Smart Baseball

by Keith Law

  How has more than a half century passed without this obviously flawed statistic losing its place in basic evaluations of player performance?

  It fails a simple logical test of its merits, and it has repeatedly produced the wrong conclusions in questions of player value. The RBI stat is responsible for more bad postseason award decisions than any other single factor in baseball history. And while I can’t confirm this, I believe strongly that it is responsible for more stupid comments by game announcers than any other statistic, too.

  Though the RBI was not an official MLB stat until 1920, it’s now enshrined in rule 10.04, which grants the batter a run batted in for every run that scores as a result of a ball he safely hit into play, or as a result of him drawing a walk or being hit by a pitch with the bases loaded, or as a result of him making an out (except for a force double play) that allows that run to score. While most RBI are awarded automatically, there are some situations, mostly involving fielding errors, that allow the official scorer to have discretion over whether to award the batter an RBI—and any stat that involves such human subjectivity is immediately reduced in value as a result. People are prone to so many cognitive biases and are so inconsistent in their judgments that allowing them to award or withhold an RBI seems like far too much responsibility.

  Rickey nailed the most fundamental problem with RBI, however: it is an individual statistic that depends far too much on the performances of other players—in this case, whether the hitters ahead of the player get on base enough for him to drive them in. Barry Bonds, MLB’s all-time career home run leader with 762, hit 450 solo home runs in his career, and had 412 baserunners on for his other home runs. Hank Aaron, whose record of 755 Bonds broke shortly before retiring, hit 399 solo home runs—and he had 482 baserunners on for his other home runs. So even though Bonds out-homered Aaron in about 300 fewer games (and got on base more often and hit for more power in total), Aaron ended up with about 300 more RBI in his career. Bonds was the better hitter, but Aaron gets the RBI glory.

  Bonds holds the dubious distinction of being one of only two hitters in MLB history to qualify for the batting average title (503 plate appearances in 162 games) and drive himself in more than he drove in his teammates. The year he set the all-time single-season home run record of 73, in 2001, he had just 137 total RBI; since he drove himself in 73 times via home runs, that means he drove in his teammates only 64 times. As a result, the season that is number one on the home run list doesn’t crack the all-time top 120 for RBI in a single year.

  Bonds hit in the third spot in the Giants’ lineup for nearly all of that historic season, 137 of his 148 games started (he batted fourth in the other eleven starts), yet came up to the plate with the bases empty in 54 percent of his plate appearances, and only had 393 baserunners aboard for him the entire year. He had seven other seasons in his career where he had more baserunners, and thus more opportunities, available to him than he did in that record-setting year. That’s because Dusty Baker, then the Giants’ manager—and, as this book goes to press, inexplicably now the Nationals’ manager—chose to split the leadoff position between two hitters, Calvin Murray and Marvin Benard, who were not good at the most important role of the leadoff hitter, getting on base:

  Neither player would play regularly in the majors again after 2001.

  The Giants did have an excellent OBP guy in the second spot, Rich Aurilia, who posted what would be by far the best on-base percentage of his career, .369, in that 2001 season—thanks to a career-best .324 batting average, because, really, don’t you want to throw him a strike with Superman standing in the on-deck circle? (More on that subject later in the book.) But Aurilia compromised Bonds’s opportunities to drive in runs by hitting a career-best 37 homers of his own that season and a career-best 37 doubles. It wasn’t all bad news; Bonds knocked Aurilia in 25 times in 2001, but the net result of two low-OBP guys splitting leadoff duties and a big power bat in the number-two spot ahead of Bonds meant that Bonds himself had unusually few RBI opportunities, and thus a relatively low RBI total for a guy with 73 home runs.

  How do we judge Barry Bonds’s 2001 season, then? Do we look at the all-time single-season home run record, what is now the ninth-best single-season on-base percentage of the modern era (Bonds himself had a higher OBP in each of the next three seasons), and the all-time single-season slugging percentage record, and say that this was the single greatest offensive season baseball has ever seen? What information does the RBI total impart to us that those other stats, ones that are independent of what his teammates did or any stupid manager tricks, could not?

  The answer, of course, is: nothing of use.

  The personal epiphany for me on the RBI’s dishonesty was Joe Carter’s season in 1990 for the San Diego Padres, his one season in California before he went to Toronto in one of the most star-laden trades in baseball history, a deal that sent him and future Hall of Famer Roberto Alomar to the Blue Jays in exchange for near Hall of Famer Fred McGriff and Tony Fernandez. Carter was very much a product of his time, a low-OBP slugger who would have a hard time holding an everyday job in the majors today but who spent twelve years as a full-time regular in the 1980s and 1990s, including playing every game from 1989 through 1991.

  Carter’s one year in San Diego was the worst of his career, as he hit for a career-low .232 batting average with a .290 OBP that was his career low until his last year as a regular at age thirty-seven. Yet Carter finished third in the National League in RBI that year with 115, even though he made more outs at the plate than any other hitter in the NL that season. How is a guy near the top of his league in RBI with all of those outs made?

