QB Continuity: What 861 Team-Seasons Say About Starting One Quarterback

The Question: Does Starting One QB All Year Matter?

Every November, some team on its third quarterback becomes the argument: "you can't win without stability at the position." It sounds true. It's also exactly the kind of claim that deserves a count rather than a vibe, and the nflverse game file makes the count easy — every game since 1999 records both starting quarterbacks. This page counts 6,967 regular-season games across 861 team-seasons (1999–2025): how many starters each team used, and what happened to their record.

The Exhibit: Win Rate by Number of Starters

Three things stand out. First, the gradient is steep and monotonic through three starters — each additional quarterback costs about a win and a half of season pace. Second, wire-to-wire seasons are not the norm: only 38.8% of team-seasons since 1999 got through a schedule with one starter. The single-quarterback season fans treat as the default is actually the minority outcome. Third, the tiny 4+ bucket (23 seasons of quarterback chaos) actually edges the 3-starter group — a reminder that at small sample sizes the noise takes over, not a reason to believe four starters beats three.

The Arrow Points Both Ways

Before anyone builds a take on that table: the causality is hopelessly entangled, and honest analysis says so. Quarterback changes happen for two big reasons — injury and benching — and both are correlated with losing through paths that have nothing to do with "continuity" as a skill.

• Benchings are a symptom. Teams bench quarterbacks when they're losing. A 2–8 team trying its backup was probably winning at a .200 clip before the switch — the second starter didn't cause the record; the record caused the second starter.
• Injuries remove good players from good teams too — but a contender losing its starter tends to stabilize with a competent backup, while a bad team losing its starter often spirals into audition mode, stacking third and fourth starters onto an already-lost year.
• Good quarterbacks are durable AND good. The players who start 17 straight games skew toward the ones worth building around, so "one starter" partially just measures "has a franchise quarterback," which everyone already knew wins games.

What this data can't do is separate those channels — the game file records who started, not why. Treat the table as descriptive: it tells you what one-starter seasons look like, not what forcing continuity would do for a team that doesn't have it.

The Trend: Continuity Is Getting Rarer Again

Wire-to-wire seasons peaked in the early 2010s — nearly half of all team-seasons — and have since fallen to about a third, the lowest sustained level in the file. The likely suspects are structural: the schedule added a 17th game in 2021 (one more week for something to go wrong), and the era of designed quarterback runs exposes starters to more contact. Whatever the mix, the practical read for fans is that a team getting 17 starts from one quarterback is now roughly a one-in-three outcome — plan your expectations, and your roster-building takes, accordingly.

How to Actually Use This

• Backup quality is a real roster line. With only ~39% of seasons surviving on one starter, the median team WILL play meaningful games with its backup. The .493 win rate of two-starter seasons says those years are far from lost — the drop-off is a cliff only at the third starter.
• Don't credit "stability" for what talent did. When a one-starter team wins 12 games, continuity gets the headline; the quarterback being good is the story.
• Midseason QB changes on bad teams aren't self-harm — they're usually the consequence of a season already lost, and the win% table reflects that selection, not the cost of the switch itself.

Related machinery on this site: how quarterback performance is actually measured, turnover luck (the fastest route to a benching that isn't the quarterback's fault), and one-score game records (the other big symptom-vs-cause trap in season records).

Reproduce It

One file: games.csv from nflverse nfldata, bundled at /data/games.csv. Filter to played regular-season games (game_type == "REG") with non-empty home_qb_id/away_qb_id — 6,967 games, 1999–2025. For each (season, team), count distinct starter ids and wins; group team-seasons by starter count and average the win rates. The era table is the share of team-seasons with exactly one starter per five-season block (the 2020s block includes the partial 2025 season present in the file). Computed 2026-07-12; figures shift slightly as seasons append.

Data: nflverse/nfldata, public. "Starter" means the quarterback recorded as starting the game in the file; relief appearances don't count.