Baseball and the Value of Sabermetrics: An Author’s Perspective
Last year, Author Alan Hirsch was kind enough to answer and respond to questions and criticisms of his book, The Beauty of Short Hops: How Chance and Circumstance Confound the Moneyball Approach to Baseball. Alan co-wrote The Beauty of Short Hops with his brother, Sheldon Hirsch.
Q: Billy Beane didn’t/doesn’t watch A’s games because, in your words, “He can’t bear seeing the damn players muck up what should be a perfectly predictable contest.”
Don’t all GMs wish their moves would work out as planned and wish the game was predictable in some sense? And is the wish and goal of GMs and sabermetrics in general actually to make things perfectly predictable or to just gain as much insight as possible into who players are and what they are capable of?
AH: Yes, GMs are in the business of winning, and when they hire sabermetricians they try to improve their teams’ chances via statistical study. There’s obviously no problem with that. We were on the Bill James bandwagon early, and we hope the teams we root for find edges wherever they can.
It’s the excesses we argue against, and the failure to recognize limitations.
Here’s Moneyball’s description of Billy Beane’s perspective: ‘The game can be reduced to a social science…It is simply a matter of figuring out the odds, and exploiting the laws of probability’ because ‘baseball players follow strikingly predictable patterns.’ As for other GMs, I can’t speak for them but I know that many of them watch the games!
Q: I think one of the strong points of this book is your critique of Moneyball. But don’t the problems of Moneyball have more to do with oversimplifying the A’s and their use of sabermetrics into a narrative of stats versus scouts or sabermetrics versus tradition (for lack of a better word)?
AH: It’s true that the flaws of Moneyball don’t necessarily carry over to sabermetrics. At the end of the chapter on Moneyball we specifically note that the errors we identified by Michael Lewis and Beane could be their own and thus it would be unfair to judge sabermetrics accordingly. Then we turn to a chapter which discusses sabermetrics more broadly and directly.
Q: Also, it seems Michael Lewis suffers from a lack of perspective on how sabermetrics influenced the game. It was published in 2003, when sabermetrics was shedding the label of being a dirty word among baseball insiders. Shouldn’t Moneyball be viewed differently than sabermetrics?
AH: It’s true that a lot has happened in sabermetrics since 2003.
That’s why we have a chapter called “The Third Wave,” devoted to post-Moneyball developments. But your question also raises the relationship between Moneyball and sabermetrics more broadly. Moneyball revolves around Beane’s success with a small budget, principally due to insights allegedly gleaned from sabermetrics.
If Michael Lewis had just written a book that looked at how some low-budget team succeeded, without introducing a new paradigm for success, it wouldn’t have had nearly the impact it did.
But if sabermetrics is central to Moneyball, how about the converse—is Moneyball important to sabermetrics? As an historical matter, yes (Moneyball publicized and accelerated the sabermetric revolution), but as an analytic matter, you’re right—you can’t judge sabermetrics by Moneyball. We don’t.
Q: You focus a lot of attention on Jeremy Brown, the slow catcher drafted in the 2002 “Moneyball” draft, and how he was a prominent figure in Moneyball. But you conveniently fail to note that the A’s took Nick Swisher, Joe Blanton and Mark Teahen in that draft. Again, I think this points out the failures of Moneyball focusing too much on narrative more so than it points out the failures of sabermetrics.
AH: We focus on Jeremy Brown rather than Swisher and company because Lewis does, and he does because Brown illustrates his central point: Beane won with little money in large part because sabermetrics enabled him to identify undervalued players.
The issue is not whether Beane won on a limited budget, which is indisputable, but how he did so. In that regard, Jeremy Brown took on symbolic significance. Beane craved him because of a new paradigm of how to recognize undervalued talent (which was not the case with Swisher, who was widely recognized as a top prospect).
Thus Brown is not just one player whom one general manager misevaluated. In fact, Beane didn’t evaluate him at all—he thought Brown’s college statistics were all he needed to know.
Brown and Brant Colamarino (another player Beane craved based on statistics despite his non-athleticism) are pretty good examples of one way in which statistics can be over-valued—in this case at the expense of old-fashioned scouting.
