When an NFL Player Is Listed “Questionable,” Do They Play—and Do the Odds Keep Up?
We count how often key players listed “Questionable” in 2016–2025 actually played, and check whether the odds the night before already reflected these public injury statuses.
- Result status
- Preliminary results
- Last updated
- Start date
- Duration
- Day 1 of 5
- Research team
- Mysports.AI Research Team
- Plan registered
NFL Questionable Players and Odds: Key Findings
2016–2025 regular seasons: the player played in 6,297 of 8,756 cases (range 70.7%–73.1%). Key player = on the field, on average, for over half of offensive or defensive plays in the last 4 games; played = at least one play. All players: 67%. Similar overall figures have been published elsewhere, so this part is not a blind test.
- The last practice matters most: 48% of Questionable key players played after not practicing, 74% after a limited practice, 89% after a full one. 63% played after missing the previous game, 76% after playing in it.
- Odds (pre-registered main test, supported): for each extra key Questionable player on the home side versus the visitors (net count), the home team’s implied win chance fell about 0.10 points on average from 16 hours to 5 minutes before kickoff (e.g., with 3 more: 60% → 59.7%, odds about 1.67 → 1.68). For 2024, where report timing can be verified, the single-season estimate points the same way with a range excluding zero; 2025 points the same way but is uncertain. This is an association, not cause and effect.
- Starting point: measured from 6 hours out, the estimate is about half; from within 3 hours, it nearly disappears (game counts differ slightly by starting time, so this does not pin down when the move happens). In 80 games with a single Questionable key player, the average odds for “he played” and “he sat” already differed 2 hours before kickoff and did not clearly widen around the official inactive list 90 minutes out (small sample, wide ranges, descriptive only).
In plain words
Why study this?
“Questionable” is the most common and most uncertain status on the NFL injury report. Knowing how often it really means playing, and whether the odds keep up, helps a model estimate lineups and helps readers understand why pre-game odds keep moving.
What do we want to know?
Before each NFL game, teams file an official injury report listing injured players as “Out,” “Doubtful,” or “Questionable.” The league never says what “Questionable” means in numbers. We ask: what share of these players actually play? And did the odds the night before already account for these public statuses?
How do we study it?
Using official injury reports (via nflverse) and play-by-play participation for the 2016–2025 regular seasons, we measure how often key players (on the field, on average, for over half of their team’s offensive or defensive plays in their last 4 games) listed “Questionable” actually played. For odds, we use Polymarket prices for 523 games in 2024–2025, comparing 16 hours before kickoff (9 p.m. Saturday for a 1 p.m. Sunday game) with 5 minutes before. A Polymarket price is a win chance: $0.60 ≈ 60% ≈ odds of about 1.67; lower odds mean more favored. The plan and the analysis code were timestamped before any results were computed.
What have we found?
About 72% of key players listed “Questionable” ended up playing; only 48% of those who missed the last practice did, versus 89% after a full practice, and 63% of those who missed the previous game. “Doubtful” players almost never played (1%). On odds: the more key “Questionable” players the home team had than the visitors on the final report, the more the home team’s implied win chance tended to slip from the night before to kickoff (odds slightly longer, a little less favored)—about 0.1 percentage points per extra net player. Measured from closer to kickoff, the association gets smaller.
How does this help with predictions?
Treat an NFL “Questionable” tag as “about 7 in 10 will play,” not a coin flip; then check the last practice: roughly half play after missing it, about 9 in 10 after a full one. The odds effect is tiny: it only suggests the night-before odds did not fully reflect these statuses, not that there is a reliable edge.
What can’t we say yet?
This is a retrospective study: the data are public and others have published playing rates, so the playing rates are not a blind test. We only have the final injury status, not the time of every report; 2024 timing can be verified and the 2024-only estimate points the same way with a range excluding zero; 2025 cannot be checked row by row and is uncertain on its own. The association shrinks when measured from closer to kickoff, but the game counts differ slightly by starting time, so this does not pin down when the move happens. Polymarket only prices “will the team win,” so we do not judge what the market thinks a player’s chance of playing is. The rest of the 2026 season will be re-tested with the same rules.
Terms used on this page
- Questionable
- An NFL injury report status meaning it is uncertain whether the player will play.
- Doubtful / Out
- “Doubtful” means unlikely to play; “Out” means will not play.
- Key player
- A player who, on average, was on the field for more than half of his team’s offensive or defensive plays in his last 4 games.
- Odds (Polymarket)
- A Polymarket price is a win chance: $0.60 ≈ 60% ≈ odds of about 1.67 (odds = 1 ÷ price). Lower odds = more favored.
- Implied win chance, percentage point
- The win chance implied by the odds; 1 percentage point is, for example, 60% to 59%.
