The 2024-25 season is upon us, and it's time to delve into a key question for basketball data analytics enthusiasts: when can we start analyzing team efficiencies with meaningful results?
It's interesting that it happens so fast. In my work (dealing with American football) it takes better than half of the season (11 games is the generally accepted norm) for stats to crystallise. Due to the NFL's extremely limited sample size, it takes much longer (relative to season length) for everything to come into focus.
I honestly wonder if it'd help a sport like basketball if things took a little bit longer to settle, because most of everything that happens after November kind of feels like waiting, except for maybe a big game here and there that people want to see, but it's also not a culture like baseball has, where individual games never mean anything ever, and everybody knows it, so waiting is the whole point of the sport.
There's obviously no possible way to make this true, but imagine an NBA where we were at the 50 game mark before teams' identities were finally settled. It sounds pretty fun to me.
There are some parameters that stabilize quite quickly, but 3P%, for example, needs a bit more time. I'm not sure what results we would get in the NBA, but I have a feeling it would be something similar.
This is a really helpful analysis! This type of information can help teams know when they should reevaluate their team to know if any structural (roster and/or coaching) changes are needed.
It's interesting that it happens so fast. In my work (dealing with American football) it takes better than half of the season (11 games is the generally accepted norm) for stats to crystallise. Due to the NFL's extremely limited sample size, it takes much longer (relative to season length) for everything to come into focus.
I honestly wonder if it'd help a sport like basketball if things took a little bit longer to settle, because most of everything that happens after November kind of feels like waiting, except for maybe a big game here and there that people want to see, but it's also not a culture like baseball has, where individual games never mean anything ever, and everybody knows it, so waiting is the whole point of the sport.
There's obviously no possible way to make this true, but imagine an NBA where we were at the 50 game mark before teams' identities were finally settled. It sounds pretty fun to me.
There are some parameters that stabilize quite quickly, but 3P%, for example, needs a bit more time. I'm not sure what results we would get in the NBA, but I have a feeling it would be something similar.
This is a really helpful analysis! This type of information can help teams know when they should reevaluate their team to know if any structural (roster and/or coaching) changes are needed.