The Reasons NBA Over-Under Totals Are Much Harder to Predict

In case they end at 114 or less, you lose. An under: you win if it ends at 114 or less. Lose if it ends at 116. ^[1]

NBA bettors typically don’t touch over-unders! 

But why?

This is the piece that explains that.

Pace Variance Changed with Time, But Models Didn’t

For a very long time, wagering on NBA games was easy because it was predictable. Watching one film would tell you how fast a team would play, and from that, one could estimate the possession since it was right there in their face. All good then.

Today, it's different. You still can get a lot of insight from a game’s pace, but the pace variance is different. A team with Russel Westbrook in the game pans out differently if he is not there. A team having a 3-point shooter changes the whole offense's speed. And a defense switch-up in the off-season affects the whole ecosystem.

While oddsmakers set totals assuming these average game speeds, this metric is not a fixed factor anymore in the modern NBA. Now we’ve got ultra-fast teams like the Suns pushing pace hard, and slower, more tactical teams grinding the games out. Pace is no longer linear; it is dynamic, being influenced by matchups. The same team plays 8 possessions per game faster against some opponents than others based purely on defensive scheme fit.

Bookies take this in and adjust with a slightly higher vigorish on totals that spread. However, they can’t perfectly model matchup-specific pace adjustments. This is where I have noticed the real prediction problem lives.

Three-Pointers Creates Unquantifiable Variance

In the early 2000s, basketball teams had an 18-20 shot attempt from the 3-point line and made about 6-7 successful baskets. Today, players take 35-40 shots and 12-14 successful baskets. ^[2]

I have played the game, and to be honest, making a 3-pointer is more volatile. I have missed more shots from this distance than from a 2-pointer distance. Mathematically, in real games, a team shoting 35% from this far instead of 33% might seem like a small difference. However, with over forty attempts, that’s around 2 extra 3-pointers/game. The extras: 6 points.

One NBA team may hit an easy 60% from 3-point range in a match against a weak perimeter defense. Then drop to 20% against length, a few games later. The total swing is based on variance, never real skill difference.

Fouls can Influence Variance

A game starts. All the great players are on the court. But 15 minutes later, one gets a foul. The next 30, another. Now two players remain! The other team can now gain an offensive or defensive advantage. This right here is foul trouble in action, and it dramatically compresses scoring. ^[3] The variance is real, but it is mostly predictable in hindsight.

Again, bookies set the margins higher because they can not see this before a game.

Garbage Time Scores Have Become Unpredictable

Before, garbage time, the final stretch of a game, had little impact on the outcome of a game. The subs or less experienced players would enter the game and just occupy the court until the whistle. The teams now have amazing players, whether they play first or last. And they could score the same or better in the fourth quarter as in the first, affecting the variance. This is yet another factor bookies cannot predict, but have to compensate for.

Bench Depth and Rotations Influence Odds

Some sports are simple to predict. Take football, you can diagram the substitution patterns. Same for baseball, which has clear defensive alignments. The NBA bench depth is a strong indicator for total predictions. And bettors, both casual and sharp, cannot fully account for. In one game, a team with 3 good scorers off the bench may score 8-10 more points than a team with two. It’s just logic.

The only downside is that bench rotations change weekly based on injuries, form, and coach preference. And unless you are a fortune teller or mind reader, you’ll rarely know. Sportsbooks meet this swing variance with price totals a bit higher on the margins.