Did you miss a session from GamesBeat Summit 2022? All sessions are available to stream now. Watch now.
What’s a sports game developer supposed to do when the sport they simulate stops making sense? This is the question facing 2K Sports after a bizarre offseason in the NBA left the league in chaos. But 2K has to put out their yearly product, and NBA 2K20 came out like clockwork, over a month before the start of the season and with a league it’s supposed to model that may unmodelable. Although when your goal is selling loot boxes, why bother?
Well, it is the only major basketball sim in town, so: What does NBA 2K20 predict for the future?
First, let’s talk about what the chaos means. Since the Golden State Warriors won their first championship of their current run four years ago, the NBA was stable: The Warriors were major favorites, and a small handful of teams served as their competition — usually the Cleveland Cavaliers, the San Antonio Spurs, and the Houston Rockets.
In the 2018-19 season, a few new competitors arose after Cleveland fell apart, and there was a clear set of six clubs at the top of the league: the Warriors and Rockets in the West, and the Philadelphia 76ers, Boston Celtics, Milwaukee Bucks, and Toronto Raptors in the East. The season largely progressed along these predictive lines — the Celtics and Rockets disappointed in the regular season, yes, but by the time the playoffs came around, these teams only lost to one another, with the Toronto Raptors beating the injury-riddled Warriors in the Finals.
NBA 2K19 largely modeled this — using its last patch (with a few injuries added), those six teams were in the top nine, with only the Lost Angeles Lakers, Spurs, and the disappointing Oklahoma Thunder having high team ratings. In other words, the 2K19 season was fairly easy to predict in broad strokes from the outside, and NBA 2K19 mostly did so.
But in the offseason, the league went wild with trades and free agency — the Warriors, the dominant team of the past half-decade, got shattered by injury and free agency. The basic trend of the offseason what that the superteam model — a team with three or four All-Stars and not much depth — gave way to multiple “dynamic duos” with two stars and a strong cast of role-players. Almost every superteam like the Warriors or Celtics got weaker, while almost every pretty good team like the Utah Jazz or the Brooklyn Nets got better, and a few mediocre teams like the Lakers got much better. And several of the consistent contenders took big risks that might or might not work, like the Rockets trading for mercurial guard Russell Westbrook.
It all adds up to this: The NBA has up to a dozen teams that seem like they could have a shot to win the title, and a bunch more who should be quite good. This is especially the case in the Western Conference, where seven teams — the Warriors, Lakers, Clippers, Rockets, Jazz, Nuggets, and Trail Blazers — can all go into the season thinking they should at least make the conference finals (only two teams will). The NBA itself is advertising how “the basketball world has changed” with an ad showcasing five of the most-altered teams to build hype for the season.
In a league that’s traditionally been one where massive, years-long dynasties clash with one another, this is a level of chaos that hasn’t been seen since probably the 1999 season, when Michael Jordan’s retirement and a lockout-shortened season essentially guaranteed chaos.
So how does NBA 2K attempt to model this? Well, in one way, oddly — long-term injuries are one of the biggest stories of the league, in particular with the Kevin Durant, arguably the best player alive right now, almost certain to miss the entire year — but 2K doesn’t actually have those injuries in the game right now (though I assume they’ll show up when the season starts). So I had to manually add six injuries to stars: roughly two months for Victor Oladipo (Indiana), four months for Jusuf Nurkic (Portland), six for Klay Thompson (Golden State), eight for Durant (Brooklyn), 10 for John Wall (Washington), and the season for DeMarcus Cousins (LA Lakers) — estimates all, but pretty essential for the story of the league right now. I also fiddled with these numbers up and down to give different potential outcomes, such as if Thompson comes back midseason or doesn’t come at all. (Note: These simulations were mostly done in mid-September, so new patches may change things.)
I also fiddled with a few players’ positions because NBA 2K20 would consistently put much worse players over good players’ secondary roles. For example, I switched the Sixers’ Josh Richardson to shooting guard, so that the giant five starters Philadelphia expects to run with will load into the starting lineup, instead of unproven rookie Matisse Thybulle.
That done, there’s a major way NBA 2K20 makes an initial prediction: the team rating. Now, this is an imperfect representation of how good a team is, basically a messy average of the game’s complicated set of ratings for individual players. In other words, it might say a team has a 98 rating because of players’ stats, but it won’t necessarily simulate that team as being better than a team with a 90 rating. But it’s still a good quick shorthand.
Which is all a long way of saying that NBA 2K really, really likes the Los Angeles Lakers.
A year ago, the Lakers signed Lebron James, finally arresting their post-Kobe Bryant collapse that had made them hilariously incompetent. NBA 2K19 gave them a team rating of 90 — a point above the Bucks, who had the best record in the league — and they hilariously and not that surprisingly missed the playoffs. It turns out even Lebron needs teammates. So this past offseason, the Lakers traded almost all their promising young players to New Orleans for another superstar, Anthony Davis. 2K loves this, with NBA 2K20 giving the Lakers its highest rating of 97. This is despite that, although James and Davis are indeed superstars, the Lakers traded … almost every other decent player on their roster. In real life, this is an interesting team with a ton of question marks that has championship potential if it can answer those. In 2K’s world, the Lakers are the massive favorites.
Some more strange choices await. The Bucks have an 89 rating, the ninth-best team in the league, same as the prior season … despite that they had the best record in the league and are almost certain to compete at the same level again. 2K still loves the Spurs, a team that’s been stuck at “pretty good” since losing Kawhi Leonard in real life, but rated them as the fifth-best team in the league both last season and this.
Further weirdness awaits deeper down. The Nuggets were the surprise team of last season, finishing second in the West with a young core that should only get better. And somehow, they have a lower rating in 2K20 than in 2K19, with an 84 that puts them right in the middle of the league. They’re just ahead of the Washington Wizards with an 83 rating, despite the Wizards being widely expected to be one of the few teams without any hope this season. (The simulation, however, successfully separates these two.)
Next page: how I tested the simulation, and a summary of results! After that, a team-by-team analysis.