Last week I wrote an NBA-related article about the league’s shift toward conference parity and, at the end of the piece, I speculated that the regression of the Western Conference was one of the reasons the Golden State Warriors were able to achieve a record-setting 73 wins.
A friend, discussing the article with me, argued that the primary reason (apart from their talent) that Golden State had managed to win 73 games was simple luck: in extremely close matches, the team had enjoyed bizarre good fortune all season long, the last-second shot nearly always going in or out as needed for a Warriors victory. I acknowledged that he was surely right but remained vaguely dismissive of the point. Any all-time sports achievement works that way, I told him: you can’t break any record without having most of the breaks go your way.
Still, I became curious as to how much, exactly, the Warriors had benefited from their good luck. Despite clichés about the importance of clutch performance, it’s widely accepted among basketball analysts that close games all pretty much amount to coin-flips. If the Warriors and the 76ers are tied in the last two minutes, the 76ers are as likely to win as the Warriors, no matter if Philadelphia is a 10-win team and Golden State a 73-win team. In that sense, these games aren’t meaningful: what defines a team are the games won or lost by larger margins.
So I had an idea: what if we attempted to negate the “luck factor” by neutralizing all the games decided by five points or fewer (an arbitrary cutoff, but it feels right to me) and then comparing the ’15-’16 Warriors to other historically great NBA teams? This approach, obviously, is imperfect: a five-point victory may not actually reflect a close game if the winning team was ahead by 11 with 25 seconds to go, and one of the opponent’s bench players just happened to hit two garbage-time three-pointers in the closing moments. Similarly, a seven-point defeat may, in reality, have been a toss-up, decided by a few 50/50 moments in the final minutes. To get a really good sense of which games were meaningfully won and which weren’t, you’d have to go through the tapes one by one—but even then, each would amount to a judgment call, and the experiment would probably be worse for it in the end.
As it turns out, my simpler methodology was nevertheless very tedious to apply—mainly because I couldn’t find anyone on the Internet who had already done the work for me and so I had to comb through each team’s schedule, game by game, looking for close box scores, and also because I’m clumsy with numbers and had to check my work several times to make sure it was right. In order to create an adjusted 82-game record for each team measured, I simply counted every close game (decided by five points or fewer, or decided in overtime irrespective of final score), whether won or lost in real life, as half a win and half a loss.
So here it is: the ’15-’16 Warriors played in 18 close games, winning 15 of them and losing three (indeed very lucky at .833), making their adjusted record 67-15 (.817).
The previous regular-season record-holder, the ’95-’96 Chicago Bulls (72 real-life wins), played in 17 close games, winning 13 of them: slightly less lucky than the Warriors (.765). The same formula also gives them 67 “true” wins, evening out the two teams.
With 69 real-life wins apiece, the ’96-’97 Bulls and the ’71-’72 Lakers trail the ’95-’96 Bulls and the ’15-’16 Warriors on the list of the greatest regular seasons. These Bulls were, predictably, less lucky than their counterparts of the previous season, winning 15 of 21 close games (.714), but even after adjusting for fairness, they lag behind at 64.5 “true” wins. The Lakers are a different story: they weren’t nearly as lucky as their peers, winning only 11 of 18 close games (.611). When we adjust for luck, their record ties that of the ’15-’16 Warriors and the ’95-’96 Bulls: 67-15.
Additionally, I used the formula to assess two of the other great teams of recent times: the ’15-’16 Spurs (67 real-life wins) and the ’14-’15 Warriors (again, 67 real-life wins). The adjustment shows that, in fact (or “fact”), the ’15-’16 Spurs were roughly equal to the historic ’15-’16 Warriors. San Antonio played in 15 close games, winning eight and losing seven (0.533); thus, their record stays virtually the same after adjustment (66.5-15.5, .811). The ’14-’15 Warriors, on the other hand, were almost as lucky as their counterparts of ’15-’16, winning 13 of 16 close games (.813), leaving them with an adjusted record of just 62-20 (.756).
I have no idea whether this is really the best way of assessing a team’s dominance—you might want to tweak the formula a bit: try it for margins under three instead of five. Or how about comparing records for various teams while excluding games in which one or more of the team’s starters was injured? Which team in NBA history had the potential to be the best without the bad luck of ill health?
Of course, it isn’t actually possible to eliminate luck from the equation: every moment of every game is in some sense a product of luck. Still, I wish I were smart enough to write a computer program that, by extracting the raw data from historical schedules, could instantaneously calculate every single team’s “true” winning percentage—maybe the “best” squad was a team that won only 64 games. What I’ve learned, though, is that most of the all-time great teams were probably about the same: each deserved to win about 67 games, assuming the standard assumption about the meaninglessness of close games is correct.
But who knows? The Warriors have had two disproportionately lucky seasons in a row: maybe they’re winning close games for a reason.