Baseball has always been a numbers game, but players and teams are getting smarter than ever — and fans are, too. Throughout The Ringer’s 2017 MLB preview, in a series we’re calling “How to Baseball,” our experts will explore the developments that stand to change the way the game is played and consumed. We’ve never known more, and while knowledge is power, it’s also a wellspring of questions. We hope to answer some of them — and to remind you all to bunt wisely.
In 2006, the most popular statistical category in Yahoo fantasy baseball was runs batted in, which was used in 99.50 percent of Yahoo leagues. By then, RBIs had been a staple of rotisserie baseball’s original eight categories for more than 25 years, and the early stirrings of the sabermetric movement couldn’t dislodge it from the fantasy firmament overnight.
2006 wasn’t the dark ages, sabermetrically speaking — Moneyball had been a best seller for years, Theo Epstein’s (and Bill James’s) Red Sox had won the World Series, and Andrew Friedman and Joe Maddon had just started running the (Devil) Rays. Baseball Prospectus had just turned 10, and FanGraphs had debuted (albeit barely). Still, those weren’t quite enlightened times. Anti-stat sportscaster Joe Morgan was still active, as was the blog about him. MIT Sloan hadn’t yet held its first, sparsely attended sports analytics conference. Many MLB teams were still years away from hiring their first full-time quant.
In the intervening decade, RBIs’ reputation eroded. Baseball embraced advanced analytics, both in front offices and in the larger cultural conversation. It became common knowledge that runs batted in were a poor measure of player performance, dependent as they are on timing, quality of teammates, and lineup position, among other factors. Naturally, RBIs’ Yahoo usage rate dropped — all the way down to 99.31 percent.
Even in 2016, RBIs remained the most popular category in Yahoo fantasy baseball, according to data provided by Yahoo. (Yahoo, which will say only that it has “tens of millions” of registered users across all of its fantasy sports, has offered fantasy baseball since 1999, but pre-2006 usage rates weren’t easily retrievable.) The top 10 categories in 2006 — the bedrock stats of the standard “5x5” league — were still the top 10 categories last year, in a slightly different order. Almost 40 years after its founding, fantasy baseball is straddling analytical eras, clinging to archaic scoring categories even as fantasy prep grows more sophisticated by the season.
The 5x5’s hegemony did slip slightly between 2006 and 2016: Each of the 10 basic statistical categories was used in 92.8 percent of Yahoo leagues last season, on average, down from 96.3 percent in 2006. Among the 55 categories offered by Yahoo in both 2006 and 2016, the biggest losers by reduction in usage rates are old-school stats that have fallen out of favor due to sabermetrics or changes in player deployment: shutouts (minus-11.0 percentage points); batting average (minus-10.5); complete games (minus-8.8); earned run average (minus-6.7); hits (minus-6.6); and errors (minus-5.8).
But it’s not as if they’re losing ground to sabermetric mainstays such as on-base percentage (which is almost unchanged, slipping to 12.2 percent from 12.3 percent). Mostly, they’ve ceded territory to the 30 new categories Yahoo has added since 2006, led in adoption rate by several not-nerdy stats, some of which are rare enough to be almost frivolous in fantasy: quality starts; no-hitters; perfect games; grand slams; blown saves; and hitting for the cycle.
Every spring, serious fantasy players pore over data-driven projections online and in stat-packed books such as the Baseball Prospectus annual and Ron Shandler’s Baseball Forecaster. They read about regression candidates who had high or low BABIPs or, more recently, launch angles and exit velocities that don’t support the previous season’s surface stats. They scrutinize spring training stats for signs of new pitches or fastballs that have gained or lost speed. And then the bulk of them still use all of the new-age intelligence they’ve assembled to construct teams with the most pitcher wins or best batting average.
The process sounds a bit backward, like buying a 4K TV to display a standard-definition program. Or, to bring things back to baseball, like building a high-tech tracking system capable of recording pitch locations with a computer’s precision, and then using it to evaluate human umpires’ performance rather than make ball-and-strike calls directly. But just like the much-maligned human element of umpiring, which gives fans something extra to argue about and leaves a little leeway in the strike zone for skilled players to exploit, fantasy’s primitive underpinnings are a sometime source of frustration that on balance may make it more fun.
“I think the allure of fantasy is that it tests our ability to project the future,” Shandler, a pioneer in applying advanced stats to fantasy, tells me via email. “But we naturally think in terms of traditional surface stats. … Most serious players know that projecting HRs from HRs alone is not the optimal process, so we do gravitate to hard-hit-ball rate, and exit velocity, and soon barrel rate … but only as a means to project HRs. It has more inherent meaning to use advanced gauges to attempt to project HRs than to project a metric like isolated power.”
Once the lineups are set, fantasy turns into spectator experience. And it’s hard to argue that that experience — monitoring each player’s production in real time, assessing its impact on team totals, and trash-talking accordingly — would be better with complex stats that couldn’t be calculated or easily intuited on the fly.
“We are wired to see home runs, stolen bases and strikeouts,” Shandler says. “Those are visual event outcomes, and as such are more accessible. … I still want to be able to go to the park, watch Giancarlo Stanton hit one over the fence, and know that his HR has direct impact on my fantasy team’s bottom line.”
