The Bill James of Japan
How DeltaGraphs founder Yusuke Okada fell in love with advanced analytics and brought sabermetrics to Nippon Professional BaseballThe Bill James of Japan
How DeltaGraphs founder Yusuke Okada fell in love with advanced analytics and brought sabermetrics to Nippon Professional Baseball
If you study the box scores from the 2016 Japan Series, the season-ending showdown between the last surviving teams from Nippon Professional Baseball’s Pacific and Central leagues, you’ll spot two patterns that look like anomalies to American eyes.
First, there are the repeated appearances of Nippon Ham Fighters star Shohei Otani, who filled a triple role in the series as a starting pitcher, pinch hitter, and DH. But Otani is a singular sensation, as unique in Japan as he would (or will) be in the states. Another tidbit from those box scores is more representative of the broader differences between the MLB and NPB brands of baseball: In the six games it took to decide the series, the Fighters and the Hiroshima Carp combined for 16 sacrifice bunts. One was credited to Yoshihiro Maru, the Carp’s no. 3 hitter, who recorded a team-best .322/.408/.547 slash line during the regular season but bunted in the sixth inning of Game 2, with a runner on second, no outs, and a 2–1 lead.
By contrast, there were only three sacrifice bunts in last year’s World Series, which went seven games. That rate of 0.21 bunts per team game was a perfect match for the MLB-wide regular-season rate. In the majors, the sacrifice bunt is an endangered play; teams are bunting only half as often as they did as recently as 1994, and they’re setting new all-time lows with almost every passing year. This season, MLB players have bunted only 0.19 times per team game. NPB players, meanwhile, have bunted about 0.80 times per team game, 4.2 times the MLB rate. MLB teams haven’t bunted that frequently since 1939.
There are cultural causes of Japan’s continued embrace of the bunt. As Robert Whiting, author of the acclaimed book about Japanese baseball and culture, You Gotta Have Wa, told me for a previous article about Japan’s most extravagant bunter, Japanese baseball "has always been a team sport" that places a high "value on harmony, everybody contributing." In that environment, bunting is encouraged; in Whiting’s words, "the more the better."
But NPB bunting continues to flourish for another reason: The sabermetric movement that’s driving a stake into the sacrifice here hasn’t caught on to nearly the same extent in Japan. Each bunt is a symptom of an old-school, statistically suboptimal style of play. Whiting also told me that "Japanese baseball is conservative, orthodox, and tends to stay within its own M.O.," adding that "it takes a while for new wrinkles from MLB to catch on in Japan."
One man, Yusuke Okada, is trying to speed up that process. As Jim Allen, a Kyodo News writer who’s covered NPB for more than two decades, tells me via direct message, Okada is "on the cutting edge of sabermetrics and sports analysis in Japan." He’s Japan’s closest equivalent to sabermetric trailblazer Bill James, and in his own field, he’s as exceptional as Otani.
Okada, 42, was born in Chiba Prefecture and majored in political science at Seikei University, alma mater of Japan’s long-serving prime minister, Shinzō Abe. By the time Okada finished school, though, he was already gravitating toward a career that wouldn’t draw on his degree. Okada began following MLB on a casual basis in middle school, but while he was in college, the stateside success of Japanese ace Hideo Nomo, who won the Rookie of the Year award after joining the Dodgers in 1995 and finished fourth in NL Cy Young voting in each of his first two major league seasons, made him want to know more about baseball in America. A friend introduced him to Total Baseball, the baseball encyclopedia compiled by baseball statisticians and historians Pete Palmer and John Thorn, which in addition to presenting basic stats with unprecedented accuracy also included sabermetric stats such as James’s Runs Created, Tom Boswell’s Total Average, and Ted Oliver’s Wins Above Team, as well as Palmer’s linear weights and ballpark and era adjustments.
"I didn’t speak English at all at the time, but I read the book day in and day out, with help of an English-to-Japanese dictionary," Okada tells me via email and direct message responses transcribed and translated by Kazuto Yamazaki, a baseball writer who works with Okada. "The stat-based approach made it easier for me to read. … What was appealing to me was its effort to equate the values of hitting, pitching, fielding, [and] baserunning and put them together, which seemed revolutionary to me and pumped me up." At first, Okada didn’t have enough money to own his own copy. Now he has six editions.
In the pre-Moneyball era, sabermetrics in the U.S. was largely the province of outsiders, but at least those outsiders had company. In Japan, it was an even lonelier pursuit. "No one seemed to be aware of the existence of sabermetrics at the time," Okada says. "Hence there was no community, though maybe a scant few people were literally analyzing data using paper and calculators." Like James, who worked as a night watchman at a pork-and-beans cannery before he came to Sports Illustrated’s attention, Okada didn’t break into baseball immediately. After graduation, he worked midnight shifts at a call center, fielding calls from customers who were involved in car accidents.
