Public scorecard · Updated Nov 12, 2026
How accurate are our projections and calls?
We save every prediction with a timestamp and grade it once the season plays out. First rookie grades: after the 2027 season. Meanwhile, here’s how our models did when tested on past drafts. The live log holds 935 saved predictions so far.
Tested on past drafts: the rookie projection replayed on past classes
Tested on past drafts, not live. We replayed every draft class since 2014: for each one, the projection was built only from the other classes, then compared with what the players really did. The range we show is meant to catch the real result 8 times in 10.
Rookie projections: how often the real result landed inside our range
Across every class since 2014 the real result landed inside our range 8, 8, 8 times out of 10 in years 1, 2 and 3. For the 2021-and-later classes, the newest college era, it was 8, 8, 8 out of 10.
| Draft classes | Year 1 | Year 2 | Year 3 |
|---|---|---|---|
| All past classes (2014–2025) | 8 in 10 | 8 in 10 | 8 in 10 |
| Earlier classes (through 2020) | 8 in 10 | 8 in 10 | 8 in 10 |
| Recent classes (2021 and later) | 8 in 10 | 8 in 10 | 8 in 10 |
The range is built to catch the real result 8 times in 10, so “8 in 10” is exactly on target.
What the college record adds
A player’s college record adds a small, real signal beyond where he was drafted. It improved the projection on every one of the first three NFL years for the recent classes, but draft position still does most of the work.
We tested player comparisons on past drafts and they weren’t reliable enough to show, so we don’t.
Does the newer college era change things?
We checked whether college production predicts the NFL differently for players drafted in 2021 or later (more seasons, the transfer portal, the extra COVID year). The data shows no clear difference, but only three recent classes have finished, so we cannot rule out a small one.
How we keep eras fair
Since 2020 many players spend five to seven years in college, so we never compare career totals: a sixth-year senior would win on volume alone. Instead we use per-game rates from a player’s two best seasons, and we give more credit to production at a younger age. Every player is measured against that season’s power-conference starters at his position, and we tested the recent classes separately from the earlier ones.
What this does not cover
- Projections use draft position and college record. The NFL team a player lands on is unknown until the draft, so it is not counted yet.
- Birth dates are not available for the 2027 class, so ages are estimated from when each player first appears in college stats.
- Three recent classes are too few to be certain about the newest era; the page says so rather than hiding it.
Live predictions saved so far
Live, not a past-draft test. These were saved before the games or seasons they cover, with the date they were made.
advisor-v1
67 logged · 0 scored
Waiting for the games or days that score these calls.
Rookie projections
How many fantasy points a drafted rookie scores per week in each of his first three NFL seasons.
354 logged · 0 scored
The first scores arrive in March 2028, when the first NFL season for the 2027 class is over.
Rookie rankings
The order of The Big Board Rookie Rankings, saved every week before we know how the players do.
496 logged · 0 scored
The last ranking before the 2027 draft is graded in March 2028, against how the players really did in the NFL and against the simple draft-order guess.
Usage-shift calls
Who gets the targets or carries an injured player leaves behind (our News cards).
14 logged · 6 scored
Right 4 of 6 calls that moved; 0 calls saw no clear move.
Closer than “nothing changes” in 7 of 10 scored calls.
Value calls
Whether a player’s dynasty value goes up or down after an injury or return (our News cards).
4 logged · 1 scored
Right 1 of 1 calls that moved; 0 calls saw no clear move.
Calls from the News cards are scored every week during the season. Rookie projections are scored once a season is over.
How we project this
Rookie projections start from where a player was drafted, using every drafted quarterback, running back, receiver and tight end since 2014.
A small adjustment for his college record can nudge the projection up or down, but draft position still does most of the work.
Every prediction goes into a public log with the date it was made, so it cannot be changed afterwards.
The full method is written up in our public notes (docs/methods/accuracy.md in the project).
Frequently asked questions
Can you change a prediction after the fact?
No. Predictions go into a public, append-only log. A change in our numbers adds a new entry; the old one stays and is scored too. The log computes its own fingerprint of each entry, so an entry cannot be quietly edited.
What does “tested on past drafts” mean here?
We rebuild the projection for a past draft class using only the other classes, then compare it with what those players really did. It shows how the method would have done, but it is not the same as a live prediction, so we keep the two apart.
Why are there no live rookie scores yet?
A rookie projection covers his first three NFL seasons. The first season of the 2027 class is scored after it ends, in March 2028.
What are we comparing against?
The simplest honest guess: for rookies, what players drafted at the same spot usually score; for news calls, “nothing changes”. We report whether we beat it.