Q: You say, at the highest level Bill James’s doctrine comes down to idea that baseball decision-makers can’t know what they’re doing without numbers. How can one objectively break down everything that has happened in major league baseball, in a meaningful way, without measuring everything that happened (i.e., without numbers or statistics)?
AH: We were praising James.
That’s the part of the book where we talk about how he rescued baseball from a tradition of ignorance. We’re all for objective data. It’s true that elsewhere in the book we give many examples of data that are useless and things that simply cannot be quantified.
For the record, I am pretty sure that James would agree with a good deal of what’s in the book. He’s publicly expressed misgivings about sabermetrics that track closely some of our criticisms.
Q: You are critical of Voros McCraken’s ideas of Defense-Independent Pitching Statistics (DIPS), the idea that besides strikeouts, walks and home runs, pitchers basically have little or no control over what else happens.
You point out that Sandy Koufax, for example, had a much lower BABiP (Batting Average on Balls in Play) than most other pitchers, therefore McCracken’s theory doesn’t hold water. His theory was built around evaluating pitchers without looking at hits allowed or statistics that are heavily influenced by hits allowed (ERA, WHIP).
Yes, certain pitchers are better at preventing hits but that will almost always show up in home runs allowed and strikeouts.
AH: It’s not just Koufax, of course.
We offer substantial evidence to refute the suggestion that pitchers have no control over outcomes except home runs, walks, and strikeouts. It’s no surprise that Mariano Rivera has a low BABiP—all those broken bats tend to produce weakly hit balls. I disagree with your suggestion that BABiP can be dissociated from the other metrics of DIPS; they are all of a piece.
We discuss all this in the Moneyball chapter. Michael Lewis argues that, thanks to his attention to McCracken’s idea, Beane was able to identify undervalued pitchers — guys whose ERAs were high solely due to a randomly high BABIP, while their more reliable numbers suggested their true quality.
In fact, when followed long enough, BABIP is not random—one of the ways pitchers can succeed is by inducing weakly hit balls. As for your suggestion that this skill almost perfectly tracks pitchers’ ability with respect to strikeouts and avoiding home runs, look at Dave Stieb and Catfish Hunter—not big strikeout pitchers and gave up plenty of home runs, but succeeded in large part because of low BABiP.
Q: You bring up the fact that Roger Maris had no intentional walks in 1961 hitting in front of Mickey Mantle, and conclude that one can’t quantify value with precision because of variables like Mantle helping Maris to get better pitches or increasing his opportunities to hit with runners on base and not walk.
But maybe we can’t quantify the value of these players in terms of overall influence on the team but can’t we quantify the value of these players in terms of their results? Doesn’t a distinction need to be pointed out there?
Statistically Ben Zobrist was one of the most valuable players in the game in 2008, but that doesn’t mean his value was representative of his skills rather than factors outside his control.
AH: You can certainly limit yourself to Mantle’s and Maris’ statistics, but precisely the point we were making is just how many variables go into a player’s value that one can’t even begin to quantify. If you were ranking these two in 1961, how do you factor in what Mantle did for Maris by batting behind him?
Bill James has said that there’s no evidence suggesting that a player can help the batter in front of him. Mantle and Maris are an apparent counterexample, as we show. But we also show that the extent to which Mantle helped Maris can’t be quantified. I don’t just mean it can’t be quantified with precision. I mean that any effort to begin to estimate it runs into several problems that apply to many sabermetric projects and that have not been acknowledged.
Q: What about the fact that Maris, by some measures, was actually as good or better in 1960 than in 1961?
In 1960 Mantle mostly hit in front of Maris, not behind him. And Maris only had four intentional walks in 1960 hitting mostly in front of the rather mediocre Bill Skowron. Should we question the impact of Skowron on Maris’ performance in 1960, the season in which he was probably more deserving of the MVP award?
AH: First, I’d take issue with the suggestion that Maris was as good in 1960 as in 1961. His slugging percentage and OPS were significantly better in ‘61, and he hit 22 more homes runs.
In terms of the intentional walks, keep in mind that ‘60 was his breakout season—he was quite ordinary until then. In ‘61, he was the reigning MVP and quickly established himself as a truly feared slugger.
