Basketball is a fairly simple game

Inspired by Dave Berri's post, I wanted to capture his explanation for how basketball works (that is, how games are won).

Teams win basketball games by acquiring possession without the other team scoring, keeping possession of the ball, and turning possessions into points better than the other team. It follows that players win their teams games by doing these, too.

The game is that simple. Its fundamental mechanics are defined by a rulebook that hasn't changed much in decades. Yes, there's nuance, gamesmanship, coaching, human error, heroism, heart, heartbreak, injury, and luck. Dave's point is that all of it still passes through the narrow physics of a possession and he's right!

A player who does an average job contributes 0️⃣ zero to their team's lead (win) or deficit (loss) for a game. Not nothing, just no difference in score; a tie.

A team of average players tallies up zero margin of victory over a season. They would be a .500 team with the same number of wins as losses. ⚖️

A player who does better than average helps their team rise above .500. 📈 A player who does worse causes their team to fall below .500. 📉

Knowing whether a player is adding to leads (and wins) or deficits (and losses) is very hard to see with the naked eye. 🫣

A typical NBA game features about ⏩ 200 possessions between two teams over 48 minutes of play (or 2 hours 14 minutes of real time), but is often decided by a single point or two!

Can we stare close enough at a game to tell that one player grabbed one less rebound, had one more turnover, or missed one more shot than an average player would? 🕵️ Not really.

An NBA season pairs its 30 teams into 82 games each, working out to almost six hours of game play (or 16 hours of clock time) we can watch on average each day over almost six months. Can we watch all of that? Not even close!

Good news: We can watch highlights.

Bad news: Highlights are incomplete and biased.

Good news: The NBA employs scorekeepers to 👀 notice and 🧮 count what every player does for us whether we watch or not. Then they publish game logs and tally events in a "boxscore." This lets us see what our eyes miss, as well as "see" games we didn't lay eyes on at all.

More good news: The scorekeepers count and tally the three kinds of possession outcome for each player! They are:

1. defensive rebounds and steals = acquiring possession without the other team scoring
2. avoiding turnovers and grabbing do overs (offensive rebounds) = keeping possession of the ball
3. shooting percentage from the field and free throw line = turning possessions into points

Bad news: Scorekeepers don't publish average player statistics! They also don't compare player statistics to average. Their boxscores and team stats force another kind of eye test on us where we eyeball which metrics "look good" or "look bad" without a point of reference.

More bad news: It is very hard to tell what is good or bad when each player's amount of playing time and position vary. If a player grabs 10 rebounds in a game, did they help or hurt their team? That depends. If they played 20 minutes at point guard, the answer is probably yes, they helped *a lot*. If they played 40 minutes at center, the answer is probably no, they hurt their team.

We could do a much better job eyeballing how each player is helping their team win or lose if we replaced game totals and game averages with net differences (+2 rebounds, -3 field goals, -2 turnovers, +3 steals, etc.).

Even better, we could do what Dave Berri did and figure out the 🍎s to 🍎s value for each possession outcome scorekeepers count, then turn them directly into player point margin. That way we'd see exactly how much a player helps or hurts. (Good news: That's what Boxscore Geeks is for!)