The One-Score Season: Favorites Are Still Winning, Just Not by Enough

Through five weeks of the 2026 season, 46 of 79 games have been decided by eight points or fewer. That is 58.2%: more than half the league's games have ended with the losing team one possession from at least a tie.

It feels like a different kind of season, and in one specific way it is. But it is not the way most people would guess, and the part that is genuinely unusual is the part history says is least likely to last.

Close games are the new normal, not the new thing

Start with the headline number in context. Here is how the first five weeks of every season since 2002 compare:

2026 is well above the long-run average. It is also only the fourth-highest start in 25 seasons, and the three above it are all recent:

Widen the lens to full seasons and the shift is plain:

The league has been getting closer for several years. The second column matters as much as the first: the market saw it coming. The average point spread has shrunk by half a point as games tightened. Over those four close seasons, underdogs matched the spread 49.4% of the time, which is a coin flip. A league full of close games did not, on its own, make underdogs outperform expectations, because the expectations had already moved.

So "lots of close games" is not the 2026 story. Something more specific is.

Where 2026 is different: the games that were not supposed to be close

Split the season by the size of the point spread and the picture changes completely.

In games the market expected to be tight, 2026 has actually been slightly less close than usual. All of the surplus, and then some, comes from games with a favorite of seven or more points. Historically, about four in ten of those finish within one score. This season it is nearly eight in ten.

Here is what that has done to the favorites:

The big favorites are winning. They are simply not winning by enough. Three of those games landed exactly on a seven-point margin with a seven-point spread: Lions over the Jets in Week 3, Seahawks over the Chargers in Week 4, Bengals over the Dolphins in Week 5.

The same shape holds across the whole league. Favorites have won 62.7% of 2026 games straight up through Week 5, against 63.7% for the same stretch across 2002-2025: essentially normal. Against the spread, favorites are 33-40-6 (45.6%). The market has picked the right winners about as often as it always does; it has overestimated the margins.

Why a close game is almost always an underdog cover

This is the link between the two numbers, and it is close to arithmetic.

If a game ends within one score, a favorite of nine or more cannot have covered: the margin is too small. A favorite of three to six-and-a-half needs to win by just enough inside a narrow window. Across every regular-season game since 2002:

Whether a game is close decides most of what happens against the spread. So the question that matters for the rest of 2026 is not whether underdogs are "hot." It is whether the big favorites will keep playing close games.

Close games, not close to the spread

A close game is not the same thing as a game that landed close to the spread. One measures the distance between the two teams; the other measures the distance between the result and the market's number. With a seven-point favorite that wins by one, the game is close and the spread is missed by six.

Because the typical spread is around five points, most close games still finish well away from it:

Over the last five full seasons, 2021 through 2025, 76.6% of all games (1,091 of 1,424, playoffs included) finished four or more points from the spread.

Two things follow.

• The closer era did pull results slightly toward the number. The share of games missing the spread by four or more fell from about 79% to about 76% as games tightened and spreads shrank. That is the market keeping pace, as the era table above showed.
• 2026's misses are normal in size, not in direction. At 77.2%, results this season are landing a perfectly ordinary distance from the spread. What is unusual is which way they land: favorites are 33-40-6 against the spread, and favorites of seven or more are 4-12-3. The market is not less accurate this year. It has been off in the same direction all season, and a pattern like that is exactly what the history below says does not last.

What 24 seasons say happens next

We found every season since 2002 that opened like this one and measured what happened from Week 6 onward.

Seasons that started with a high one-score rate

Seven seasons opened Weeks 1-5 at 55% or higher: 2012, 2016, 2018, 2019, 2022, 2024 and 2025. Together they line up almost exactly with 2026:

The early numbers are a near-perfect match for 2026, which sits at 58.2% one-score and 54.4% for underdogs against the spread.

Two things happened next, and they point in different directions:

• Some of the closeness stuck. Those seasons stayed about three points closer than the rest from Week 6 on (52.7% against 49.6%). Across all 24 seasons, a season's early one-score rate correlates with its later one-score rate at r = 0.46. Part of a close start reflects something real about that season, such as the era, rule changes or parity.
• The underdog surplus did not. From Week 6 on, underdogs in those seasons matched the spread 50.1% of the time, the same as every other season. Across all 24 seasons, the correlation between a season's early and later underdog rates is r = −0.23: slightly negative, which is to say none at all.

Seasons where big favorites started close

The more specific 2026 pattern, big favorites playing close, has happened before, just not often. Five seasons opened with at least 55% of their seven-plus spreads finishing within one score: 2003, 2012, 2015, 2022 and 2024.

From Week 6 on, those seasons' big favorites went right back to being big favorites. They finished within one score at nearly the normal rate and matched the spread at nearly the normal rate. Here the early-to-later correlation across 24 seasons is r = 0.18 for closeness and r = −0.13 for the favorites' record against the spread: close to nothing in both cases.

2024 is the cleanest example. Seven of its eight early big-favorite games finished within one score, and those favorites went 1-6-1 against the spread. From Week 6 on, the same category finished within one score 48% of the time and the favorites matched the spread at 51%.

So what is 2026 actually telling us?

Three things, in order of confidence.

1. The NFL is a closer league than it used to be, and the spread knows it. One-score games have run around 55% for four straight seasons and point spreads have shrunk to match. This is the durable part, and it is already priced.

2. 2026's real anomaly is big favorites playing close. Fifteen of 19 games with a favorite of seven or more finished within one score, against a historical 42%. The favorites are still winning those games; they are winning them by a field goal or a touchdown instead of two scores.

3. History says that anomaly fades. Every season that opened this way saw its big favorites return to roughly normal from Week 6 on. Not all of the closeness reverts, but the gap between what the market expected and what happened does. This is the same lesson as our look at two-week starts and five-game form: an early-season pattern is real as a description and close to worthless as a forecast, because the market reads it too.

Week 6 opens Thursday with five games carrying a favorite of seven or more as of Wednesday. If the last 24 seasons are a guide, those games will look more like 2002-2025 than like September 2026.

The honest caveats

• Seventy-nine games is a small sample. The 15-of-19 big-favorite figure in particular rests on 19 games. It is a real description of what happened and nothing more; we are not claiming it is statistically distinguishable from a hot streak, which is in fact the argument.
• "One score" means eight points or fewer, the most a single possession can produce (touchdown plus two-point conversion). Ties count as one-score games.
• Comparison seasons are small too. The "seasons that started like this" groups contain seven and five seasons. The Week 6-onward samples behind them are large (1,293 and 245 games), but the number of seasons is not, and the r values above are each computed across only 24 seasons.
• Closing numbers only. All spreads are closing numbers. Nothing here addresses opening lines or how they moved.
• Pushes count as half in every percentage, the same rule the site uses. The raw win-loss-push records are in the tables.
• Regular season only. Weeks 1-5 are compared with Weeks 1-5; full-season figures exclude the playoffs.
• The 55% thresholds were set before looking at the later results, to find the seasons that resembled 2026. We then re-ran the comparison at every cutoff from 52% to 58%. The close seasons always stayed between two and three and a half points closer from Week 6 on, and underdogs in them always landed between 49.8% and 50.7% against the spread. A different cutoff changes which seasons qualify, but not the result.

See it yourself

• Spreads: how results break down by the size of the spread
• League Trends: league-wide records against the spread, season by season
• Standings: 2026 records against the spread and straight up for every team
• Predictions: our graded record, published in full, wins and losses alike