Even though the fantasy norms at Yahoo have hardly evolved, the ongoing proliferation of fantasy formats has produced variants whose scoring systems are more closely linked to real-life value. Daily fantasy, for instance, is largely governed by luck in small samples, but for those who keep coming back, it prioritizes performance in categories that resemble wOBA more than batting average.
ESPN daily-fantasy sports analyst Derek Carty describes daily fantasy via direct message as “much closer to real-life than 5x5 but not precisely. I think a good comparison is that it’s kind of like slugging percentage. It’s much closer to valuing each event’s contribution properly than batting average is in that it weighs events differently, but the weighting is still somewhat arbitrary.”
Carty, a former writer for The Hardball Times and manager of fantasy content at Baseball Prospectus, developed his own DFS projection system, which he uses to search for small edges that add up over time. Paradoxically, divorcing the process from the results by using advanced stats to project simple ones makes it easier for him to set himself apart from the pack of casual players.
“There are people who want [fantasy] to reflect real-life value as closely as possible, but I don’t see the point,” Carty says. “Especially if you try to tease out the effects of defense and whatnot. What’s the fun in only trying to predict, say, [defense-independent statistics] for a pitcher? They’re so stable that it would be way too easy. Sure, there’s variance in runs and wins, but they are predictable to an extent, and by including the effects of defense and parks and umpires … it gives you more things to analyze and more ways to gain an edge.”
Craig Goldstein, one of the hosts of the Baseball Prospectus fantasy podcast There Is No Offseason, says that every league he plays in uses RBIs and batting average. Despite his affiliation with the site whose PECOTA projection system tries to make baseball less unpredictable, he appreciates the randomness endemic to traditional fantasy stats. “I like getting at the core of baseball, but fantasy is fun,” Goldstein says in an instant message. “I like the gamble involved.”
Like Carty, Goldstein also believes that the further fantasy is removed from state-of-the-art stats, the more sabermetrics can help him steal a march on a less attentive owner. “If I know someone’s [true average] is better than their BA, I can count on a bounceback,” he says, referring to BP’s primary rate metric for hitters, which is calibrated to the batting average scale but properly weights all aspects of offensive performance. “But if you just use [true average] you end up with a sortable table. That’s boring.”
For those attached to the sortable table, there’s Ottoneu, FanGraphs’ fantasy game. Ottoneu’s creator, Niv Shah, began to develop the game in 2005. After reading Moneyball, he found the 5x5 format “annoying on multiple fronts,” and his answer was Ottoneu: a game that replaced the snake draft and single-season leagues with an auction, the ability to keep as many players as the budget permits, 12-team leagues with 40-man rosters, and an unusually deep player pool that includes the minor leagues. It also promised to reward owners for the total value of the offense and pitching their players produced on the field.
Ottoneu offers four scoring systems: 4x4; 5x5; FanGraphs points (a linear-weights-based method that assigns run values to each event); and SABR points (like FanGraphs points, but with FIP-based pitching scoring that rewards pitchers based on how many runs their peripherals suggest they should have allowed). Shah tells me via email that Ottoneu has 231 active baseball leagues, of which 136 use FanGraphs points, leaving 53 for 5x5 and 21 apiece for SABR points and 4x4.
This is seemingly what Shah wanted when he founded his fantasy game, but he’s discovered that, for all the realism the alternative scoring systems offer, “4x4 and 5x5 make for better gameplay” than the sabermetric route. “There’s really just one answer to player valuation in that kind of [linear-weights] system,” Shah says. “Points are points, after all. In a 4x4 or a 5x5, I might need HRs more than you need HRs, so valuation for the same player that has the same production can differ based on our rosters.”
Even though obtaining that diversity of incentives forces owners to dumb down the stats, it fosters the kind of creativity that sabermetric scoring can kill. “[Points-league] owners don’t have to think as much when deciding how much to pay for a player or what a player’s trade value is,” Shah says. “4x4 and 5x5 add more variance. … People think they don’t want variance, but variance is what makes these kinds of games fun.”
As Carty puts it, “The goal isn’t necessarily to use advanced stats to figure out which players are best at helping their team win, but rather to figure out which players are best at accruing the stats that our game uses.” That might be the most lifelike aspect of fantasy. In baseball, a hitter who drives in a run, or scores one, may just have been in the right place at the right time. A pitcher who’s credited with a win may not have helped his team. But they can still be the heroes, because runs and wins are the goals of the game. In real life, it’s illogical to conflate a player’s “win” with his team’s. In fantasy, they’re the same.
And that’s why fantasy may never need to adapt, no matter how statistically literate its audience grows. An online league isn’t really a refuge for late adopters who still see pitcher wins as a worthwhile statistic — they’re the marks that the stat-savvy players prey on. But in fantasy, the anachronistic numbers that in real life we’ve learned to discount are as vital as most fans once believed them to be. “There are some leagues that use straight sabermetrics,” Shandler says, citing Ottoneu. “But I don’t see these ever becoming mainstream.”
Fantasy’s stasis is an acknowledgement that we crave something more than certainty in sports. Even with yardsticks such as WAR and JAWS that could be argument-enders, we still debate award winners and draw conflicting conclusions. As ESPN’s Sam Miller wrote after the latest perplexing Hall of Fame voting results were released, “It has never been less clear what a Hall of Famer is. Oddly, it has never been more clear what a Hall of Famer did.” Baseball stats have never been so precise, but there’s still a place for ambiguity.
Thanks to Sean Hamel of Yahoo for research assistance.
