After an NBA Game Busts the Total: Next-Game Over/Under Rates

A game flies 25 points past the total and the instinct kicks in: the shooting was unsustainable, pace will cool, hit the under next time. I tested that instinct across 21,218 regular-season team-games from 2017-18 to 2025-26, each with a posted total in both the trigger game and the team's next game. The day-to-day calls live on my NBA predictions page. This file is about the pattern underneath them.

The headline answer is that the next game barely notices. After a 15+ point over, the next game went over 51.2% of the time. After a 15+ point under, it went over 49.1%. Against a baseline of 50.4%, that is noise dressed up as a story, and the one place something sharper shows is when both teams walk in off the same kind of miss.

Aftermath · 01Next-game over rates barely move after a big miss

This chart sorts every team-game by how far its previous game missed the total, from under by 25+ to over by 25+. The bars sit in a tight band. After a 25+ point over, the next game went over 50.0% (1,853 decided games, ±2.3), and after a 25+ point under it went over 48.5% (1,597 decided games). The middle bands hover close to the 50.4% baseline.

The contrarian play does not survive the numbers. Backing the under after a 15+ point over returned −6.7% at −110, and backing the over after a 15+ point under returned −6.3%. You pay the margin and get nothing back for spotting regression, because the market had already priced it.

Aftermath · 02Scoring snaps back hard while the line barely flinches

This chart explains why. After a 25+ point over, the next posted total moved +1.1 pts from the last line, but the actual total changed by −32.9 pts. After a 25+ point under, the line moved −1.7 pts and scoring jumped +30.0 pts. Scoring reverts sharply, and the line was never built on last night's outlier.

That looks like the market missing something, but it is the reverse. Totals are priced off pace, efficiency and rotation patterns, not one box score, so the number already sits near where scoring lands on average. The snap back is baked in before tip-off.

My read is that a big miss tells you about variance, not about the next number. Hot three-point nights and foul-fest fourth quarters do not carry over, and the books price them as if they won't.

Aftermath · 03Both teams off the same miss is where it bites

This table looks at the matchup rather than one side. When both teams came off a 15+ point over, the game went over 53.9% (513 games, ±4.3), on an average line of 226.8 against an average total of 228.5. When both came off a 15+ point under, the over hit just 44.4% (497 games), with the average total of 220.9 landing below the 222.7 line.

Read the ranges before getting excited. A ±4.3 range on 513 games leaves room for the true rate to sit nearer a coin flip, and the mixed row, one 15+ over against one 15+ under at 49.1%, shows nothing at all. I treat this as directional: when both sides share a scoring rhythm, the line seems to lag slightly, so I take a harder look at pace and schedule rather than firing automatically.

Aftermath · 04After a 15-point miss, seasons swing both ways

The seasonal lines show how fragile any single-year read is. After a 15+ point over, the next-game over rate ranged from 48.0% in 2022-23 to 54.3% in 2020-21. After a 15+ point under it swung from 44.7% to 53.5%, and the two lines cross more than once.

The main caveat sits here. Each season holds roughly 400-700 games, overtime pads some trigger games and some next games, and everything in this file is a historical frequency rather than a betting system. A hot year for any angle is mostly noise.

Written byMarcus Sullivan

I write NBA previews with a strong focus on pace, matchup advantages, and how the modern game shifts betting markets.

View profile →