So if your implication with the Skowron stat is that the zero intentional walks in ’61 wasn’t because of Mantle, I’d respectfully disagree. It’s staggering that, in the midst of a record-breaking home run season, Maris received zero intentional walks.
But the 1960/61 inquiry is a diversion from out main point. We provide significant data suggesting that Mantle’s presence in ‘61 helped Maris, but we fully acknowledge, indeed emphasize, that the extent of the benefit cannot be quantified. Moreover, we explain why additional data (from 1960 or 1962 or any other year) won’t help much, if at all.
This is one of several examples we cite in which potentially important aspects of a player’s contribution simply can’t be measured.
Q: Regarding Ricky Henderson’s baserunning, you point out that many sabermetricians discount what he did to disrupt opposing pitchers and help his teammates at the plate. You point out that several hitters—Dwayne Murphy, Don Mattingly and Dave Winfield—had their best seasons with Henderson batting in front of them.
This simply isn’t true.
Mattingly was as good in 1984 without Henderson as he was in 1985 with him. And Mattingly’s best season was 1986, Henderson’s worst or second-worst.
Winfield’s best season was clearly 1979, without Henderson. Murphy hit behind Henderson from age 24 to age 29. Is it really saying anything that his best season was one in which he hit behind Henderson? What about the other five, rather mediocre seasons behind Henderson?
AH: I think if you look at the data comprehensively (and don’t forget Edgardo Alfonzo, who may be the clearest example), you will find that overall players batting behind Henderson seemed to prosper.
But let’s put this in context.
For a long time, sabermetricians argued that stolen bases were attempted too often because the negative effect of a caught stealing was insufficiently considered. They were probably right, but their analyses neglected the fact that the threat of a steal might unnerve a pitcher and produce better results for the next few batters.
Then a prominent sabermetrician wrote an article which did consider this dynamic but nevertheless concluded that Henderson (because he was caught so often) was barely more valuable on the bases than guys who never steal.
The problem is that, in considering a player like Henderson’s effect on subsequent batters, he ignored several variables. This wasn’t just a failure that can be corrected by the next study. Rather, there are simply too many variables to consider, and no way to do a prospective study even if you somehow cataloged them all. More data is not always the answer.
Sometimes you’re just spinning your wheels and not getting any closer to the truth. I doubt that we’re closer today than we were 20 years ago to quantifying Mantle’s value to Maris or Henderson’s impact on a game.
You gave a good example why (and we actually made this very point). Dwayne Murphy had his best years playing with Henderson, but there’s no way of knowing how much of that was for reasons unrelated to Henderson.
Q: You discuss Babe Ruth’s stolen base attempt in Game 7 of the 1926 World Series. You write that Ruth’s odds of a successful attempt in that situation were probably anywhere from 20 percent to 80 percent. And if he’d have stolen successfully, then only the batter at the plate (Bob Meusel) needed to get a hit to tie the game.
But with Ruth on first, the Yankees needed hits from both Meusel and on-deck hitter Lou Gehrig to tie the game. So Ruth attempting a steal may have given the Yankees a better shot at winning.
You seem to be guilty of that which you criticize sabermetricians, you fail to take the specific situation. Meusel led the American League in homers in 1925 and posted a respectable slugging percentage in 1926. Plus, if by chance Meusel got on base (he had posted a .361 on-base percentage to that point in his career), then Gehrig comes up with the tying run at least on second.
AH: Meusel’s power is actually one of the many variables we consider.
There are additional variables we don’t consider. Part of our point is that you couldn’t possibly know them all. We argue that numerical analysis simply cannot help Ruth decide whether to steal, which is just one example of the larger point: Sabermetrics generally does not provide much help with respect to in-game decision-making such as whether to steal or bunt. The conclusion to the contrary rests on over-extrapolation from base rate data.
Look at this way. If Ruth is a 55 percent successful stealer and sabermetricians find that you need to be successful roughly 75 percent of the time to make a stolen base attempt worthwhile, isn’t it obvious that Ruth should not have attempted the steal?
Both the 55 percent and 75 percent figures are highly variable depending on the specifics of the situation—score, inning, pitcher, catcher, and any number of other things…Ruth’s likelihood of stealing the base in that very specific situation was a virtual guess.
