How to Make Sure the 'Best' Team Wins

Any Dallas Mavericks fans who groaned when the execrable Golden State Warriors knocked their team out of the NBA playoffs this spring, or New York Yankees diehards still shaking their heads that their 97-65 (regular season) team watched as the 83-78 St. Louis Cardinals went home with World Series rings last year, take heart: two physicists have devised a way to make 99 percent sure that the “best” team really does win.

Sure, upsets spice up the game. But let’s get real: the 91-71 Florida Marlins as the 2003 world champions and not, say, the 101-61 Atlanta Braves or Yankees? Over the last 100 years, find Eli Ben-Naim and Nicolas Hengartner of Los Alamos National Laboratory, the lower-ranked baseball team “had an astounding 44% chance of defeating” a higher-ranked team. Fans know that this reflects the fact that even a worse team can have a terrific pitcher, and a better team can suddenly find itself putting on the mound someone who just flew in from the minor leagues to fill a gap in the rotation, or that a few bad bounces can determine a game, among other quirks that make the sport exciting. The result, of course, is that “even after a long series of competitions, the best team does not always finish first,” the scientists write in an upcoming paper in the journal Physical Review E.

Should we ever get fed up with travesties like the 116-46 Seattle Mariners not winning the 2001 World Series, the Los Alamos battery has calculated how we can ensure that the best team emerges as champion at season’s end. The first point is that you need to increase the sample size. Given the randomness inherent in any one match-up, a league with X teams would have to have X times X times X games to guarantee that the best team will end the season and post-season with the most wins. In the 16-team National League, that means 4096 regular-season games; the 14-team American League would have to cram in 2744 games.

Since that’s not exactly practical, the physicists have another solution: add a preliminary round to the season. That would help eliminate the weakest teams before regular league play begins and ensure that a top team is not wiped out of the playoffs through bad bounces, bad luck or just too many off-nights in a row (I’m talking about you, Mavericks). Using numerical simulations, the physicists showed that it would then take on the order of 100 or so games—the current 162 in baseball or 82 in basketball would work just fine—to make 98-percent sure that the champion at the end of the season is among that year’s top two or three teams.

As for football, the 16 games in the regular National Football League season means that the pro football season outcome is even more random than baseball’s.