There is a fascinating analysis of NBA scores by Jeff Ely and Toomas Hinnosaar on Ely’s blog Cheap Talk. Read through the comments as well which try to explain the data.
The data plotted in the chart is the margin of victory for the home team at the end of regulation (negative if they lose). There is a very pronounced spike at zero, implying that the % of games that are tied and go to OT is much higher than if it was closer to a normal distribution around a mean of zero.
There is a great video if you click through to the post, which shows how the distribution of the lead of the home team changes in the last 40 seconds of the game. Things look pretty normal till about 20 seconds are left. Something happens in the last 20 seconds that makes things converge to a tie.
The comments from readers try to explain this. The explanations that ring true to me are:
1. Shot selection – If the trailing team is within 3, it will attempt a 3 point shot. If it is within 2, it will typically go for a higher percentage shot for 2 points.
2. Fouling to stop the clock – The trailing team will keep fouling to stop the clock in the last 20s which will typically expand the lead. If they luck out and pull even, they stop fouling. If they pull ahead, the other team starts fouling.
Said differently, if the two teams are separated by 3 or fewer points in the final 20s and the trailing team has the ball, they will try to run out the clock and make a final shot attempt to tie the game (or win it if they are trailing by 1). If the leading team has the ball, the other team will foul them in the hope that they will miss a free shot and they get possession of the ball.