Attempts and dropbacks

What it measures

Passing volume, counted two ways, and the difference between them is where quarterbacks hide.

• Attempts match the official passing line: completions, incompletions (spikes included), and interceptions. Sacks are not attempts. Neither are scrambles.
• Dropbacks count every play where the quarterback dropped back to pass: attempts plus sacks, scrambles, and throwaways.

A quarterback who takes 45 sacks has 45 bad plays that never touch his attempt-based stats. Per-attempt stats quietly forgive them; per-dropback stats do not. Whenever two rate stats disagree about a quarterback, check the denominator first.

How it is computed here

Attempts count completions, incompletions, and interceptions with two-point tries excluded, matching official passing attempts. Dropbacks count plays with the nflverse qb_dropback flag set for that passer, which is also the denominator our EPA per dropback uses. Team and league views carry the same distinction: off_dropbacks at the team grain, and the league_baselines plays column for the all-snaps picture.

How fast it stabilizes

Immediately, in the sense that counting stats do not need to stabilize: they are exact records of usage. The useful signal is in the ratio. Sack rate (the gap between dropbacks and attempts plus throwaways) is sticky for a quarterback across seasons, more so than for offensive lines, which is itself a famous analytics finding.

How it gets misused

• Quoting per-attempt efficiency for sack-prone quarterbacks. Yards per attempt ignores every sack taken; per-dropback numbers are the honest version.
• Using attempts as a proxy for how often a team passes. Scrambles and sacks are pass plays. Pass rate built on dropbacks is the real tendency number.
• Comparing volume across eras or season lengths without normalizing. A 17 game season adds roughly 35 attempts of pure schedule.