In 2002, an acquaintance helped him land his first full-time sports-related job, overseeing stats for broadcasts on Nippon TV, a company that aired NPB games and had hired him part-time in college. The available data was limited, and anything beyond basic outcome stats required laborious manual work. "I had to provide stats that were appealing to viewers — numbers in individual hitter-pitcher matchups, whether players had been hot or cold, some other split stats, along with anecdotes from the players’ amateur years and milestones they were close to, which I passed along to the announcers and directors," Okada recalls. "Most [of the] audience couldn’t understand advanced stats, even after Moneyball made the rounds in Japan."
By the mid-aughts, Okada was absorbing the best of the American sabermetric movement, reading James as well as the sites that had sprung up to carry the analytical torch online, including Baseball Prospectus, The Hardball Times, and FanGraphs. No longer was he completely alone; "after the internet came around," he says, "statheads contacted each other online, and the Japanese sabermetrics scene came to existence." Frustrated by the rudimentary data he had access to with Nippon TV, he left in 2006 to supervise the baseball department of a company called Data Stadium. Founded in 2001, Data Stadium was a Japanese analogue to American companies like Inside Edge and Stats LLC. It was the country’s sole provider of pitch-by-pitch and batted-ball data, which it sold to media companies and NPB teams. That was an upgrade from what he’d had at his disposal before, but it still wasn’t enough to satiate his statistical appetite.
"Data Stadium profited from providing data and spent little to no effort on analyzing it," Okada says. "On top of that, the system they had was complicated, so I had a tough time getting the data I needed for advanced analysis I wanted to do, especially on fielding." Okada wanted to adopt American methods to slice and dice NPB data the way many were doing with MLB data, but he encountered resistance at every turn. "I tried to explain the significance of advanced research to my bosses at Data Stadium as well as people working for NPB clubs, but they rarely listened to me," he says. "For about 10 years from 2002, no one tried to understand, or even cared about sabermetrics. … I was rejected right away. It was really hard to explain advanced analysis to NPB club officials who were former players. Those were the toughest times in my career."
In his spare time, Okada single-handedly calculated defensive ratings for Japanese players, borrowing the framework from American fielding metrics. He also increased his outreach efforts in hopes of fostering a collaborative group. "I contacted some statheads outside of the company to build the keystone of the Japanese sabermetrics community," Okada says. "I thought the future was bright, if we could follow the path the American community had gone through."
Okada decided that to indulge his interest and advance the larger cause, he would have to leave Data Stadium and start his own site, one where he could gather all the data he wanted and wouldn’t have to traffic in trivial stats and small-sample ephemera. In September 2011, he founded DeltaGraphs, a company whose mission statement (via Google Translate) lays out an intention to "tackle the development of infrastructure aiming at [the] dissemination of sabermetrics in Japan." Okada and his collaborators, who rallied around DeltaGraphs’ banner and consolidated what had been a scattered community, have published several books devoted to statistics and research, and they’ve also created a website called 1.02 Essence of Baseball, a reference to MLB’s bedrock rule 1.02.
As the name suggests, DeltaGraphs is essentially FanGraphs for NPB (although it’s not affiliated with FanGraphs). The site includes a sabermetric glossary, live win-expectancy stats, projections, and advanced-stat leaderboards, as well as written analysis. Okada is circumspect about his sources, but he will divulge that DeltaGraphs’ data is collected by stringers — that is, human observers who record information either in person or from video. He says DeltaGraphs’ site’s database goes back to 2012, although the site only displays stats back to 2014. If you’ve used FanGraphs, DeltaGraphs’ layout and assortment of offerings — including WAR, UZR, plate-discipline, and pitch-arsenal stats, and so on — would be extremely familiar. (David Appelman, the founder of FanGraphs, doesn’t mind that DeltaGraphs borrowed FanGraphs’ look, although he hopes to add NPB stats to his own site.)
The main difference is that FanGraphs is free and DeltaGraphs requires a subscription to access all of its data, for a fee of 1,000 yen (roughly $9) per month. However, DeltaGraphs does allow users without subscriptions to see stats of qualified players. "Ideally, I want to make all data free to everyone," says Okada, who adds that he considers it his duty to make more sabermetric converts. "Before DeltaGraphs, there weren’t many stat-based websites [in Japan] in the first place. … Unlike [companies] in the states, we haven’t been able to make enough money from data-tracking to pay the stringers and contributors. I don’t think ad-based revenue works in this market, at least not yet."
