In a 1000-year-old tradition that focuses on honor such as sumo wrestling, cheating would seem to be an outrageous accusation. Formerly in the sumo world, two men separately came forward with accounts of cheating. Both mysteriously died on the same day in the same hospital with no investigation afterword. Mark Duggan and Steven D. Levitt speculate that the structure of promotion in rankings gives incentives to the sumo wrestlers to rig matches. Using data on every match in a tournament between 1989 and 2000, Duggan and Levitt find that wrestlers who are one win away from moving up in the rankings are more likely to win than usual. The authors discount increased effort as the reason for better performance. The increased amount of wins decreases when a wrestler is in the beginning or last year of his career and media attention is focused on cheating in wrestling.
Each year numerous tournaments occur to decide the rankings. All tournaments have the same structure: 15 bouts in 15 days for each wrestler, an 8-7 record warrants a ranking increase of one place and a 7-8 record decreases a wrestler’s ranking by one spot. With the eighth win worth the most, a wrester going into the final with 7-7 has more to lose than one with 8-6. This non-linear payoff structure generates a situation where conspiracy for two wrestlers is attractive. Use close to 32,000 bouts in 67 tournaments over 11 years, Duggan and Levitt predict a number of truths about the data they analysis.
Incorporating a binomial distribution to find the real probability of a wrestler winning, Duggan and Levitt discover wrestler win disproportionate amount of matches when they are on the “bubble”, or 7-7. On the last day of the tournament, wrestlers with 7-7 win more than 70% of the time, 77% if the authors exclude data on a match with both wrestlers having 7-7 (both need the win, no cheating should occur). This does not happen against wrestlers who have 10 or more wins, as they have a chance of winning entire tournament, which brings more money and prestige. On day 14, wrestlers with 6 or 7 wins win over 60% of the time. If all excess wins are due to conspiracy, 40% of the matches on day 15 and 20% on day 14 are rigged by wrestler on the bubble.
Checking their results with a regression for the bubble wrestlers, Duggan and Levitt encounter similar evidence of match rigging. On day 15, bubble wrestler win 25% more often, day 14 15% more, day 13 11% more and day 11 has 5% more winnings. If all excess winnings were due to riggings, day 15 has almost 50% rigged matches, day 14 had 38% rigged, day 13 has 22% rigged and day 12 has 10% rigged. Further support comes from evidence that wrestlers with 8 wins on day 15 lose more than they should, meaning they collaborate with bubble wrestlers.
Some many argue that those on the bubble might have increased their effort in order to win the crucial 8th win. However, Duggan and Levitt find that 74% of the rematches result with the former loser winning. When the authors split all the bubble matches into whether the bubble person won or lost, the authors find that bubble wrestler who lost do not lose in the rematch as much as those who win the bubble match. It appears that if the match wasn’t rigged for the bubble wrestler, he has no reason to lose to his opponent next time, like those who win have a payment to give. The authors speculate that if this were a case of increased effort, the number of wrestlers losing in the rematch wouldn’t spike downward.
Who is more likely to win on the bubble also substantiates the rigging theory. The number of matches between two wresters has a positive relation to a wrestler’s chance of winning on the bubble against that opponent. It must take time to build trust and learn who is a willing participant. The number of wrestlers who win on the bubble rises through one’s career. Duggan and Levitt even think the stables, or clubs, of the wrestlers play a role in facilitating the corruption. Using the top 10 stables, stables that do well on the bubble lose in the rematch.
The incentive to collude also comes into play. Highly ranked wrestler, or those in the running to win the tournament, do poorly on the bubble because they are unwilling to participate. Those in the last year of their career are 4-5% less likely to win on the bubble because wrestlers view these leaving wrestlers as not useful for the future.
Another fascinating aspect is that of media scrutiny. When the two former wrestlers came forward with tales of cheating and the media began to examine the sumo world, the amount of match rigging decreased.
As I mentioned in the introduction, the idea of cheating in sport that is based in 1000 years of the theme of honor is humorous. Yet, the collusion seems to appear. The theme of cheating to win occurs in many sports events, though usually in a different form: drugs. From the Olympics, to Major League Baseball to the Tour de France, the use of a drug to gain the edge and win pervades media coverage of many events. Though it is a different form of cheating, it brings up similar questions. Cheating: Does it hurt of help the game? Are different forms of cheating different?
Who is really losing when the players cheat? Is it the fans or the players? When players are better via drugs, one could argue that the fans win because the competition is more competitive, faster, stronger, more exciting. Some claim steroids revived baseball because it brought big home run hitters that made the game exciting to watch. Or are the viewers being cheated because the excitement is fake? The players are putting their health in danger to appease the viewers taste for the fantastical, to win the gold metal, or to win the yellow jersey.
Cheating through rigging effects the experience in a different way than drugs. The biggest argument would be that the fans lose because the competitions are not as good as they could possibly be. The outcome is not what is should be. Yet, people love WWF wrestling and the entire competition is a giant show. Everything is fake and people can’t get enough.
While I enjoyed this random topic, how the authors calculated the probabilities of how much some one theoretically should win was confusing. They used a binomial process, but I would think those cannot be 100% correct. Sometimes the Cinderella team or wrestler does win against all odds.