RealGM

To determine the "impact" of a player apart from the raw box score stats?

For example, the most attractive metric for any player is points per game - the more one player scores, the better they look. But of course, there are many ways to skin a cat - and many many ways to put points on the board. Layups and dunks are better than free throws which are better than corner threes which are better than other threes which are better than midrange shots which are better than long twos, so a player who puts up 20 ppg on midrange jumpers and long twos is less effective than a player who only shoots from beyond the arc or at the line or in the restricted area.

Then there's the factor of pace, which is why stats like AST% can be more important than raw assists per game. Fast paced teams create more possessions, which in turn create more opportunities for scoring, passing, and rebounding.

And of course, there are various statistics that attempt to find the value of a player on defense. Opponent PER, RAPM, defensive rating - take your pick. While none of them are perfect, they each contribute to the valuation of players who while not putting up gaudy box score stats still manage to be in a rotation.

So, what to you is the most important part of advanced stats?

I wouldn't say "apart" from raw stats, but to refine the broad-strokes picture that raw stats creates. In some cases it's clear that the raw stuff creates a glaring systematic bias, but in general, we're just getting into more granular detail.

TwentyOne920 wrote:To determine the "impact" of a player apart from the raw box score stats?

For example, the most attractive metric for any player is points per game - the more one player scores, the better they look. But of course, there are many ways to skin a cat - and many many ways to put points on the board. Layups and dunks are better than free throws which are better than corner threes which are better than other threes which are better than midrange shots which are better than long twos, so a player who puts up 20 ppg on midrange jumpers and long twos is less effective than a player who only shoots from beyond the arc or at the line or in the restricted area.

Then there's the factor of pace, which is why stats like AST% can be more important than raw assists per game. Fast paced teams create more possessions, which in turn create more opportunities for scoring, passing, and rebounding.

And of course, there are various statistics that attempt to find the value of a player on defense. Opponent PER, RAPM, defensive rating - take your pick. While none of them are perfect, they each contribute to the valuation of players who while not putting up gaudy box score stats still manage to be in a rotation.

So, what to you is the most important part of advanced stats?

One of the best uses of advance stats is to determine what a player is good at statistically and to use that sort of information to see if it helps or hurts his team (and by what amount).

For instance, you've implied that because mid-range shots are less efficient, they are less valuable. But that might not be the case! If a player shoots X standard deviations better than the league average, and his team does better with him than without him, then we can state that the midrange shooting is not necessarily detrimental.

Then we can ask ourselves "Why isn't this bad for a team?" Maybe the mid-range shots come near the end of the shot clock and go in at a high enough percentage that they are more valuable than 3s shot at the same time. Maybe those mid-range shots force defenders defenders away from the perimeter, allowing for more assisted 3s. Etc.

Through this process, we learn more about the game. It allows us to determine which assumptions are wrong and which are right, and furthermore it creates new questions that can lead to further refinements of our understanding of the game.