As Okada had hoped, DeltaGraphs’ sophisticated work has helped spur a gradual increase in the number of people producing and consuming sabermetric content in Japan. Okada estimates that between the Tokyo chapter of the Society for American Baseball Research, the approximately 15 analysts contributing to DeltaGraphs (many of whom have day jobs and conduct baseball research as a hobby), and independent writers and researchers, there are about 100 sabermetric analysts and researchers in the country. Some of them have started their own statistical resources, such as Nul Data, a site operated by a single hobbyist who publishes player game logs and other useful references. "Basically, we replicate the research done by American analysts and see if the same trend comes up or there’s a different outcome," Okada says. "Since we are behind stateside analysis, we have to do it that way."
Early in his career, Okada envisioned his work as a springboard to an NPB front office, which would in turn confer the credibility needed to sway sabermetric doubters. "When I was struggling to spread sabermetrics, I thought it would be easier to do so if I was involved with a team," he says. To that point, though, his prior interactions with NPB teams had gone poorly. "When I was set to have a meeting with an NPB club, their baseball operations person, who was a former player, yelled at me," Okada says. "He said, ‘What are you doing here? You didn’t play at the professional level. Go home!’ On another occasion, I presented a team visualized data of their position-by-position [depth] chart compared to the other teams in the league. They said they didn’t need me to point it out. All they wanted to know was strategic aspects, namely individual hitter-vs.-pitcher stats. The team did nothing on roster construction and kept struggling."
That began to change soon after he founded DeltaGraphs. The year after he started the site, he heard from Akihito Sasaki, the director of baseball strategy and analytics for the Pacific League’s Tohoku Rakuten Golden Eagles, who offered him the chance to consult for the club. Sasaki tells me via email that Okada, whom he calls "the best sabermetrics guy in Japan," inspired his interest in advanced stats. "In 2012 he helped me to create a new department, team strategy and analytics, for our club," Sasaki says.
Okada remembers this as a thrilling, albeit intimidating time. "When we founded Rakuten’s analytics department in 2012, the only people involved were Sasaki-san, myself, and another newly hired employee," he says. "Initially, no one in the organization understood what we were doing and was skeptical. No one listened to us." Okada says the fledgling crew benefited from forward-thinking Rakuten founder and CEO Hiroshi Mikitani’s hiring of a new team president, Yozo Tachibana, a former financial executive who wasn’t familiar with baseball or beholden to traditional methods. The new exec welcomed Okada’s input. "During the 2013 season, I had meetings with the president [and the] field manager, along with other executives," Okada says. "I believe I was the first non-former-player analyst in NPB history to have such opportunities. During the meetings, it became clear that the manager agreed with my thoughts on the team’s weakness. The team climbed up the standings as he fixed the problem."
Okada doesn’t specify what that weakness was, but with his help, that Rakuten team — which featured Masahiro Tanaka, Takashi Saito, and Kazuo Matsui, as well as American exports Andruw Jones, Brandon Duckworth, and Casey McGehee — went 82–59 and won its first Pacific League pennant and Japan Series championship. Just like the James-aided Red Sox in 2004, Rakuten had broken through in its second season after hiring its country’s most prominent sabermetric pioneer. That success made sabermetrics a core component of Rakuten’s front office. "Luck played a huge part of the run in 2013, [but] I think analysis helped them too," Okada says.
Okada says many NPB teams still make major mistakes that most MLB teams have moved past, such as evaluating batters based on batting average, home runs, RBI, and performance with runners in scoring position, and pitchers based on ERA; embracing excessive small ball; batting powerless slap-hitters second; failing to accurately evaluate defensive value; investing too much in relievers relative to starters and in veteran free agents who are aging out of their primes; and building around pitching and defense, regardless of circumstance.
Consulting for Rakuten and other clubs has given Okada more insight into teams’ reluctance to adopt sabermetric strategies, which has helped him tailor his communication methods. Because of this experience, he says, "I could come up with new ways to preach sabermetrics in the Japanese industry. I couldn’t have done these things as an outsider." His success as a consultant has also changed his mind about working full-time for a team. "Working for one team would narrow the ways to spread sabermetrics to the industry," he says. "I thought it would be better to help multiple teams as DeltaGraphs. Our goal is to make DeltaGraphs all 12 teams’ go-to analytical source."
