Going without a viable superflex QB cuts weekly win chances by 8.8 percentage points
In 403,792 superflex lineups across 2,594 leagues, 17.7% had no viable QB in a superflex slot. Compared with the same team's other weeks, those weeks were won 8.8 percentage points less often.
By Elliot Pollaro · Published 2026-10-06 · Pre-registered (DOI 10.17605/OSF.IO/F8K59)
Key findings
- A superflex slot had no viable QB in 17.7% of lineups (95% CI 17.4%–18.1%): a non-QB started there, the slot was empty, or the QB scored 0. The rate rose from 8.0% in week 1 to 29.4% in week 8.
- Compared with the same team's other weeks, going without one cut the chance of winning by 8.8 percentage points (95% CI 8.3–9.2). In odds, that is 0.67 times the odds of winning (95% CI 0.66–0.69).
- Exploratory: starting a WR, RB or TE there (15.8% of lineups) cost 7.7 percentage points, and a QB who scored 0 cost 16.2.
- Replicated in 220,936 superflex lineups from 2024: 18.0% had no viable superflex QB, and going without one cut weekly win chances by 8.9 percentage points (odds ratio 0.67).
How often it happens
The study covers 403,792 regular-season team-weeks in 2,594 superflex leagues (28,968 team-seasons). The rate climbs through the middle of the season, when NFL bye weeks and injuries pile up, and peaks at 29.4% in week 8.
| Week | Teams without a viable SF QB | 95% CI | Team-weeks |
|---|---|---|---|
| 1 | 8.0% | 7.7%–8.4% | 28,728 |
| 2 | 9.0% | 8.6%–9.4% | 28,784 |
| 3 | 11.6% | 11.2%–12.0% | 28,786 |
| 4 | 10.9% | 10.5%–11.3% | 28,830 |
| 5 | 20.8% | 20.4%–21.3% | 28,834 |
| 6 | 18.3% | 17.8%–18.7% | 28,856 |
| 7 | 17.0% | 16.6%–17.5% | 28,882 |
| 8 | 29.4% | 28.9%–29.9% | 28,890 |
| 9 | 21.6% | 21.1%–22.1% | 28,914 |
| 10 | 22.6% | 22.1%–23.0% | 28,924 |
| 11 | 16.4% | 15.9%–16.9% | 28,930 |
| 12 | 25.8% | 25.3%–26.3% | 28,876 |
| 13 | 15.4% | 15.0%–15.8% | 28,858 |
| 14 | 21.0% | 20.5%–21.5% | 28,068 |
| 15 | 12.7% | 9.5%–15.9% | 632 |
What it costs
Raw win rates were 37.1% without a viable superflex QB and 52.8% with one. The pre-registered test compares each team only with itself across its own season, so team strength drops out. Going without a viable superflex QB cut the odds of winning to 0.67 times (95% CI 0.66–0.69), a drop of 8.8 percentage points in win probability (95% CI 8.3–9.2).
| Week type | Win rate |
|---|---|
| Viable SF QB | 52.8% |
| No viable SF QB | 37.1% |
Why the slot had no QB (exploratory)
Most of these weeks were managers starting a WR, RB or TE in the superflex slot (15.8% of all lineups), and those weeks cost 7.7 percentage points. A QB who started and scored 0 cost far more, 16.2 points. This split was frozen into the analysis before unblinding but is not in the registered plan.
| Superflex slot that week | Team-weeks | Share | Won | Change vs the same team's viable weeks, points (95% CI) |
|---|---|---|---|---|
| Viable QB in every superflex slot | 332,290 | 82.3% | 52.8% | baseline |
| A WR, RB or TE started there | 63,727 | 15.8% | 38.7% | -7.7 (-8.2 to -7.2) |
| A QB who scored 0 | 6,904 | 1.7% | 24.4% | -16.2 (-17.3 to -15.0) |
| Slot left empty | 871 | 0.2% | 18.1% | -16.8 (-21.0 to -12.5) |
Does it hold in 2024? Yes
The same frozen code was run on an independent sample of 2024 superflex leagues, covering 220,936 team-weeks in 1,397 leagues. A superflex slot had no viable QB in 18.0% of lineups (95% CI 17.5%–18.5%). Going without one cut the chance of winning that week by 8.9 percentage points (95% CI 8.3–9.5), an odds ratio of 0.67 (95% CI 0.65–0.69). The finding replicated almost exactly.
How the study works
Starters are mapped to lineup slots in the platform's own lineup order. In the data, 4,138,784 of 4,138,785 checkable starts were eligible for their mapped slot. A week counts as exposed when a superflex slot started a non-QB, was empty, or started a QB who scored exactly 0. The result uses the official score, including commissioner overrides, and ties are dropped.
Who is in the sample
Sampled leagues, and how many each eligibility rule excluded.
| Sampled leagues | Count |
|---|---|
| No games played | 446 |
| More than one team without an owner | 137 |
| Not a completed dynasty season (or best ball) | 162 |
| No decided championship | 20 |
| Analyzed | 4,235 |
Limitations
- Some exposed weeks reflect injuries or byes, and others reflect roster construction. Comparing a team with itself removes stable team strength but not every week-specific shock, such as an injury that also hit other starters.
- A QB with 0 points is often a late scratch or an in-game injury, which can hurt a lineup in other ways too.
- One platform's public superflex dynasty leagues in one regular season.
Data, method and verification
Public dynasty leagues on a major fantasy football platform, read from the platform’s own public league records (settings, rosters, weekly lineups and scores, transactions, playoff brackets). None of Dynasty Dealer’s player values, trade database or users are used.
The plan was registered before any result was seen: Dynasty League Case Studies: Pre-Registered Analysis Plan (DOI 10.17605/OSF.IO/F8K59). Every team-week's exposure and result was reproduced by a second, independently written implementation, and randomly chosen lineups were checked by hand against the raw league records.
How to cite
Elliot Pollaro (2026). Going without a viable superflex QB cuts weekly win chances by 8.8 percentage points. Dynasty Dealer. https://www.dynastydealer.com/case-studies/superflex-qb-shortage. Pre-registration: https://doi.org/10.17605/OSF.IO/F8K59
FAQ
How much does it cost to start a non-QB in superflex?
Across 403,792 superflex lineups, starting a WR, RB or TE in the superflex slot went with a 7.7 percentage point lower chance of winning that week, compared with the same team's weeks with a viable QB (exploratory). For any week without a viable QB there, the pre-registered estimate is 8.8 points.
Where does the data come from?
From the public league records of a major fantasy football platform: weekly lineups, player points and matchup results. None of Dynasty Dealer's data is used.
Was the analysis planned in advance?
Yes. The plan was registered on OSF before any result was seen (DOI 10.17605/OSF.IO/F8K59).