Accountability Report
Every projection locks to a hash-chained ledger before kickoff. Every outcome is scored after. No retroactive edits are possible. A single changed digit breaks the chain verifiably. This page publishes every Tuesday in season.
⚠FIXTURE MODE. This is the report format running on the synthetic test event with invented actuals, a rehearsal of the real Tuesday page, including a losing call, because that is the point of the page.
What do these terms mean?
MAE
Mean Absolute Error: the average size of the miss between projection and actual. Lower is more accurate.
Coverage
How often the actual result landed inside the projected floor-to-ceiling band. The target is about 80 percent.
Calibration
Whether stated confidence matches reality: things called 70 percent should happen about 70 percent of the time.
Surprise probability
The modeled chance of an outcome far from projection, either a big beat or a total bust.
Tier-1 % ★
Nova's modeled probability he finishes as a tier-1 (elite) player at his position, from 5000 simulated seasons.
Verdict
Whether the actual result landed inside the projected floor-to-ceiling band (IN RANGE) or outside it, high or low (MISS HIGH / MISS LOW).
★ = Nova’s own model, the secret sauce. The rest are the market reference and raw stats Nova is measured against.
3.97
MAE (PPR points) · 12 players scored
3.14
CRPS (full distribution) · beats MAE
83%
actuals inside our P10–P90 band · target 80%
55-11
start/sit record this week
69%
avg claimed confidence · calibration compares these as weeks accumulate
Distribution scoring
The product is the distribution, so we score the distribution, not just the median. CRPS is the proper score for it, and Brier grades the boom/bust probabilities. A positive skill number means the probabilities beat simply predicting the base rate.
| Claim | Score | Skill | Base rate | Reading |
|---|---|---|---|---|
| Full distribution (CRPS) | 3.14 | n/a | MAE 3.97 | beats the median-only guess |
| Boom probability (≥20 PPR) | 0.0779 | +0.584 | 25% | Brier, lower is better |
| Bust probability (<10 PPR) | 0.1416 | +0.418 | 42% | Brier, lower is better |
Confidence calibration
When we claim a start/sit call is 70% to win, does it win about 70% of the time? This table fills in week by week. Claimed and realized should converge.
| Claimed tier | Claimed avg | Realized | Calls |
|---|---|---|---|
| 50-60% | 56% | 67% | 18 |
| 60-70% | 64% | 73% | 15 |
| 70-80% | 74% | 94% | 18 |
| 80-90% | 85% | 100% | 12 |
| 90-100% | 93% | 100% | 3 |
Every projection, scored
| Player | Projected (median) | Band (P10–P90) | Actual | Error | Actual landed at | Verdict |
|---|---|---|---|---|---|---|
| Joe Burrow | 22.2 | 12.3–34.6 | 37.8 | +15.6 | P94 | MISS HIGH |
| Emeka Egbuka | 10.2 | 4.1–19.9 | 2.6 | -7.6 | P4 | MISS LOW |
| Bucky Irving | 8.7 | 3.8–19.7 | 14.2 | +5.5 | P69 | IN RANGE |
| Chris Godwin | 10.8 | 3.7–21.6 | 6.8 | -4.0 | P26 | IN RANGE |
| Ja'Marr Chase | 18.4 | 7.3–35.9 | 21.4 | +3.0 | P58 | IN RANGE |
| Tee Higgins | 12.7 | 5.0–26.6 | 15.7 | +3.0 | P62 | IN RANGE |
| Baker Mayfield | 19.5 | 10.6–30.5 | 22.4 | +2.9 | P64 | IN RANGE |
| Chase Brown | 11.2 | 3.2–24.9 | 12.8 | +1.6 | P54 | IN RANGE |
| Mike Gesicki | 7.8 | 2.6–20.3 | 6.2 | -1.6 | P38 | IN RANGE |
| Cade Otton | 7.6 | 2.3–16.8 | 8.9 | +1.3 | P60 | IN RANGE |
| Mike Evans | 12.4 | 4.8–26.7 | 11.2 | -1.2 | P44 | IN RANGE |
| Rachaad White | 5.1 | 0.8–14.0 | 5.4 | +0.3 | P52 | IN RANGE |
Start/sit calls, scored
| Call | Claimed | Tier | Result | Outcome |
|---|---|---|---|---|
| Baker Mayfield | 83% | STRONG START | 22.4 vs 14.2 | CORRECT |
| Baker Mayfield | 89% | STRONG START | 22.4 vs 8.9 | CORRECT |
| Baker Mayfield | 75% | STRONG START | 22.4 vs 12.8 | CORRECT |
| Baker Mayfield | 79% | STRONG START | 22.4 vs 6.8 | CORRECT |