Shingo Murata, the manager of Rakuten’s strategic planning group, says via email that he believes that even now, only a minority of NPB teams use DeltaGraphs. He does note that seven of the 12 NPB teams have TrackMan systems (the radar component of MLB’s Statcast system, which records pitched and batted balls) installed. But that doesn’t mean that they know what to do with the data, in part because of the difficulty of finding analytical and computer programming talents in Japan who are also well-versed in baseball. Programming hasn’t been as widely accepted as a path to front offices as it has in the states, and teams are often staffed with employees of their parent companies in temporary positions.
"NPB as a whole is still quite a bit behind MLB in terms of having analytical/technical staff within the organization to do analysis," Murata says. "A team often has a hybrid of analyst/video coordinator/some form of scout. Our strategy office has nine full-time staff including four advance scouts, and two more that also hold other responsibilities. I’d say we have 3.5 analysts, which is more than any other team, and I believe we’re the only team that has an independent, standalone database dedicated solely for analysis (other teams usually rely on third-party companies for providing analytical tools). We also have former players who help gather input from players/coaches, and also distill the analytical output to them."
Murata says that his group, which owes its trusted status to Okada (who continues to advise Rakuten), offers data-driven input in four main areas: outfield positioning; evaluating player performance based on batted-ball speed and launch angle; determining which relievers can and can’t handle outings on consecutive days; and, yes, calculating the costs and benefits of sacrifice bunts. Perhaps not surprisingly, the Golden Eagles have laid down the fewest sac bunts in the Pacific League this season.
Thanks in part to Okada’s efforts, change is arriving incrementally. Jason Coskrey, who covers baseball for The Japan Times, says via direct message that he’s seen the NPB’s audience become more statistically literate during his 10 years in Japan. "The All-Star broadcasts [this season] were at stadiums with TrackMan, so it was heavy on spin rates and batted-ball exit velocity and angles, and [the broadcasters] talked about it every chance they got during Game 1," he says. "Overall, you see a lot more sabermetric stuff than when I got here, which was basically none. A lot of the pre- and postseason baseball guides feature sabermetrics as well, so it’s definitely growing, just not at the heights it is in North America right now."
Adam Guttridge, a former Brewers analyst who now licenses his NEIFI player-evaluation system to teams, says that comparing Okada to any one person — even James — undersells how much he’s accomplished. Because he’s built his own network of stringers, played a leading role in promulgating and popularizing the data they collect, and remained at the forefront of Japan’s sabermetric movement, Guttridge suggests that Okada is really like John Dewan (the founder of STATS and Baseball Info Solutions), Sean Forman (the creator of Baseball-Reference), and Tom Tango (an author and consultant turned senior data architect for MLB Advanced Media) all rolled into one.
Granted, Okada benefited from the work of the North American quants who preceded him. But Guttridge, who has met extensively with both Okada and NPB teams in the process of pitching his own system, appreciates how high a hurdle Okada has had to clear. Guttridge suggests (and Yamazaki confirms) that because of cultural norms, Okada, in speaking with teams, "would have to be very polite about presenting material which suggests one of the outfielders on the bench should be playing more than the starter, because that would imply somebody who outranks him is incorrect about something." According to Guttridge, NPB teams still eschew infield shifts not because they aren’t aware that the tactic exists, but because shifting would imply that previous managers who didn’t shift were wrong not to have done so. In that environment, Guttridge says, Okada "[had] to not only create a product, but formulate an entire market."
Coskrey says that compared to the U.S., there’s less outright dismissal in Japan of numbers as a nerdy pursuit. But like Guttridge, he also believes that the country’s emphasis on senpai and kōhai, which stresses the importance of seniority, makes the resistance even more difficult to overcome. The result, Coskrey says, is that "if you have younger people who are running this data and coming up with ideas, maybe they’re not really pushing too hard to be heard. … Japan is still traditional in many sectors, baseball being one, so the new ways are also pushing back against a strong current of ‘This is the way we’ve always done things.’" Murata agrees that "[The] baseball world is a rather closed and very hierarchical (based on age) in Japan, and it is somewhat shocking that someone like myself or other analysts can have casual conversations about data [with coaches or managers]."
Okada, however, is hopeful. It’s hard not to be, given the ascendance of analytics in U.S. sports, the progress he’s already made, and the advantages awaiting any NPB team bold enough to overcome cultural inertia. "Many more advanced stats and studies are public, [and] the younger people try to understand sabermetrics more than their elder counterparts," he says. "More people [have shown] interest in sabermetrics lately. Additionally, those who used to reject me have started to contact me. As someone who’s dedicated the majority of his career to preaching sabermetrics in Japan, I can say things have changed in the right direction over the years." Sacrifice bunts may be much farther from the endangered list in Japan than they are in the majors, but Okada is coming for them.
