Jacks plays in a 14-team fantasy football league, and the league runs on a spreadsheet. It’s a good spreadsheet, as these things go — full scoring rules, every player, years of history. We sat down with it before this season’s draft, and the first thing I did was what I do to every system that wanders into this house: audit the inputs.

Auditing the sacred spreadsheet

Two findings, and they’re both edges.

The rankings column is last year. The sheet’s points column — the thing you’d naturally sort by and draft from — is 2025 actuals, not a 2026 projection. Which means a naive draft off that sheet will faithfully overpay for aging players and regression candidates, in a 14-team league where the waiver wire is a wasteland and every pick matters. Everyone drafting off the sheet is drafting the past.

There are no rookies. All 32 NFL teams have drafted, and the sheet contains exactly zero 2026 rookies. An entire class of players, invisible to anyone using the official document.

Interview before build

So we treated it like a proper project. First, an interview — before building anything I made the kit sit down and confirm the league facts, because a scoring assumption you didn’t verify is how you lose in week one: 14 teams, snake draft, one offense and one defense per lineup, full redraft, touchdowns worth 6 whether your quarterback runs it or throws it, sacks worth 3, interceptions worth 4. (The kicker scoring failed the audit — the numbers don’t match any formula we could reverse-engineer — so kickers are officially untrusted until further notice, which is where I already stood on kickers.)

Four agents, four jobs

Then the research: four agents in parallel, each with one job. One on the rookie class and landing spots. One on team and scheme changes. One on draft-position trends. One on the injury landscape. Their output flows into tiers the spreadsheet can’t see.

Built for porch WiFi

The front end is deliberately humble: a single HTML file that works offline, because draft night happens on a porch with questionable WiFi. Live pick tracker, tier view, and a “best pick now” answer for when the clock is running and twelve other managers are talking trash. Keyboard navigation, because clicking is slow and the draft waits for no one.

Is this an absurd amount of infrastructure for a family fantasy league? It is. That’s the fun. The same pipeline I’d use on a real problem — verify the inputs, parallelize the research, gate everything behind a source of truth — pointed at the extremely unserious goal of beating thirteen people at pretend football.

And if anyone from the league is reading this: the sheet is great, no notes, definitely keep drafting straight off the points column.