Context is Key: “Stats per Game” is Misleading

By Jack Van Deventer • August 6, 2021

You have the following stats available for your college team.  Which player is your best scorer?

Based on these numbers, you would reasonably conclude that Anthony is your best scorer.  Indeed, it appears that Anthony is twice as effective at scoring as Chris is. 

But what if these players averaged 30, 20, and 10 minutes of playing time each?  Then, it’s a different story.  Since this is a college team and 40 minutes is the length of a regulation college game, I’ll calculate a “PTS per 40” metric:

PTS per 40 = (PTS / MIN) * 40

That is, I calculate the “points per minute” times the number of minutes in a game (32 minutes for high school, 40 for college, and 48 for NBA).  Now, I have more context to work with when evaluating players: 

When playing time is factored in, now Chris is shown to be the most prolific scorer.  That is, he scores at a higher rate than the others.  His rate of scoring is 50% higher than Anthony’s. 

The information advantage you gain as a coach comes from a more precise CONTEXT (time played instead of games played).  We multiply by the game length (e.g., points per 40 minutes) because it’s a familiar time framework.  Again, this adds meaning to the metric, because you already have expectations as to what a reasonable scoring rate in a game should be.    

When it comes to stats at BasketballScience.Net, we provide the standard metrics:  TOTALS and STATS per GAME.  Most importantly, we also include STATS per 40 (or per 48 for NBA teams).  The latter metric is the most accurate for comparing players.  We do this for REB, TOV, AST, BLK, Plus-Minus, and every other metric. 

Of course, other context factors need to be considered:  Is Chris only playing against weak competition at the end of a game?  Is Anthony only playing with the best players on the team?  Is Brian playing against the hardest teams? 

All of this matters.  Context is key.  BasketballScience.Net provides the context.