And while sabermetricians can tell us that, on balance, you need to succeed 75 percent of the time to justify the steal, you don’t face “on balance” situations. The percentage needed to justify a steal when Grover Cleveland Alexander is throwing the way he is in a one-run game in the ninth inning—good luck figuring that out. Even if you could, by the time you did the inning would be long over.
Q: The way I understand it, 75 percent is kind of an estimated break-even point over the course of a season. Obviously you can’t know if a guy is going to be successful that often except through trial and error. If a guy has the speed and baserunning skills, he should utilize it until it’s proven he shouldn’t.
But I would argue, and I think most sabermetricians would argue, that a guy shouldn’t steal in any specific situation unless he’s almost certain he’s going to succeed, especially when the hitter at the plate has a decent shot for an extra-base hit and an out would end the game.
Even 80 percent certainty of success wouldn’t have been good enough in that situation. The odds are still probably against the Yankees even if Ruth steals the base and the batter in that situation has a pretty good shot at a game-tying extra-base hit, whether Ruth is on first or on second.
AH: I disagree that one should always be “almost certain of success” before stealing.
That really depends on inning, score, pitcher, batter, and more. You’re down one run in the ninth inning, two outs, a singles hitter at bat against a dominant closer—you should be willing to gamble quite a bit.
What percentage is needed to justify an attempt in that situation? It can’t be known, just as you can’t know the likelihood that the runner will be successful: his overall success rate may be a poor predictor in the specific situation.
That’s why when you quibble over the particulars of the Ruth example (e.g., whether Meusel’s OBP of .361 tips the balance), you seem to me to miss our main point. When Babe is standing on first base deciding whether to steal, he has to take into account whether or not the pitcher is holding him on tightly, short-striding or not, throwing fast balls or breaking balls, and a host of other situation-specific variables which he can’t think about because he doesn’t even know. The decision whether to steal necessarily rests on old-fashioned judgment and intuition.
Q: You seem to argue that from a sabermetric and statistical perspective, Pete Rose doesn’t appear to have Hall of Fame credentials because his career on-base percentage and slugging percentage were both too low and that the primary reason he’s considered a solid candidate statistically is because he played for so long and racked up impressive counting statistics.
Context matters, and sabermetric stats that attempt to adjust for context suggest that Rose is indeed at least a decent Hall of Fame candidate. Plus, longevity matters to some degree.
AH: This was the chapter in which we praised sabermetrics’ major contributions but also argued that some people overrate those contributions. We used Rose as a case in point of a player who, with the benefit of sabermetrics, we realize was overrated.
Just compare his OPS to many players who no one thinks worthy of Hall of Fame consideration. But we also point out that Rose had spectacular intangibles, and these must be taken into account when evaluating a player. Rose is hard to rate right—considerably overrated if you don’t crunch the numbers, considerably underrated if all you do is crunch them.
Q: You argue that Rose is a Hall of Famer largely because of intangibles. If intangibles are a primary reason why a solid player like Rose should be in the Hall of Fame, why not put someone like Brett Butler in the Hall? I don’t see how taking intangibles into account makes Rose a Hall of Famer when, as you claim, he doesn’t have the statistics, but intangibles do not make someone like Brett Butler a Hall of Famer.
AH: Judgments about who belongs in the Hall of Fame are extremely subjective but I’m not sure what we said that you disagree with.
Both statistics and intangibles are relevant to assessing whether a player belongs in the Hall. Rose clearly belongs (putting aside gambling, an issue we don’t touch) and Butler obviously doesn’t. Rose has better statistics than Butler, and may have better intangibles too.
Q: You write, “When data trumps all else, you end up…underrating Rickey Henderson and Mickey Mantle.”
I don’t know many sabermetricians who underrates Henderson and Mantle. In some respects, sabermetricians argue that Henderson and Mantle were underrated and belong in a tier right at or just below elite-level Hall of Famers like Ruth, Williams and Mays.
AH: You’re right that by emphasizing OBP, sabermetricians enhanced appreciation of both Mantle and Henderson.
In context, we were making a specific point about the value of Mantle batting behind Maris and the value of Henderson in unnerving and tiring pitchers. I think we make a strong case in the book that these non-measurable contributions (and, of course, similar contributions by other players) have been underrated by sabermetricians.
