We read all 16 NFL preseason games at once: turnover margin picked the winner in just 7 — and 4 won it backward

We read every box score from this week's NFL preseason slate — all 16 games, so you don't have to scroll 16 game pages — and one column behaved nothing like its reputation. Between 13 and 16 August 2026, all 32 NFL teams played exactly once: 16 games, every one now final, with one tie (Indianapolis 13, New England 13). That makes a complete population, not a sample — every club, one game each, the whole round. Turnover margin is the number football reaches for first. Give the ball away fewer times than the other side, the thinking goes, and you have all but won ^mmodeled. Across the 15 decided games this week, the team that committed fewer turnovers won 7. In 4 games the turnover counts were even (Pittsburgh–Green Bay, Arizona–Las Vegas, Tennessee–San Francisco, Los Angeles Rams–Kansas City), and in 4 more the winner lost the turnover battle outright: the Los Angeles Chargers beat Houston 27-7 while giving it away twice to Houston's once; Jacksonville beat New Orleans 24-20 on 2 turnovers to 1; Baltimore beat Philadelphia 24-7 despite 2 giveaways to Philadelphia's 0; and Dallas beat Seattle 17-7 with the game's only turnover. Read the other way: of the 11 games in which one side gave the ball away more than the other, the team that protected it better won 7. Compare that with the plainer columns. The team with more first downs won 13 of the 15 decided games; the team that ran more offensive plays won 12; the team with more total yards, a higher yards-per-play, and a longer time of possession each won 11. Turnover margin — 7 — sat alone at the bottom. Every one of those turnover figures reconciles: we re-derived each team's total from fumbles lost plus interceptions thrown, and all 16 games balanced (Indianapolis's four, for instance, is 2 fumbles lost and 2 interceptions; New England's is zero of each). The whole slate produced 33 turnovers across 32 team-performances, roughly 2 a game. The tie is the cleanest illustration. Indianapolis out-gained New England 355 yards to 275, made 21 first downs to 16, and averaged 5.6 yards a play to 4.0 — and turned the ball over four times to New England's none. It was enough to tie, not win: Spencer Shrader kicked a 46-yard field goal early in the fourth quarter and a 61-yarder with 1:16 left to reach 13-13, and neither team scored again. The team that lost the ball four times controlled the game by each volume measure here — yards, first downs, yards per play; the team that never lost it could not win. The turnover column pointed one way and the scoreboard sat still.

Strip the giveaways out of the ledger and the winner of this slate was, more often than not, simply the team that moved the chains [analysis]. First downs matched the result in 13 of 15 decided games, offensive plays in 12; turnover margin, the column that carries the most weight when games count, matched it in 7 — the loosest fit on the sheet. That inverts the usual hierarchy, in which a takeaway is worth more than a first down because it is worth a possession ^mmodeled. A likely explanation is preseason itself, though we did not measure it ^sspeculative. Starters typically play a series or two, and many second-half turnovers plausibly come from roster-fringe players and quarterbacks who will not make the team, on drives that change no standing — but we did not analyze the drive- or play-level records that would confirm who committed a giveaway or when ^sspeculative. If a giveaway costs a possession, an August possession may be cheaper — benches empty, no one is chasing the win — yet that too is inference, not something these data isolate ^sspeculative. Two of the four winners who gave the ball away more still won by 17 and 20, alongside dominant volume: Baltimore out-gained Philadelphia 440 to 159 and made 31 first downs to 6, its two turnovers a small figure beside that margin. These are co-occurrences in the box score, not a demonstrated causal trade — the data do not isolate what the turnovers cost. What the complete read surfaces is a column coming loose from the result [analysis]. In the games designed not to matter, the single stat that matters most when they do — the one a broadcast cuts to after a fumble as if the game just turned — was, of the five columns we tested across the 15 decided games, the least attached to who won [analysis]. The volume columns, the ones that measure who spent the day with the ball and made it count, held. We are not claiming first downs "predict" better than yards here; at 15 games those columns (13, 12, 11, 11, 11) cluster too tightly to rank against one another, and a game or two would reshuffle them ^mmodeled. The finding that survives its own small sample is narrower and firmer: across the whole round, the turnover battle picked the winner in fewer than half the games, and four teams won it flat backwards.

The honest ceiling here is the sample and the setting ^sspeculative. Fifteen decided games is a small number, and the gaps between the volume columns — 13, 12, 11 — are well inside the noise; anyone reading a stable hierarchy into "first downs beat yards" is reading too hard, and we do not ^mmodeled. The turnover finding is sturdier only because it stands further out (7 against 11-13) and because four teams overturning the margin is a fact, not a rate. Preseason is also its own country: these are not the lineups that will play in September, and none of this should be carried into the regular season, where the takeaway's leverage is real and, by long convention, large ^mmodeled. We did not pull a regular-season baseline this run, so the contrast we draw is against turnover margin's reputation, not against a figure we re-derived today ^sspeculative. We did not analyze the drive-level records that would say whether a given giveaway came before or after a game was effectively decided, which is exactly the distinction that would explain why the column came loose ^sspeculative. What is not in doubt is the closed count: 32 teams, 16 final games, one tie, and among the 15 that produced a winner the turnover battle was won by the winner 7 times — every figure re-derived from fumbles and interceptions, every game balanced.