Net points at the team level is a metric that is simple: it's how much you outscore an opponent or they outscore you. At the individual level, though, net points (or Net Pts) involves dividing the credit or blame on every offensive and defensive event.
On made shots, for example, it divides credit across the shooter, the passer, the spacers. More difficult shots get more credit than spacing on the perimeter. More difficult passes that lead to easier shots get relatively more credit.
On missed shots, it divides the blame across the shooter and teammates involved. It also divides credit/blame on rebounds, fouls, and turnovers. On defense, it divides credit on stops. It divides blame when a score is allowed.
Net points adds up over all players to be very close to the game scoring margin or season scoring margin, which ensures a level of accuracy. But the meaning of net points for individuals is essentially the same as for teams, it's how many points a player adds to (or costs) the team's scoring margin. It accounts for shooting, turnovers, rebounding, free throws, individual and defense, passing, and some smaller factors.
It uses play-by-play data to do this, which is a big step up from a box score, but even that data is still missing things like actual matchups, details of screens, nuances of how good a pass was or how close a defender was. Without that, net points makes estimates, estimates that get improved when tracking data is available and incorporated into the algorithm.
Like all methods, it has uncertainty and variability. But it is granular, applying at the game level, the half level, the clutch situation level, the in-conference level, and so forth.
That means that it is helpful to answer questions like these:
It is not currently designed as a predictive metric, though we are working on that version. The current descriptive version forms a basis for talking about the game in greater detail, which should also lead to better predictions.