Q: You spend a great deal of time on whether hitters own certain pitchers. I’ve read sabermetricians who argue that while we can’t say for certain whether hitters own particular pitchers, we may be able to determine whether hitters may own certain pitches.
For all intents and purposes, this may be a minor distinction but a distinction nonetheless. And I think sabermetrics is closer to your view on this subject than you realize.
AH: This was in the context of whether Joe Torre should have played Enrique Wilson against Pedro Martinez in the 2003 ALCS when Wilson seemed to own Pedro, but based on a very small sample size.
We talked about the way such decisions were traditionally approached, and contrasted that with how we think sabermetricians would have approached it based on an interesting article by James Click.
And we proposed a “third way” which synthesized aspects of the traditional approach and Click’s perspective. If you’re saying sabermetricians would actually embrace our analysis, my answer is: I hope so. We’re not looking to pick fights for the sake of it. There are any number of places in the book where we express agreement with sabermetricians.
Q: You bring up the Minnesota Twins as an example of a very successful anti-sabermetric team in the “Moneyball” era. I would argue, in a broad sense, the Twins are in fact a “Moneyball” team, although I agree they are not really a sabermetric team.
Again, I think this points out the flaws of the narrative within Moneyball of stats versus scouts. Sabermetrics is more about meaningful evidence (mostly statistics) versus seemingly intuitive guessing or meaningless statistics. The Twins and other quality organizations, like the Phillies, fall into neither of these categories. And most serious sabermetricians will tell you that it’s better to look at no stats than the wrong stats.
AH: We quote Twins manager Ron Gardenhire and their assistant general manager Rob Anthony about their contempt for sabermetrics. Rob Neyer says they show an “utter lack of sophistication regarding statistical analysis.” In any event, we can agree that they’re doing something right.
Q: You make the claim that one reason sabermetrics is misguided is because there is not a narrow path or formula for team success.
I would argue there is.
The formula is outscoring your opponents through good offense, good pitching/defense or both. There is strong correlation with some statistics and team success. A team doesn’t necessarily have to use sabermetrics to outscore opponents, but I think you have to admit sabermetrics made a significant contribution into which player attributes were overrated and which were underrated.
AH: Yes, the formula for winning is to outscore your opponent!
We point out that there’s enormous variety in the construction of successful teams (regular season and postseason alike).
Teams win with great offense or great defense or both, and offense built around power or small ball or both—every permutation. When you say we “have to admit” sabermetrics has made a contribution to baseball understanding, I’ll go further: we not only “have to” admit it but we do so without reluctance.
Q: It seems you misinterpret Dayn Perry and Nate Silver’s study on postseason success.
I don’t think any sabermetrician would argue that luck isn’t a huge factor in winning over the course of 5-7 games. The study was about factors that may influence postseason success, not coming up with a definitive formula for guaranteeing postseason success.
AH: In a way I hope you’re right, because I’m a fan of Nate Silver—particularly his political analysis. If you convince me that FRAA and WXRL are in fact meaningful statistics, and weren’t used by them tendentiously, I’ll admit the error. But we may want to have that conversation in private lest we put most of your readers to sleep.
Q: Regarding Ultimate Zone Rating (UZR), I think the section of the book on it speaks to a misunderstanding of sabermetrics as claiming it to be final and complete.
UZR measures something and attempts to adjust those measurements for context. No one claims that it’s flawless. But neither is watching every play a defender makes. Does that mean we should discount watching games? In the same way, we shouldn’t discount UZR. Both ways of analyzing fielders is useful.
AH: I readily agree that just watching defense will not always yield reliable assessments. That point extends to everything.
We mention a scene in Ball Four when Bouton starts reciting statistics to let his manager know how well he’s been pitching. Joe Schultz says, “Aw ****, I don’t want to see any statistics. I know what’s going on out there just by watching the games.”
We do not endorse Schultz! Rather, I agree with what you said earlier—the key distinction is between statistics that are meaningful and those that are not.
With respect to those that are, there’s a question of how meaningful. In the book, we try to show why UZR is not very meaningful. Is it possible that better fielding statistics will be developed that don’t share some of UZR’s flaws? I’m skeptical (so, apparently, is Bill James)—this may be a case where more and more data simply do not help overcome inherent limitations in the measurement.
