In basketball and many other sports, the concept of a “comeback effect” is not often discussed, but the effects are often seen when it comes to live betting. The term refers to the phenomenon where a team that is significantly behind in score begins to narrow the gap more than expected from their opponents. The expectation should be relatively obvious. Let’s say the closing line on a game is eight points. We would roughly expect, that team to win every ten minute increment by two points. This trend, on average, holds true for close games. However, if that same team were up, say, twenty points at the end of the first half, we would expect the average final to be around twenty-four points based on our original expectation. With the comeback effect, this isn’t true!
What Causes the Comeback Effect?
While it is hard to nail down the exact effects of this phenomenon, there are three things that I think play the biggest role.
- Adjustments in Strategy: Coaches make tactical adjustments during the game. When one team is up big, these changes will usually benefit the trailing team. The winning team may unexpectedly sit or rest their starters while the losing team will continue to play with their best players in order to not get embarrassed.
- Referee Bias: We sometimes forget that referees are human. Regardless of which side is the home team, the referees will begin to give all the 50/50 calls to the team that is getting crushed. It’s just human nature.
- Psychological Focus: Teams that go up big don’t play as hard. They may start to mess around. This is true everywhere from the NBA to your local YMCA pickup game. Again, to save from embarrassment, the team losing heavily will try harder.
The Projected Comeback Effect
Taking data from over sixty thousand college basketball games has allowed me to give a rough estimate on what the comeback effect is in college basketball. While I have not done this same work on the NBA, one could make the assumption that you could take these numbers and multiply by 1.2x to get the NBA comeback effect. Let me give a quick explainer on what the data below means. This is the added effect of the comeback ON TOP of the expected results based on how good the teams are alone.
As an example, let’s pretend that a eight-point favorite UNC is down by fifteen to NC State at halftime. What would the expectation be for the second half? As an eight-point favorite we expect UNC to make up four points in second half with talent, then per our chart below, another 2.75 points in “comeback effect”. Thus, on average, UNC would be expected to lose the game by 15-4-2.75= 8.25 points.
Interesting Observations
There were a few observations within the data that are worth noting.
- Home/Away Had No Effect: These trends are consistent for both home and away teams.
- Consistent Across Talent Level: The comeback effect stays roughly consistent no matter how big or small the favorite is.
- The Effect Maxes Out: Once the halftime deficit reaches and exceeds twenty, the effect levels off. I could not prove any difference in effect for any values over twenty.