Every pick. Locked in public. Receipts all season.

“Week 6 is locked, and this card ended up being a little more aggressive than I expected. There are a handful of teams I’m comfortable riding with this week. Baltimore, Indianapolis, Kansas City, Buffalo, the Rams and San Francisco all made the cut without a ton of internal debate. I also like Denver at home against Seattle and New England against the Jets. The more interesting part of this week is where I’m willing to go against what probably feels like the safer public pick. I’m taking Houston over Jacksonville, New Orleans over the Giants, Carolina to go into Philadelphia and win, and Pittsburgh over Tampa Bay. Those are the games that will probably determine whether this ends up being a great week or one where I’m staring at the board Sunday night wondering what I was thinking. And then, of course, there’s Dallas. The Cowboys are 2-3, they just lost to Tampa Bay, and the model is taking Green Bay. I’m still taking Dallas. That is exactly why I built the Cowboys Bias part of this experiment in the first place. I know I’m capable of talking myself into the Cowboys when the numbers say otherwise, so rather than pretending that bias doesn’t exist, we’re putting a scoreboard on it. If Dallas goes into Green Bay and wins, I’ll happily take the receipt. If they lose, Homer Tax gets another data point. Fourteen games are published. No changing the story after the fact. Now we see what happens.”

A published pick receives a timestamp. At kickoff it becomes immutable in Postgres. The season record is the record.