How I tested 2K20
So, clearly, the limited preseason predictions are rather different from what most experts might say. But who’s to say the experts are more right than the game? That’s one major question. The second is: What happens when you actually simulate the games in 2K?
Actually, simulating the season isn’t enough. Injuries are on, which makes it far too random to do just once. What happens when you simulate 20 seasons in NBA 2K20? (My initial goal was to sim at least 30 seasons, but the differences in each season were so slim that this seemed unnecessary. But it’s a hell of a lot better than one season: Suck it, New York Times!)
Here’s what I did. I simmed two seasons where I adjusted rotations for injury and performed trades myself. I also did two more where I did trades, but let the rotations occur automatically. Then 16 more where the game controlled those details and I just let the chips fall. I did continually adjust the CPU’s trade tendencies until I could get stars like Bradley Beal or Kevin Love actually traded, but this didn’t happen (with one exception) no matter how far I raised it — the biggest name I saw traded consistently, without my input was the Thunder’s Steven Adams. (The exception was one season where the Warriors traded Draymond Green to the Bulls for Zach LaVine. That was weird, but an extreme aberration.)
At the end, I noted each team’s regular season record, which I averaged out. I also noted the highest and lowest win totals to give a good idea of how varied teams could go. (The raw data is available in the spreadsheet.)
Then, in order to calculate playoff success, I used a simple linear measure of giving a point for a team making the playoffs, then another point for each series they won, which seems, subjectively, to be a reasonable representation of how success is perceived. The championship-winning Raptors would get 5 points, while the runner-up Warriors getting 4 points would have had twice as successful of a playoff run as the Rocket team they beat in the second round. With 20 seasons simmed, this made for a really good conceptual set of numbers: 100 would be a team that wins every championship, 20 would be a team that just makes the playoffs every year, while anything above 40 would be a superb team that wins more playoff series than it loses. As a point of comparison, the Popovich Spurs, our real-life multigenerational dynasty, would have had a Playoff Point total of 56 from 1999-2018 — that’s five championships and most years as a contender.
Also worth noting: This method would be biased toward Eastern teams, as thanks to the NBA’s unbalanced schedule, they’d play each other more often in the regular season and only each other in the playoffs until the Finals. By every subjective estimate, the East is somewhat weaker both in terms of number of championship contenders (the West has the seven mentioned above with a few maybes, while the East has two definite contenders and a bunch of maybes) and in top-to-bottom strength — four of the five teams expected to definitely just be bad, the Hornets, Knicks, Cavaliers, and Wizards — are in the East (and the Grizzlies in the West).
This pattern continues in NBA 2K20, with the West being generally stronger and occasionally, hilariously so — the best of these being one simmed season when the Spurs missed the playoffs with 48 wins in the West, a total which would have have had them in third, one game out of second place, in the East that season. This is just the way the NBA works and has worked for two decades now, and only advanced statistical analysis beyond the scope of this article could begin to adjust for it.
Results
You can see all 30 teams ranked, with their wins and playoff totals, as well as analysis on the next page. The Google Doc with the raw data is also here. But here are the biggest conclusions:
- For better or for worse, 2K didn’t adjust its simulation after an offseason left the league without a clear hierarchy. There’s a pecking order in the game’s simulation.
- The Lakers are at the top of that list — the game treats them as dominant force to the level that the question is “can they lose?” — much more on this on the next page.
- Three other teams are contenders who win championships and make multiple Finals: the 76ers, Clippers, and Rockets. The Sixers are the only Eastern team to win it all, although the Bucks and the Nets (with a recovered KD) also make the Finals. The Blazers also win once, but that seems to be a (welcome!) fluke.
- The six teams above, minus Portland, are the only teams to make the playoffs every year. The Blazers, plus the Warriors, Nuggets, and Celtics all make the playoffs in at least 17 of the 20 simmed seasons.
- Beyond the Lakers, the simulation seems to really like the Thunder and the Bulls a lot more than most oddsmakers would. It also really dislikes the Pacers, Heat, and Jazz.
- Six teams fail to make the playoffs in any of the 20 simmed seasons: the Kings, Mavericks, Wizards, Grizzlies, Hornets, and Cavaliers. The latter four are not a surprise — they’re probably going to be the worst teams in the league. But the Kings and Mavs, however, are — these are exciting young teams for which it wouldn’t be a surprise at all to see take the next step.
- Yes, Knicks fans, this means 2K thinks you can make the playoffs (once, with 38 wins). The Suns and Hawks surprisingly sneak in a couple times.
- I did not see huge differences in games where I controlled rotations and/or trades versus pure simulations — that Blazers championship came without me trading for Kevin Love or the like. The one exception to this is the Thunder improving notably when I started both Roberson and Gilgeous-Alexander — something that continued when I adjusted Roberson’s position so the game would automatically slot him in at small forward.
- While most of the decisions made within the game are defensible individually, they add up to a much more predictable simulation than NBA observers or oddsmakers might expect. Seeing the Lakers and Sixers in the Finals almost every time may test players’ patience, depending on how they play and how the real season plays out — although both may lead to patches from 2K.
On the next page, the first set of team-by-team results, plus an experiment