Detailed results
Share of “Questionable” key NFL players who played
2016–2025 regular seasons; participation in the last practice separates players the most.
Pre-registered description | Key player = on the field, on average, for over half of plays in the last 4 games; played = at least one play. Practice status is missing for 96 cases. Similar figures have been published elsewhere; not a blind test. This is research, not betting advice.
HypothesesFor deeper reading
- H1The home-minus-away number of key “Questionable” players on the final report is associated with how the home team’s odds move from 16 hours to 5 minutes before kickoff → supported: each extra net player goes with about 0.10 percentage points less implied win chance on average (range −0.19 to −0.02; odds slightly longer); the effect is smallSupported in this sample
- H2The number of “Doubtful” players also predicts odds moves → not seenInsufficient evidence
- H3Questionable key players who missed the last practice play at a different rate → supported: 48% if they did not practice, 74% limited, 89% full (similar figures were published elsewhere; this is a replication)Supported in this sample
- H4Core starters (on the field for 75%+ of plays) play at a different rate → holds once position, practice and other factors are accounted for (about 5 points higher); the raw rates are almost the same (72% vs. 72%). A split added after seeing the results shows core starters are concentrated in positions with lower playing rates when listed (e.g., offensive line, defensive backs), which may relate to the differenceSupported in this sample
- H5A short week (4 days or less since the last game) is associated with a different playing rate → not seenInsufficient evidence
- H6Questionable key players who missed the previous game play at a different rate → supported: 63% vs. 76% for those who played last gameSupported in this sample
NFL Questionable Players: Research Methods
Research questions
- What share of key players listed “Questionable” on the final injury report actually play? Does it depend on whether they practiced, whether they played last game, how central they are, or a short week?
- Did the odds the night before already account for these public “Questionable” statuses, or did they keep drifting in a predictable direction until kickoff?
Method summary
- Playing rate: 8,756 cases of key players listed “Questionable” on the final report, 2016–2025 regular seasons; “played” = appears on at least one play (including special teams).
- Odds: Polymarket odds 16 hours and 5 minutes before kickoff; every game checked against play-by-play to make sure the final reading came before the actual kickoff. Over 99.6% of injury statuses in 2016–2024 were final at least 16 hours before kickoff.
- One main test (odds); the other five are corrected together for multiple testing. All five data checkpoints were met, and the placebo comparison raised no warning.
- The plan went through four rounds and the results two rounds of external code review; the analysis code was timestamped before it was run.
Data and limitations
Data we use (all public and free)
- Official NFL injury reports (rights belong to the NFL; since 2009, organized by nflverse)
- Play-level participation: NFL Next Gen Stats (2016–2022) and FTN Data (2023–2025), via nflverse, CC BY-SA 4.0
- Schedules, play-by-play (actual kickoff times) and weekly rosters: nflverse
- Polymarket single-game NFL odds (mid prices every 5 minutes; 2024 and 2025 regular seasons)
Limitations
- Retrospective, not blind: the data are public and similar playing rates have been published.
- Only final statuses, not the time of each report; 2025 timing cannot be checked row by row, and 2025 alone is inconclusive.
- The change in implied win chance is tiny (about 0.1 points per net player) and does not imply an edge after spreads and fees; this is not betting advice.
- Polymarket only prices “will the team win,” so we do not judge the market’s view of whether a player will play.
- This page shows only aggregated numbers, no raw data.
Latest progress
Latest：Main results are in: 72% of Questionable key players played; the main odds test holds but the effect is small (about 0.1 points per player); the plan and analysis code were timestamped first, and results were externally reviewed
Next step：Collecting the remaining ~178 games of the 2026 season; re-test with the frozen rules once 2026 participation data are released
Study timeline
- Novelty and data licensing checked; plan finalized and timestamped after four external code reviews
- Analysis code timestamped before running; all five data checkpoints met and no placebo warning
- Main results in; corrected after external code review
- Collecting the remaining ~178 games of 2026; re-test with the same rules once 2026 participation data are released (expected Feb–Mar 2027)In progress
Update log
Version history
- v1.0-results
Main results: 72% of Questionable key players played; the main odds test holds but the effect is small; one robustness figure corrected after external review (roster-based rate 74.5% → 75.0%), main numbers unchanged
- v1.0
Plan finalized and timestamped (T0 = 2026-10-22 07:25 UTC); retrospective, not blind, with the rest of 2026 as a prospective test
Study record
4b5e756ce60ade2bdf0d32399a66bf2e0b49e3a52a22626b2445eb0f90deeb93416b6ea2cf87aeb683a98a3523176a7901824e506f75246f73bf9199495007b99857e17af99197d3a82572b37fc8f956a196570d48b07dc1ddb492e451f91494This page is a research log for reference only and is not betting advice or a guarantee of profit.