There is a comfortable piece of basketball folk wisdom that says shootouts are close and slugfests are blowouts, that when both teams are scoring, the game stays tense, and when the offense dries up, someone runs away with it. We took all 1,231 games from the 2023-24 season and tested it directly against the final margins. The folk wisdom is wrong, and not by a little. The total points scored in an NBA game tell you almost nothing about how close it will be. The correlation is basically zero.

The claim we wanted to check

The intuition has a logic to it. High-scoring games feel like both offenses are humming, so surely the score stays tight; low-scoring games feel like grinds where one team's cold spell becomes a 15-0 run and a runaway. If that were true, you would expect the combined point total of a game to predict its final margin: more total points, closer game. It is the kind of thing everyone half-believes and nobody checks. So we checked.

The setup is simple: for every game, add the two final scores to get the total, and take the absolute difference to get the margin. Then ask whether the total predicts the margin. Across 1,231 games the average total was 228.4 points and the average margin was 12.6 points. The correlation between them is −0.02. That is not a weak relationship. That is no relationship.

Scatter plot of all 1,231 NBA games in 2023-24, combined final score on the horizontal axis against final margin on the vertical. The cloud of dots is shapeless, spreading margins from 1 to about 60 points at every scoring level, and the best-fit line through it is nearly horizontal, with a correlation of minus 0.02.
All 1,231 games of 2023-24, each placed at its combined score and its final margin. The least-squares line fit through the real games is almost perfectly flat (r = −0.02), and that flatness is the finding: scoring volume carries no information about how close the game will be. Source: bundled data_layer/nba_home_results.csv (2023-24; Basketball-Reference), totals and margins from final scores. Charted by our chart script with a stamped provenance footer.

The exhibit: sort games by scoring, and the margin barely moves

A correlation near zero can still hide a curve, so we split the season into four equal groups by total points (the lowest-scoring quarter of games up to the highest) and looked at the average margin in each. If the folk wisdom held, the margin would slide down as scoring climbed. It doesn't. It just sits there.

Scoring groupGamesAvg total ptsAvg marginClose (≤5)Blowout (≥15)
Lowest quarter310203.711.925%32%
Second quarter332221.613.520%33%
Third quarter297235.312.920%37%
Highest quarter292255.511.929%27%

Source: bundled data_layer/nba_home_results.csv, all 1,231 games of 2023-24. Total = home + away points; margin = absolute point difference. Games split into four equal-size groups by total points; "close" and "blowout" are the share of games in each group decided by 5 or fewer, and 15 or more, respectively.

Look down the average-margin column: 11.9, 13.5, 12.9, 11.9. The lowest-scoring quarter of the season and the highest-scoring quarter produced identical average margins, to the decimal. The 292 highest-scoring games of the year, the 250-plus-point track meets everyone thinks of as nail-biters, were decided by 11.9 points on average, the same as the 203-point rock fights. If anything, the very highest-scoring group had the largest share of close games (29%) and the smallest share of blowouts (27%), a faint tilt in the exact opposite direction of the folk wisdom, but the effect is small enough that we would not stake anything on it.

Why the intuition fails

The reason is almost obvious once you see the numbers, and it comes down to what a margin actually is. The final margin is the difference between two teams' scoring, and the total is the sum. Those two quantities are close to independent. A game can be high-scoring because both teams are efficient and in rhythm (in which case they cancel and the margin stays modest), or high-scoring because one team is torching a defense that has stopped competing, which produces a blowout. Low totals split the same way: a defensive classic where neither team can pull ahead, or a mutual brick-fest where one team's slightly-less-awful night is enough for a 20-point win. Every scoring environment contains both close games and blowouts, in roughly the same proportions.

Pace makes this even cleaner. A lot of the variation in total points is just tempo: a fast team playing a fast team racks up a big combined score without either side being better than the other. Pace inflates the total and leaves the margin untouched, which is one more reason the two don't move together. It is the same lesson that runs through how much of an NBA game is luck: a single game's outcome is dominated by noise, and the scoreboard's total is not a lever on that noise.

The extremes agree

If the relationship were real but faint, the tails would show it: the very highest and lowest totals would part ways. They don't. The 185 games that broke 250 combined points had an average margin of 11.8. The 211 games that stayed under 210 had an average margin of 12.2. Four-tenths of a point of difference across the two extremes of the entire season is, for practical purposes, nothing. Whatever makes an NBA game close, it is not the amount of scoring in it.

Five seasons, and the pace test

One season could be a fluke, and the test above lumps tempo and efficiency together. So we ran it again on every game that counted in the standings from 2021-22 to 2025-26, 6,150 games from ESPN's public results (espn_games_2022_2027.csv, saved November 16, 2026), and then split the 2025-26 totals into their two parts, possessions and points per possession, using both teams' box scores from the league's game log.

