Trying to Draw Big Conclusions From the Early NBA Season is Difficult

If the Portland Trail Blazers do something statistically interesting, my job is to write about it. After Week 1 of the 2018-19 NBA season, I looked at Portland’s numbers and found some interesting stuff, but honestly, my heart wasn’t really in it. Most NBA teams have played four games. A few have played five, and some have still only played three (the Trail Blazers obviously among them). It’s just really tough taking much away from such a small sample size, especially when you look from a statistical angle. Variance rules sample sizes that aren’t large enough. Therefore, instead of trying to take some things away that probably won’t hold true, here’s a few reasons why early-season stats can be misleading.

1. Shooting is important, and it’s inherently prone to variance

One of the most fundamental truisms of basketball is that you need to score points to win games. In today’s NBA, a lot of scoring is done with outside shooting. Outside shooting is difficult in NBA situations, even for the best of the best. That means it’s very streaky, and thus prone to variance when looking at small samples. For example, Steph Curry might miss all eight of his threes in a game. He might even go 2-20 over a three-game period. That doesn’t mean he’s not the best shooter of all time, or that he’s suddenly regressed, or that it means anything at all outside of “he had a couple bad shooting games”. Of course, at the start of the season, even a single poor game can have momentous weight on a player or a team’s stats. So not only are most player’s shooting stats (especially if they go against the large sample of the player’s career) to be held in suspicion, so are team’s shooting numbers as a rule. Shooting is a huge part of basketball, and if it’s hard to evaluate players on their shooting after just a few games, it’s therefore pretty difficult to get much statistical evaluation at all.

2. Teams are still figuring out rotations

For most teams, the rotations they use to start the year are different than the ones they use to close the season. The players might be the same, but they’re used in different combinations and played at different times. At the beginning of the season, coaches play it somewhat safe. They run with veterans who have “earned” their spots through play in previous years or preseason, and stick with them until a chance is clearly necessary. Now, it can be clear relatively early on that certain lineups don’t seem to be working well – but that’s not necessarily a referendum on the team or even the players in the lineup. Sometimes a group of players just isn’t a good fit, and once the coach figures out the right combinations, a team can come together. Even still, three to five games is too early to even judge a lineup, due to the shooting variability mentioned above.

3. Lack of scouting

For established players, this isn’t as much a thing. NBA teams and veteran defenders know that certain players like going left, or that others prefer to take shots from the right elbow. But rookies and young players who are still adding a lot to their game can be difficult to plan for without the proper amount of film and reps. Similarly, most teams change their playbooks at least a little from season to season. This has a two-fold effect. First, opposing teams will be surprised by plays, or by plays having different and new twists to them. Second, players might not know all the plays in the book, especially if their offense underwent drastic change over the summer. All this means that there’s a lot of mistakes on both ends early in the season that won’t necessarily remain that way after teams settle in a bit. Once scouting comes out on players and tactics, teams get smarter, and the way the games are played changes a bit. The early season is a testing ground, and that means some teams get away with things that won’t work for them for long.

It’s tough to evaluate basketball at all. It’s even tougher to make large generalizations about teams or players when the sample size of the season is so small. Some things can be gleaned from these games; this is the regular season, so every game matters. Each game that’s added to the books creates a larger data set, and sooner or later real statistical takeaways can be made. But for right now, while a lot of NBA stats are fun, they don’t say anything too substantial. Just last year, the Orlando Magic started out 8-4 (a much larger sample size than what we have now), and all the talk was of how they were a changed team. They promptly lost their next eight games and ended the season deep in the lottery. So, when you see a stat or take that offends you, or even one that excites you, just remember that the season has just begun, and nothing is set in stone.