With the playoffs kicking off this weekend and every basketball-related writer on Earth turning into a Magic 8 Ball, I couldn’t help but think of the vagaries of basketball and how they impact predictive analysis. It’s commonly believed that the better team wins in a 7-game series. Pick a series incorrectly and you’ve falsely identified the weaker team.
That’s not really true, though.
Ask yourself how often the best team would win in a single-possession series. Clearly, whatever number you come up with, it isn’t 100%. After all, the Bobcats did win seven times this year and the winning team doesn’t always score first. This is because it takes more than one possession to determine who is better in basketball.
It also takes more than one quarter to determine who is better. And it takes more than 48 minutes. And more than seven games. Upsets happen all season long, it doesn’t mean the winner is suddenly a superior team. Instead, they occur because basketball is a high variance sport. Even the worst teams are able to score 100 points on a given night. Even the best teams have their shooting percentages waver. The difference between a good offense (1.10 points per possession) and a bad offense (1.00 points per possession) is an extra basket every 20 possessions.
In 2012, the average margin of victory (MOV) in the NBA was 11.1 points per game. Process that for a second…the average NBA game was an 11-point game, but average MOV of all the teams was 0. Because of this variability, the average game isn’t that close to the overall average performance. Basically, if you stop a game after 48 minutes, it’s not highly likely that the best team won. It’s no where close to a statistically significant threshold used in science (95% certain at the least).
Using Home/Away MOV to Examine Upset Frequency
One way we can observe this is by comparing all of the weaker teams who defeated superior teams throughout a season. Instead of simply looking at the difference between two teams in terms of Margin of Victory (MOV), let’s include where the game was played; Some teams are much better at home than they are on the road. We can compare the difference between two teams in a given game by looking at their home MOV for the season versus the opponent’s away MOV.
For instance, on the second day of the season, Denver traveled to Dallas. The Nuggets were a +1.9 MOV team on the road this year. The Mavericks were a +4.6 MOV team at home this season. The difference between the two teams was 2.6 points (in favor of Dallas) – we can call the Nuggets a 2.6 point underdog, and say they won in an “upset” (by 23 points). It’s not a perfect indicator of the strength difference between two teams, but it’s a good ballpark measurement and we’ll use it to make our point...
Get the rest of this article over at Back Picks.