Which office knows football best?

Every company on this table is one office running its own NFL pick'em pool in Slack or Google Chat. Coworkers call the straight-up winner of each game in a private message, the bot scores the week as the games finish, and what appears here is the office total. Nothing about an individual player is published.

The player list locks at the first kickoff of the week. Whoever is in the pool when the opening game of the week starts is the roster for that company that week, and it cannot change afterwards. That is what stops a company adding five confident colleagues on a Sunday night, and it is why a company that grows mid-season sees the new arrivals counted from the following week rather than retrospectively.

A company needs at least five players in that locked list to be ranked. Smaller pools keep playing exactly as normal, and their office table works the same way; they simply stay private until the fifth colleague joins.

A missing pick counts as a wrong pick. Accuracy is correct picks divided by every pick that was available to the locked roster, so the denominator is the roster multiplied by the games that have actually finished. A future game is never counted as a miss, and an office cannot improve its accuracy by having half the room sit the week out.

The weekly view ranks accuracy for that single week and is the fairest comparison on the page: every ranked company is judged on the same games. The season view adds up the correct picks per player from every week played, which rewards companies that have played more weeks. That is deliberate rather than accidental, and it is why the week each company joined is shown next to its name.

The table refreshes about every five minutes while games are being played, and a week is marked final once every game in it has finished.