| Baker Mayfield | 82% | STRONG START | 22.4 vs 2.6 | CORRECT |
| Baker Mayfield | 52% | COIN FLIP | 22.4 vs 21.4 | CORRECT |
| Joe Burrow | 59% | START | 37.8 vs 22.4 | CORRECT |
| Baker Mayfield | 71% | STRONG START | 22.4 vs 11.2 | CORRECT |
| Baker Mayfield | 85% | STRONG START | 22.4 vs 6.2 | CORRECT |
| Baker Mayfield | 93% | STRONG START | 22.4 vs 5.4 | CORRECT |
| Baker Mayfield | 71% | STRONG START | 22.4 vs 15.7 | CORRECT |
| Bucky Irving | 59% | START | 14.2 vs 8.9 | CORRECT |
| Chase Brown | 55% | LEAN | 12.8 vs 14.2 | WRONG |
| Chris Godwin | 56% | LEAN | 6.8 vs 14.2 | WRONG |
| Emeka Egbuka | 54% | LEAN | 2.6 vs 14.2 | WRONG |
| Ja'Marr Chase | 77% | STRONG START | 21.4 vs 14.2 | CORRECT |
| Joe Burrow | 87% | STRONG START | 37.8 vs 14.2 | CORRECT |
| Mike Evans | 63% | START | 11.2 vs 14.2 | WRONG |
| Bucky Irving | 56% | LEAN | 14.2 vs 6.2 | CORRECT |
| Bucky Irving | 70% | STRONG START | 14.2 vs 5.4 | CORRECT |
| Tee Higgins | 64% | START | 15.7 vs 14.2 | CORRECT |
| Chase Brown | 63% | START | 12.8 vs 8.9 | CORRECT |
| Chris Godwin | 64% | START | 6.8 vs 8.9 | WRONG |
| Emeka Egbuka | 63% | START | 2.6 vs 8.9 | WRONG |
| Ja'Marr Chase | 83% | STRONG START | 21.4 vs 8.9 | CORRECT |
| Joe Burrow | 92% | STRONG START | 37.8 vs 8.9 | CORRECT |
| Mike Evans | 71% | STRONG START | 11.2 vs 8.9 | CORRECT |
| Mike Gesicki | 52% | COIN FLIP | 6.2 vs 8.9 | WRONG |
| Cade Otton | 62% | START | 8.9 vs 5.4 | CORRECT |
| Tee Higgins | 72% | STRONG START | 15.7 vs 8.9 | CORRECT |
| Chris Godwin | 50% | COIN FLIP | 6.8 vs 12.8 | WRONG |
| Chase Brown | 52% | COIN FLIP | 12.8 vs 2.6 | CORRECT |
| Ja'Marr Chase | 71% | STRONG START | 21.4 vs 12.8 | CORRECT |
| Joe Burrow | 80% | STRONG START | 37.8 vs 12.8 | CORRECT |
| Mike Evans | 57% | LEAN | 11.2 vs 12.8 | WRONG |
| Chase Brown | 60% | START | 12.8 vs 6.2 | CORRECT |
| Chase Brown | 72% | STRONG START | 12.8 vs 5.4 | CORRECT |
| Tee Higgins | 57% | LEAN | 15.7 vs 12.8 | CORRECT |
| Chris Godwin | 52% | COIN FLIP | 6.8 vs 2.6 | CORRECT |
| Ja'Marr Chase | 73% | STRONG START | 21.4 vs 6.8 | CORRECT |
| Joe Burrow | 84% | STRONG START | 37.8 vs 6.8 | CORRECT |
| Mike Evans | 58% | START | 11.2 vs 6.8 | CORRECT |
| Chris Godwin | 62% | START | 6.8 vs 6.2 | CORRECT |
| Chris Godwin | 73% | STRONG START | 6.8 vs 5.4 | CORRECT |
| Tee Higgins | 58% | START | 15.7 vs 6.8 | CORRECT |
| Ja'Marr Chase | 76% | STRONG START | 21.4 vs 2.6 | CORRECT |
| Joe Burrow | 87% | STRONG START | 37.8 vs 2.6 | CORRECT |
| Mike Evans | 60% | START | 11.2 vs 2.6 | CORRECT |
| Emeka Egbuka | 61% | START | 2.6 vs 6.2 | WRONG |
| Emeka Egbuka | 73% | STRONG START | 2.6 vs 5.4 | WRONG |
| Tee Higgins | 61% | START | 15.7 vs 2.6 | CORRECT |
| Joe Burrow | 60% | START | 37.8 vs 21.4 | CORRECT |
| Ja'Marr Chase | 66% | STRONG START | 21.4 vs 11.2 | CORRECT |
| Ja'Marr Chase | 80% | STRONG START | 21.4 vs 6.2 | CORRECT |
| Ja'Marr Chase | 88% | STRONG START | 21.4 vs 5.4 | CORRECT |
| Ja'Marr Chase | 65% | STRONG START | 21.4 vs 15.7 | CORRECT |
| Joe Burrow | 76% | STRONG START | 37.8 vs 11.2 | CORRECT |
| Joe Burrow | 88% | STRONG START | 37.8 vs 6.2 | CORRECT |
| Joe Burrow | 95% | STRONG START | 37.8 vs 5.4 | CORRECT |
| Joe Burrow | 76% | STRONG START | 37.8 vs 15.7 | CORRECT |
| Mike Evans | 68% | STRONG START | 11.2 vs 6.2 | CORRECT |
| Mike Evans | 79% | STRONG START | 11.2 vs 5.4 | CORRECT |
| Tee Higgins | 51% | COIN FLIP | 15.7 vs 11.2 | CORRECT |
| Mike Gesicki | 63% | START | 6.2 vs 5.4 | CORRECT |
| Tee Higgins | 69% | STRONG START | 15.7 vs 6.2 | CORRECT |
| Tee Higgins | 79% | STRONG START | 15.7 vs 5.4 | CORRECT |
A 56% call that loses is not a scandal. It is a 56% call. What matters is whether our 56% calls win 56% of the time over the season. That calibration table builds here week by week, in public, on an append-only ledger.