Football

The football record

We make no claim that these plays win. Two pre-registered studies, both published in full, found this market cannot be out-forecast at the moment we can act: a market-blind model lost to the de-vigged consensus in every season and market tested, and buying the best available price returned less than the vig it was paying. So this is not sold as an edge. It is sold as process and receipts — the whole slate reasoned through from market prices, one play we would actually act on, and every one of them graded here in public, win or lose. Football is staked at zero units.

Record reset disclosed: pre-launch test history. Official cohort starts September 11, 2026 under fp-v0.4; earlier boards retain their original rules.

The September 8 incident slate is excluded after audit, including its two featured losses. A zero official record is not an undefeated lifetime record. See retained results and the dated amendment.

Only committed plays count toward the Open Ledger record. Covered-game analysis is published for transparency but is not included in W–L or ROI.

Official football begins with eligible kickoffs on or after September 11, 2026 (America/New_York), under fp-v0.4. No retroactive selections. The weekly Saturday decision schedule remains in force; a date boundary does not guarantee a Friday pick.

Official record — committed plays

Append-only. Nothing is edited, nothing is deleted, losses publish exactly like wins. Every one of these was fingerprinted before kickoff.

Invalidated incident slate — excluded from official performance

The full September 8 slate (two featured picks and 70 additional coverage cards) was invalidated on September 13, 2026, after both featured picks settled as losses. This post-result classification corrects a defective recommendation process; it is not a claim that these picks were never published. Original selections, prices, losses and hashes remain unchanged. No result from this board contributes to official W–L, CLV or hypothetical returns.

Dated classification amendment · Full original slate · Unchanged ledger

Pre-launch / invalidated selection test

Prior selections and results remain intact, including losses. They are excluded from official W–L, CLV and hypothetical returns. This is a disclosed restart, not an undefeated lifetime record. The old grader re-ranked results rather than settling the committed board; historical tier labels are not proof of a committed play.

Reset amendment and selection rule · Full unchanged historical entries · Commitment evidence · Pilot audit manifest

Full covered slate — research, not the record

These 0 games were evaluated and the rule's chosen side is recorded, so its behaviour can be checked rather than described — but none was offered as a position and none counts toward the record above. Games that failed the coverage markers never reach this table at all; they are listed as NO MARKET on each week's page, with the reason.

D.J. Mercer Spotlight

The model above is one record. D.J. Mercer keeps another: human-researched NFL and college picks, fingerprinted before kickoff and graded on their own append-only ledger. The two never mix and neither borrows credibility from the other.

Weeks

• Week of Tuesday, September 8, 2026
• Week of Tuesday, September 1, 2026
• Week of Tuesday, August 25, 2026

Today's board · MLB ledger · The Morning Line · Live research observations →