A clean NFL record is not an edge: 93% of the NFL board's clean wallets are underwater after fees, by a median of 0.40%

Twelve days before the September 9 kickoff, our graded NFL board of Polymarket wallets holds 1,295 of them. 592 of them pass our farming-risk check — and 552 of those 592, 93.2%, have a negative real edge: after fees and the latency a copier eats, the typical clean NFL wallet is losing money. Its median real edge is −0.40%. Exactly 15 clean wallets on the board carry a real edge above 1%, and nine clear every gate for a positive copy verdict — nine accounts with $58,160 of NFL volume between them, eight of which we have observed for under 60 days.

A clean record and a good score are not the same thing as an edge. The NFL board is the clearest demonstration of that gap we have, and it is the one thing this post exists to make precise.

All wallet figures are a point-in-time pull from CopyGrade's live scored set via the read-only database, dated 2026-08-28. "Real edge" is our post-fee, post-latency estimate of what a copier would actually have kept, per the published methodology.

Clean wallets lose more often — and lose far less

Split the board by farming level and read two columns at once. The first counts signs. The second measures size.

Read naively, that first column says the flagged wallets are more often profitable than the clean ones — 81.4% underwater against 93.2% — which looks like evidence against the model. It is the opposite. Clean wallets cluster just below break-even: a fat, narrow lump around −0.40%. Severe-flagged wallets have a far fatter tail in both directions — that is what lifts their share of positive signs while making their typical outcome more than three-and-a-half times worse. Manufactured records are exactly what produce outsized apparent returns, and outsized apparent returns are the signal the detectors fire on.

So: clean NFL wallets lose more often and far less. Any post that quotes only the first half of that sentence has the finding backwards.

How many NFL wallets actually have an edge

Walk the board down from the top.

Two things in that table are worth the whole post.

First, every clean NFL wallet with a real edge above 1% also scores 60 or better — the score and the edge agree at the top of the board. They just agree on fifteen wallets out of 1,295. Below that, 507 clean wallets score under 60 and net out to −$2,761 on $8.4M of turnover; the clean cohort's entire profit is carried by its 85 best-graded accounts, the same concentration we found on a board a quarter this size on August 5.

Second, the nine positive verdicts — which are also the nine wallets clearing the canonical copyable bar (positive verdict, clean farming check, positive real edge, at least 20 trades in the last 90 days) — have traded $58,160 of NFL volume combined, are up $648 between them, and eight of the nine have been under observation for less than 60 days (median 17 days; the youngest, four). They hold positive verdicts because their overall records clear the bar, not because they have proven anything on football. That is the brand-new-wallet pattern again, and reading a score against how long we have watched the wallet is the habit that keeps it from fooling you.

This is not an NFL quirk

We grade 25 category boards. Rank them by the share of wallets with negative real edge and the same shape appears at the extremes:

The boards where the fewest wallets are underwater are boards with a median score of 30 and flag rates around 70% or above. All eight boards with a median score of 30 — tennis, WNBA, esports, commodities, UFC, IPOs, cricket, boxing — sit in the bottom twelve by negative-edge share, and every one of them has a median real edge of −3% or worse. The eleven boards with a median score of 48 or better sit between −0.40% and −1.60%. Fewer wallets losing, each losing several times more: signs versus magnitude, at board scale.

NFL's place in that table is exact and unremarkable: third of 25 on negative-edge share, behind F1 and golf; fifteenth of 25 on median score and fifteenth-lowest on flag rate. Where it does stand out is the clean cohort itself — of the 25 boards, only F1's clean wallets are underwater more often than NFL's, and by four hundredths of a point (93.28% against 93.24%).

One caveat that belongs beside every board-versus-site comparison: board membership selects for flagged wallets. Qualifying for a board means at least five trades concentrated in one topic, and concentration is exactly the repetition the detectors read. Every one of the 25 boards sits above the site-wide flag rate of 37.4% — the lowest, F1, at 42.2% — so an NFL flag rate of 54.3% is a finding about the filter as much as about football.

What changed since August 5, and what we can't say about it

On August 5 the board held 319 wallets and the clean cohort out-earned the flagged one in dollars — +$116,860 against +$99,138 for watch and severe together (that post compared clean against the severe cohort alone, +$79,060; the comparison holds either way) — which made NFL the one sports board where the honest money was not simply donating. On today's board of 1,295 that is no longer true:

The clean cohort's profit barely moved — up $1,461 while its headcount quadrupled — and the flagged cohort's profit quadrupled with it. We cannot tell you whether that is the market or our reach. Our whole graded universe went from 2,646 wallets to 8,405 over the same stretch as ingestion widened, the NFL board's share of it moved only from 12.1% to 15.4%, and farming signals are recomputed from scratch on every sync, so there is no earlier per-board state to diff against. What we can say is that it is not the games: Week 1 has not kicked off, and the $14.2M traded across the 34 preseason games already settled is a fifth of the $73.4M traded lifetime across the open NFL futures and prop events.

The commitment from August 5 stands. We will re-grade the board after kickoff and publish what changed — and if the clean cohort's dollar lead was an artifact of a small pre-season board rather than a property of football, that goes on the record too.

What to do with the twelve days

1. Vet on edge, not on record. For anyone copy trading NFL markets on Polymarket, a clean farming check is a veto passed, not a reason to copy — and it is not what makes an NFL wallet worth copying. The number that decides whether a wallet is worth following is its real edge after fees and latency, and on this board it is at or above zero for 40 clean wallets out of 592. Run the full checklist on the ones that survive.
2. Demand weeks of observation, not days. Eight of the nine positive verdicts are under 60 days old. A positive verdict on a wallet we have watched for a month is a starting point; on a wallet we have watched for a week it is barely a rumor.
3. Build the watchlist before the records get noisy. A wallet added today has a pre-season baseline the season can be measured against; a wallet added in October has a four-week record and no way to tell skill from a hot start. The season guide covers the board-specific mechanics — news latency, sides-and-totals versus longshot futures, the sports taker fee.

Caveats

These figures cover CopyGrade's covered universe, not all of Polymarket, and are a dated snapshot — the board recomputes with every sync and will have moved by the time you read this. Real edge is an estimate from public trade history against a published methodology; realized profit understates wallets holding unresolved positions, which on a board that is still mostly futures is a real limitation. Board attribution is per category, so wallets active elsewhere appear on other boards too. A farming-risk flag is an algorithmic risk assessment, not an accusation of conduct. A "positive" verdict is where diligence starts, not a recommendation — our own validation of whether the score predicts copier returns is still null.

CopyGrade is analysis-only — it never executes trades, holds funds, or custodies keys, and a CopyGrade Score is a documented research opinion, not financial advice.