  Some of it is just power; Carter hit 24 home runs that season, good for 16th in the National League, and 27 doubles, although his slugging percentage, which is just total bases divided by at bats and thus incorporates batting average, was below the NL median. But Carter had 542 baserunners on board for his league-best 697 plate appearances—the most baserunners Carter had in any single season of his career—because he was behind a trio of on-base machines:

  The 542 baserunners on base for Carter’s times at the plate were the most in the National League in 1990 by a huge margin; the next-highest mark was 496, so Carter had 46 more guys to drive in. In fact, that 542 figure for Carter would have led the National League in 1988 . . . and 1989 . . . and 1991 . . . and 1992 . . . and 1993. After Willie McGee had 544 baserunners on for him in the “rabbit ball” high-offense year of 1987, only two batters had more men on base for them than Carter did in 1990 through the 2002 season. Carter’s RBI total might have been third in the NL, but it’s a sad fraction of how many he might have driven in if he’d just been a better hitter to begin with.

  There’s zero sense in crediting Carter for having all of those extra opportunities. We can credit him for the hits he got with men on base, but we already do that in other statistics, like batting average and OBP and slugging, that don’t include the noise of RBI. There is a strain of baseball thinking that we should think differently about hitters who perform especially well or poorly with men on base, a subject to which I’ll devote some time in a later chapter, but for now I’ll assert that this isn’t actually a separate skill for batters. Hitters hit, as the late Tony Gwynn liked to say. It doesn’t really matter who’s where when they do.

  One of the most common counterarguments to my “RBI are meaningless” philosophy is that the team with the most runs wins each game, so how can a stat that keeps track of who drives in those runs be a bad stat? Like many political statements, this argument confuses team stats and individual stats and relies on the listener to fall into the same trap.

  Yes, the team that scores more runs wins the game—I can’t argue with that—and tracking runs scored at the team level is an important measure of the quality of a team’s overall offense. A lineup may look good on paper, but if it’s not generating runs, then it’s probably not as good as it originally looked. (There could be other explanations, too.) However, those ru
ns are team events, generated at the team level, and other than home runs, they require more than one batter to happen.

  Baseball essentially double-counts runs when working at the individual level: One player scores the run and gets a “run scored”; another player drives in the run and gets a “run batted in.” That’s two “runs” in individual stats for a single run on the field—maybe, since a player who knocks in a run by grounding into a double play doesn’t get an RBI, and no RBI is awarded if the player scores on a fielder’s error, a wild pitch, or other unusual event.

  If a player did something in between to advance the runner, such as a single that advanced a runner from first to third, he gets a pat on the back (or somewhere else). So it might make more sense to think about the events that lead to a run scoring in fractional terms for the hitter, which is the philosophy behind “linear weights” methods of evaluating offense. With linear weighting, any event from a hitter is worth, on average, some fraction of a run, so if you assign the right fractional amount of a run to each of those events and add them all up, you’ll get a number that measures in runs the value of all of that hitter’s actual production for a given time period. When a hitter gets an RBI, it creates (or fosters) the impression that this hitter alone was responsible for the full creation of the run, and therefore the hitter with the most RBI on a team or in a league did the most to help his team win. This mistake thrives on the confusion I mentioned above, conflating a team activity with the attempt to measure individual performance. We are not trying to glean team performance from individual stats, so we shouldn’t try to glean individual values from team stats.

  When we look at the statistics of any single player, we are looking for one or both of two things:

  1. What did the player actually do in his time on the field?

  2. What might those statistics tell us about his likely performance going forward?

  RBI give a partial answer to the first question; he did drive in those runs, so in the strict sense of counting the number of times that the hitter did something good that resulted in a run scoring, RBI more or less get the job done. But the way they’re counted has contributed to the impression that the hitter who drives in a run created that whole run, which is wrong. (You could argue that it’s not the RBI’s fault, but I’ll leave that to the philosophers.)

  As for the second question, the answer is simple: nothing. A hitter who had a lot of RBI in a specific season played a lot and batted often with men on base, but that can’t tell us that the same things will be true in the following year.

  Runs Created is a simple measure of total offense that correlates pretty well with a team’s total run-scoring, even though it lacks the precision of something like linear weights. Created by Bill James in the 1970s, Runs Created is most simply calculated as OBP * Total Bases, which is a simplified version of the formulas he initially presented in his Baseball Abstracts.

  To return to Runs Created again as a very simple measure of total offense, RBI do a much worse job of predicting a player’s offensive output the following year (as measured by RC) than his Runs Created total does. This correlation analysis measures how RBI per 100 plate appearances (PA) and RC per 100 plate appearances appear to “predict” RCs the following year:

  So, if you want to try to figure out what kind of offense a player might produce in the following season, his RBI rate tells you less than you would get from looking at his RC rate. In other words, the RBI column is just making us dumber. And that’s by comparing it to a very simplistic stat like Runs Created.

