“Collective fear stimulates herd instinct, and tends to produce ferocity toward those who are not regarded as members of the herd.”
― Bertrand Russell, Unpopular Essays
One of the white lies that we like to tell ourselves is that what happens on the field of play is a fair an accurate determination not only of which team is better but how much better each team actually. It’s the main reason we keep score past winning and losing. Tracking and predicting that margin allows us to provide the best possible predictions of future outcomes and to gauge the values of each individual player.
We can then conclude that merit and production provide as measured in the final score provide us with a fair, unbiased estimator of the worth of each team and each athletes endeavors right?
The truth, as it is with most human undertakings, is more complex and fuzzy that our scientist hearts and minds would like. Human beings are biased creatures and our motivations and ingrained behaviors make themselves know on the field of play.
Where is this coming from? When you have available a database of every NBA game played some things come to your attention that make zero sense in a purely rational world.
That is the average Points scored by a team in each season since the NBA came into being. In it we see massive fluctuations in points driven by the evolution of the game. The introduction of the shot clock in the 1954-55 season for example can be clearly seen as a boom in production. Other rule changes have over time altered the face of the game. At times, the points scored were nearly double what they were in the leagues infancy and 15 to 20% higher than they’ve been in the 21st century.
You would think that the actual margin of victory would line up to that.
Nope. No matter what rule changes the league has implemented, no matter the speed the game is played, no matter the current ongoing boom in three point attempts, the average margin of victory for an NBA game has remained around 11 points.
Even the standard deviation has remained steady and with the increases in shooting efficiency and the the three that makes even less sense.
Think about that for a minute. Given the fluctuations in total points this makes no sense from a rational point of view. Hell, it makes no sense if you simulate it either. Game margins should be way higher if external factors were not influencing behavior.
What exactly is going on? My working theory is that players, referees, coaches and fans all attach a talismanic significance to the double digit lead regardless of it’s actual worth. In the infacy of the NBA game a 10 point lead represented about 13% of the total points scored in a game. Not only that but given the lack of a shot clock, a team could sit on that kind of lead almost indefinitely.
Once the shot clock came into vogue, in the first 5 seasons (1954-5 thru 1958-9) we see 19 games (out of 1044 or 1.7% of the time) where a team comes back to win after being down ten or more points after three quarters.
Now? We have 11 such comebacks in half a season this year and 101 in the past five seasons (out of 5732 games or 1.8%).
When a team reaches that double digit plateau, they ease off. The refs put their whistles away. The opponents take more risks.
This might have made sense the shotclock. Now? It’s pure lunacy. With the move towards more and more 3 point attempts a 10 point lead can be erased in less than two minutes of game time.
It reminds me a bit of the Takeru Kobayashi story. The NBA, like competitive eating, has an artificial bias built around the socially perceived value of a double digit lead.
Someone will come along and shatter that notion eventually.
For the season to date, the average margin of victory for all home teams is 2.35 points per game. If that holds, that would be third smallest in NBA history (only 1948 and 1968 would top it). The data suggests that there will be some regression next season (take all home dogs!) but also that the historical trend is towards a more reduced Homecourt advantage.
I will, of course, continue to keep track.