Q: You write, “We are, needless to say, not opponents of data. To the contrary, as should be clear, we’re prone to traffic in numbers ourselves. But one needs to do so with a healthy dose of skepticism and awareness of limitations. One senses sabermetrics careening almost randomly from one pole to another. Baserunning and defense are overvalued, then undervalued.”
But, in a broad sense, that was pretty much the whole point of Moneyball. Players’ market values often careen almost randomly from one pole to another. I think sabermetricians are more aware of its limitations. No one only uses sabermetrics or statistics and most on the scouting side do not avoid statistics. The “holy war” is overplayed, and Moneyball certainly didn’t help to put this “struggle” into the appropriate perspective.
AH: Sabermetricians are not monolithic. But do many of them overstate the extent to which baseball decisions can be quantified? I think we make pretty good case that they do. It’s hard to discuss this in generalities, but we give examples throughout the book.
Q: I’ve never known a sabermetrician to write or say, “a walk is as good as a hit.” You make the claim that it was actually Little League that taught us what sabermetricians claim to have taught us.
But I think most sabermatricians would take a player who posts a high on-base percentage via hardly any walks, especially if that means more extra-base and home run power. How often a player gets on base or how many bases he gains at one time is more important than how a player gets on base. Sabermetricians understand this as well as anyone.
AH: To be fair, what you’re talking about was in the chapter that discusses sabermetricians’ contributions…Some of their insights did not emerge ex nihilo, and in that context we note that the value of the base on balls was apparent to many observers long before sabermetrics made OBP a point of emphasis.
But credit where credit is due and we give credit where appropriate to lots of folks, including Michael Lewis, Billy Beane and (very much so) Bill James. Contrary to what many of our critics (those who have not read the book) assume, and as I think you can attest, Short Hops isn’t a Joe Morgan-like screed against sabermetrics.
Q: You make the common mistake of equating on-base percentage with walks. But it’s about baserunners and avoiding outs. I don’t know any sabermatrician who is more concerned with how a player arrives at a high on-base percentage than if a player arrives at a high on-base percentage.
Most sabermetricians would agree with you that Kevin Youkilis was more valuable in 2008-2010 when he walked less but posted a higher cumulative on-base percentage and slugging percentage than in his previous seasons when he walked more.
In other words, most sabermetricians have always understood that walks and even on-base percentage aren’t the be-all, end-all.
AH: The Youkilis example was in the specific context of an irony in Moneyball.
Lewis emphasizes Beane’s annoyance with players who are impatient at the plate, over-valuing power and under-valuing walks. The A’s front office worshipped Youkilis (“The Greek God of walks,” though he isn’t in fact Greek), and we point out that Youkilis became a superstar only when he changed his approach at the plate in the direction that Beane generally opposed.
Q: You fail to address the fact that team on-base percentage has a very strong correlation with runs scored. I know, correlation isn’t causation, but it’s not just correlation; it’s also reasoning. The more baserunners, the more likely a team is to score runs. But slugging also matters. I’ll get to slugging later.
AH: Actually, we’re very clear that the emphasis on OBP was a major contribution by sabermetricians. That’s because it correlates with runs scored—that’s what counts.
Q: When Jack Cust finally got a chance to play regularly, with Billy Beane’s A’s, he slugged .457 during his time with the A’s and Giambi slugged .445 with the A’s. These are not outstanding slugging percentages, but hardly Eddie Yost and Eddie Stanky, especially when you consider Oakland is not really a home run park.
In your “cheers” for sabermetrics, you absolutely ignore the second key batting statistic that sabermetrics helped bring to the forefront as the sister stat to on-base percentage: slugging percentage. No sabermetrican prefers players who are like the Eddies and are likely to post higher on-base percentages than slugging percentages.
AH: It’s not true that we ignore SP.
We write, ‘Of course OBP isn’t everything. To many sabermetricians, OPS (the sum of OBP and SP) is the best gauge of offensive production.’
We agree with all sabermetricians that some combination of OBP and SP captures performance better than either statistic alone, while either alone is more revealing than batting average.