  But it has long made everyone dumber, because for much of the statistic’s history, RBI was interpreted as meaning a player was a good hitter or had some sort of magic woo-woo that gave him a special ability to drive in runs. (Whether that same woo-woo also led to the hitters ahead of him getting on base more often was generally not acknowledged.) Hitters with high RBI totals would be placed in spots in the lineup that typically get more RBI opportunities, creating a sort of offensive death spiral when the hitter in question was someone like Joe Carter, more opportunist than actual run producer.

  As we saw earlier in Carter’s extreme case, a hitter can still rack up a lot of RBI even when he’s not an especially good hitter. But when we put a good hitter behind one or more hitters who get on base often, we can get enormous RBI totals. That’s how Rickey Henderson, the best player in the American League in 1985, helped fellow Yankee Don Mattingly win the AL MVP award that year.

  Henderson reached base safely 274 times in 1985, good for fourth in the American League even though he missed 19 games that year, and led the league with 80 stolen bases. Mattingly, playing in all but three games that year, reached base 269 times—no slouch himself—but also drove in Henderson 56 times, helping him lead the AL with 145 RBI and win the award. When Hack Wilson set the major-league record for RBI in a single season in 1930 with 191—a record that still stands today—even he didn’t drive in any single teammate more than 45 times.

  In 2004, when Barry Bonds reached base an absurd 61 percent of the time, thanks in large part to 120 intentional walks (a record that I hope for all of our sakes is never broken or even approached), the cast of characters who had the good fortune to hit behind him all racked up RBI totals out of proportion to their actual performances. Edgardo Alfonzo batted directly behind Bonds more times than any other Giant that season, in 57 games, and knocked in 77 runs despite hitting just 11 homers and slugging only .407. A. J. Pierzynski had an almost identical line, with 11 homers, a .410 SLG, and 77 RBI, because he hit in the two spots behind Bonds more often than any other player.

  But any readers age forty or over have probably been wondering when I’d bring up Tommy Herr, whose 1985 season earned notice even at the time for its outlier status. Herr was the slap-hitting second baseman for the St. Louis Cardinals, who won the NL pennant that year under manager Whitey Herzog, who encouraged a style of play called “Whiteyball,” with lots of athletic, frequently African American players who could really run; the Cardinals had five players steal at least 30 bases in that season, led by Vince Coleman’s 110, and they stole 130 more bases than any other team in the NL that year.

  Herr wasn’t as fast as Coleman or Willie McGee; he did steal a career-high 31 bags in 1985, but that’s not why we’re stopping at his house on memory lane right now. Herr hit only eight home runs that year, but knocked in 110 runs, becoming the first player since 1950 to knock in 100 or more runs while failing to hit ten homers. (Paul Molitor would later do it in 1996, but no one else has done it since.) Such high RBI/HR ratios were common prior to World War II, but the rise in home runs as baseball expanded and players became stronger had made them a thing of the past, until Herr and Whiteyball.

  Herr’s slugging percentage of .416 that year remains the lowest for any hitter who had 100 RBI in a season since 1938, and he owes nearly all of this to the three men listed above and their fleet feet.

  Herr had another anomalous RBI season a few years later. The 1987 season was known as a “rabbit ball” year because home runs spiked across the game with no apparent explanation, although the widespread theory at the time was that MLB had changed the baseball. The power spike brought home runs to plenty of hitters who weren’t normally long-ball guys—Wade Boggs hit 24 home runs in 1987, but only reached double digits in one other season, the strike-shortened 1994 year, when he hit 11—but not to Herr, who hit only two balls out of the park that year. Thanks to his speedy teammates, however, Herr became the first and still only player since 1943 to drive in more than 75 runs while hitting two or fewer homers; he ended up with 83, an RBI/HR ratio that would stand almost alone in modern baseball were it not for Ozzie Smith, still Herr’s teammate that year, who had 75 RBI without a home run at all. That remains the highest RBI total for a homerless season since World War II.

  The Cardinals’ middle infielders were not great offensive players, although Smith was one of the best defensive shortstops the game has ever seen and made himself into a competent con
tact hitter. Their RBI totals were the products of their team’s system, one that put guys on base and kept them moving into scoring position, so that it didn’t take power, or even a hit, to drive them in.

  RBI just don’t tell us anything useful about a player’s individual performance in a game, a season, or a career, but they remain prevalent in the minds of writers and fans. In November 2014, Rob Neyer wrote a piece on that year’s MVP voting results for Fox Sports’ Just a Bit Outside blog where he said that “the writers’ obsession with RBI guys on first-place teams has long outlasted any excuses for it, and I’m tired of them.” High RBI totals continue to pollute MVP ballots, albeit not to the extent they did twenty years ago, and to derail discussions of actual player value. Mike Trout lost two MVP awards to players with higher RBI totals on more successful teams, even though in both years (2015 and 2012) Trout had the better seasons. Ryan Howard won the NL MVP award in 2006 thanks to his 149 RBI and 58 home runs but wasn’t even the most valuable player on the right side of the Phillies’ infield that year.

  There’s really nothing RBI tells us about the player that we couldn’t glean from other, less fuzzy statistics. The same is true about saves, perhaps the most ridiculous of all of the traditional stats because it has actually changed the way the game is played—unequivocally for the worse.