overall, I’d say that the book is about two-thirds nuggets, one-third essays. I’ve described only the historical essays, so far. I should briefly mention some of the rest. There are a couple on contemporary events— the 2009 Cy Young races, and the future of Michael Bourn. There are a couple of opinion pieces—a proposal from Bill on how to improve the All-Star game, and a prediction of how history will treat steroid users. There’s also a fun rant on the attribution problem, which explains why we don’t let kids play by themselves in the park anymore and risk getting kidnapped. Some of my favorite Bill James pieces are the ones, like that one, that aren’t primarily about baseball. Bill has a true-crime book coming out next year, and I’m very much looking forward to getting my hands on it.
And I should mention that there’s one mildly technical essay, about defensive Win Shares, where he argues that the more outs you make at the plate, the greater your responsibility for fielding. I’m not sure I agree with that (or even understand it fully), and other sabermetrics sites I’ve seen were similarly perplexed with Bill’s logic. It’s still an entertaining read, though.
Those of us who subscribe to Bill James’s website will have read all the essays already, so only the nuggets will be material we haven’t seen before. For my taste, Bill is such a good writer and sabermetrician that I’m willing to pay $23.95 even if only half the material is new. Your mileage may vary, of course, but Bill is so important a sabermetrician, and so much fun to read, that I can’t imagine any serious Bill James fan would pass this up, even at twice the price.
PHIL BIRNBAUM is editor of By the Numbers, the newsletter of the SABR Statistical Analysis Committee. He lives in Ottawa, Canada.