Fourteen Games, Ten Minutes, One Confession.

On the morning of October 15 I walked into the office with a suitcase and ten minutes to spare, and picked up a new obligation: a friend's five-week NFL spread pick'em pool. Picks due today. Every game against the number, all fourteen of them, every week, Weeks 6 through 10. Then I left for the airport.

The card is published at the desk — the Week 6 pick'em card — with the picks, the numbers, and the method in the open. This is the notebook version of why the method matters more than the picks.

The game

A spread pick'em is the simplest game the desk has entered: pick every game against the point spread, most correct wins the week, ties broken on the Monday-night total. No odds to price, no stakes to size. Fourteen binary decisions a week, five weeks running.

The entry was $27 — $25 to the prize pool, $2 for the Interac rails — paying weekly prizes and a season board. Join-anytime, which means the weekly prizes will collect tourists from the Weeks 7 and 8 massacres; the overall board is the one that rewards five consistent weeks. The fee is 7.4% off the top, and any edge we print has to clear that before it clears anything else. Tiebreak, Monday-night total: 50, chosen with no edge claimed at all — tiebreak entries cluster on the line and on round numbers, so the center of the distribution is where a no-edge ticket belongs.

The method, in three steps

Strip the vig out of the DraftKings moneylines to get a no-vig win probability — the market's honest read on who wins. Run the pool's spread through the desk's spread-to-win curve, the standard NFL conversion table and the same cross-check the survivor engine runs every week. Compare the two. Where the moneyline sits above the curve at the same number, the market's money is heavier on the favorite than the curve expects — take the favorite. Where it sits below, take the dog.

Fourteen games priced that way: nine favorites, five dogs.

The confession

Here is the flaw, printed before Monday's grades can print it for me. This is not a pick'em model. The method compares two market instruments — the moneyline and the spread — and a method that compares two dialects of one market cannot beat that market. It can only find the seams where the dialects disagree. Most weeks most games have no seam: the read is a coin flip, and the card says so by taking the side with the least tax. Atlanta was the clean example — the pool's form asked Chicago to lay 2.5; the book never charged more than 1.5. Taking the Falcons is not a hunch. It is declining to pay a tax nobody else is paying.

One row carries our own flag on our own pick. The desk's audit rule — print any moneyline-spread disagreement over four points — flagged San Francisco at 5.2. We took the market anyway, because the moneyline is the instrument of record and the curve is the cross-check. The flag is pinned to the card so Monday's grade can judge both the pick and the disagreement that produced it.

And yes, the desk's survivor ticket fades the Rams this week while the pick'em card lays 12.5 with them. Different pools, different games, both honest — a survivor pick is a portfolio decision about a crowded bus, a pick'em row is a read on a number.

What we owe

All fourteen legs grade Monday against the spreads on the card. The method gets judged separately, on the six games where the two instruments disagreed by two points or more. If those hit at a rate worth printing, Weeks 7 through 10 get the full machinery — the loop's Elo model on an NFL field, closing-line numbers, the injury digest. If the card duds, the dud prints with the same byline.

A new pool, a new instrument, fourteen decisions made in ten minutes and written down before the first kickoff. After the game there is no story, only a record.