Total points against final margin, five regular seasons. The last two columns are the average margin in the lowest-scoring and highest-scoring quarter of each season's games. Source: ESPN results, saved November 16, 2026; our calculation.
SeasonGamesAvg totalAvg marginCorrelation, total and marginMargin, lowest-scoring quarterMargin, highest-scoring quarter
2021-221,230221.212.40.0411.612.9
2022-231,230229.411.20.0011.311.1
2023-241,230228.412.6−0.0212.012.0
2024-251,230227.712.7−0.0112.512.5
2025-261,230231.213.3−0.0113.313.1

Every season says the same thing. The correlation between a game's total and its margin ran between −0.02 and 0.04, and across all 6,150 games it was 0.00. The 2023-24 numbers in this file differ slightly from the ones above because the In-Season Tournament final is left out here. Pooled into four equal groups by total, the five seasons look like this:

All 6,150 games of 2021-22 to 2025-26 in four equal groups by combined score. "Close" is a margin of 5 or fewer; "blowout" is 15 or more.
Scoring groupGamesAvg totalAvg marginCloseBlowout
Lowest quarter1,537202.512.125.1%32.3%
Second quarter1,538220.712.625.0%32.4%
Third quarter1,537233.612.723.4%34.4%
Highest quarter1,538253.612.327.0%31.5%

The highest-scoring quarter of five seasons was decided by 12.3 points on average, the lowest-scoring by 12.1. The 857 games that reached 250 combined points were decided by 12.1; the 1,111 under 210, by 12.0.

The pace test. A game's total is its possessions times how efficiently both teams scored. In 2025-26 (all 1,230 games, possessions estimated from both box scores as FGA + 0.44 × FTA − offensive rebounds + turnovers), about 32.3% of the variation in totals came from tempo and 67.7% from efficiency. Neither part predicts a close game. Combined points per 100 possessions correlated −0.01 with the margin. Holding the number of possessions fixed, the total's correlation with the margin was −0.01.

2025-26 games in four equal groups by tempo (possessions per 48 minutes, so overtime does not count as speed). Source: the league's 2025-26 game log (game_box_2025_26.csv, NBA Stats via nba_api, retrieved June 2026); our calculation.
Tempo groupPossessions per 48Avg totalAvg marginMargin per 100 possessions
Slowest quarter95.9221.212.212.8
Second quarter100.2226.112.512.5
Third quarter103.6235.014.013.5
Fastest quarter108.4242.514.413.2

Tempo does one small thing. The fastest quarter of games finished 2.1 points wider than the slowest (14.4 against 12.2; correlation 0.09). Per 100 possessions most of that gap goes away (13.2 against 12.8): more possessions give the same per-possession edge more room to add up. It runs the opposite way from the folk wisdom, and it is small. The 54 overtime games, which add possessions but start from a tie, were decided by 4.4 on average, which is why the raw possession count shows no link at all (−0.01).

What actually makes a game close

If not scoring volume, then what? Mostly the gap in team quality, and then luck. The better team wins more of the time and by more, as we laid out in how often the better team wins. A mismatch produces a blowout at any tempo, and two evenly matched teams produce a coin flip whether they score 90 apiece or 130. Total points is a property of style; margin is a property of the difference in quality plus single-game variance. Those are different axes, and the data treats them as such. This is also why raw scoring volatility is not the same as team quality, the point we made measuring the most and least consistent teams: how much a score swings and how good a team is are separate questions.

Honest limitations

Total points is a blunt instrument. The first test splits games only by combined score. The pace test above separates tempo from efficiency for 2025-26 and finds no link with closeness in either part, apart from a small widening of margins in faster games that mostly disappears per possession. That test covers one season; the box scores it needs are in our files for 2025-26 only.

Five seasons, one era. The near-zero correlation now holds in each of five seasons (2021-22 to 2025-26), but all five come from the same high-scoring era. The exact blowout and close-game shares are a snapshot of that era, not a constant, and would drift with the scoring environment.

Margin is not drama. A game decided by 12 can be a one-possession game with 90 seconds left that gets salted away at the line, and a game decided by 5 can be a 20-point lead that got cosmetically trimmed in garbage time. Final margin is a crude proxy for how close a game felt, and the "close" and "blowout" buckets inherit that crudeness. Play-by-play win probability would tell a richer story than a final score can.

How much a game scores says almost nothing about how close it is

Sort a whole NBA season from its lowest-scoring games to its highest and the average margin of victory does not budge: 11.9 at the bottom, 11.9 at the top, with a couple of points of wobble in between. The comfortable story that shootouts are tight and grind-it-out games are blowouts is just a story; the final margin is a difference and the total is a sum, and the two live on separate axes. If you want to know whether a game will be close, the amount of scoring in it is one of the least useful things you can know. Ask instead how far apart the two teams are, and then accept that most of the rest is luck.

Sources & Further Reading

  • Game-by-game scores: bundled data_layer/nba_home_results.csv (2023-24, all 1,231 games: the 1,230 regular-season games plus the In-Season Tournament final; Basketball-Reference / public data). Totals, margins, and correlations computed directly from final scores.
  • The five-season test and the pace test (added November 17, 2026): our calculation from espn_games_2022_2027.csv (ESPN's public NBA results, saved November 16, 2026; games that count in the standings) and game_box_2025_26.csv (the league's 2025-26 game log, NBA.com/stats via nba_api, retrieved June 2026).
  • Point differential, margins, and single-game variance: Dean Oliver, Basketball on Paper.
  • Live scores and box scores: Basketball-Reference and NBA.com/stats.

How this was made

Published by NBAAnalytic from public data and the sources named in this piece. How this site is made · the methodology