The discussion of the “Eddies” that you refer to involves an effort to answer this question: If Beane values players who are underrated because of high OBP, why hasn’t he acquired more of them?
Cust and Giambi are examples of such players. (The fact that they also hit home runs—more power to them, ha-ha.). The question is why Beane hasn’t found more players like them and continued to get such great bang for his buck. I think we offer some good explanations (which, for the record, do not denigrate Beane).
Q: Another aspect of sabermetrics you fail to address in your “cheers” section is sabermetrics’ attention to context, especially position scarcity and park effects.
The reason the Eddies were as valuable as they were was because they were middle infielders. It’s always been harder to find a middle infielder who can actually play even adequate defense in the majors everyday yet still post respectable on-base and/or slugging percentages.
A huge contribution of sabermetrics, I would argue at least as important a contribution as bringing on-base percentage and slugging percentage to the forefront, is its attempt to bring context to the world of statistics.
AH: Perhaps we should have talked about that more. We do note that, long before Bill James came along, there were plenty of statistics thrown around. James’ search was for meaningful statistics, and to a large extent he succeeded. But later generations of sabermetrics have also produced a plethora of less helpful statistics, as James himself has acknowledged and lamented.
Q: The last 50-or-so pages of the book deal with the strange occurrences during 2009 Red Sox games, unique qualities of some individual players and other things that help point out that the beauty of the game is its majesty, mystery and colorfulness.
You imply that sabermetricians (with the possible exception of Bill James) do not appreciate the game’s “majesty and mystery” because sabermetrics reduces the game to numbers.
Just because someone pours themselves into sabermetrics, or has an appreciation for sabermetrics, does not mean they find the mystery and majesty of the game disturbing. In fact, I would argue the game’s qualities that make it somewhat measurable and somewhat predictable—that makes its strange occurrences even more enthralling.
AH: We emphasize James’ fascination with all sorts of extra-statistical aspects of baseball, and we certainly don’t say he’s the only one. On the other hand, we give plenty of examples of sabermetricians missing the forest for the trees.
Q: Some devote themselves to sabermetrics, but that doesn’t mean those people reduce the game to pure numbers and statistics or view players as robots or a series of zeros and ones.
Bill James defines sabermetrics as a search for objective knowledge about baseball. Just because some baseball analysts prefer that search for objective knowledge about baseball, does not mean they are closed off to the subjective, the mysterious, the majestic aspects of the game.
AH: Here’s a quote from the book: “While I still believed that numbers could reveal things about the game that were invisible to the naked eye, my own eyes had glazed over as the combination of fantasy baseball and mathematical arcana conspired to squeeze the life from the game I loved.” That’s not us talking. That’s John Thorn, a leading sabermetrician.
Q: With all due respect to you and Mr. Thorn, I see no reason why in-depth statistical analysis and sabermetrics would squeeze the life out of the game unless one is reaching for something that will squeeze the life out of the game.
I’m not a sabermetrician, but I’m very sympathetic to sabermetrics and try my best to learn and understand as many sabermetric concepts as possible. Perhaps it’s because I’m not really a sabermetrican that I don’t understand the joylessness of those sad sabermetricians who are merely watching the game of zombie or robot baseball.
The predictable and the statistical have taught me a great deal about the game and give me more appreciation of its majesty and mystery, not less.
I appreciate the unpredictable as much now as I ever have, largely because I have a better understanding, through statistics and sabermetrics, of what the numbers say is supposed to yet doesn’t happen. The fact that the meaningful numbers usually get things right makes the unexpected events in baseball seem even more miraculous.
I would argue that sabermetrics in a certain sense is an anti-statistical movement in that it opens the door to the organic parts of the game. Sabermetrics make baseball statistics into a language and not just cold and limited symbols on the backs of baseball cards.
AH: If you see no merit in Thorn’s reflection, and disagree that sabermetrics has been taken to excess, I doubt I can convince you. But you gave Short Hops a thoughtful read and clearly take seriously the issues we raised. That’s all we can ask.
From the Short Hops website: “Alan Hirsch, a visiting professor at Williams College, is the author of numerous books and articles. His articles on sports and other subjects have been published in the Los Angeles Times, Washington Post, Washington Times, and Newsday, among many other publications.