All calculations are based off FiveThirtyEight win probabilities. Analysis presumes BFIG competitors will pick the available team with the highest win probability that week, which of course isn't always true. Because Moxie, because strategy, and because we don't always agree with computer algorithms.
After Week 10, with seven games still in front of us we had nine BFIG combatants with a 24.2 percent chance of running the table. You arrive at that number by simulating a pick line that takes the likeliest available team to win each week.
Let's take @billynickell, who's been featured in several Commish write-ups thus far:
- His pick line through 10 weeks: SF CAR MIA WAS IND TEN CIN OAK NO BAL.
- His projected remaining picks, optimizing for FiveThirtyEight's win probability: KC (Week 11, 84% win probability) BUF (84%) NE (81%) DET (77%) ATL (84%) LA (76%) PIT (Week 17, 86%).
- Multiply each of those percentages and you arrive at a 24.16% probability that Billy runs the table.
It's fitting that the "optimal" Week 11 pick was Kansas City, which promptly lost at home to Tampa Bay. And indeed, that very quickly reduced @billynickell's Title Belt odds, along with fellow leaders @cbroglie and @amehamolla, to zero.
There's no such thing as a "right" pick - even when using a computer algorithm. Else, survivor would be a far easier game. KC had an 84% win probability, but Detroit was 83%, New England was 82%, and both Dallas and the Giants were 75%. Lots of heavy favorites in Week 11, and in the parity-riddled NFL, a 9% difference in win probability basically amounts to making vs. missing a field goal in the first half.
Here was our top 20 after Week 10, with Week 11 picks also shown:
Oakland is yet to play (16 picks - 4% of the title pool), but if we presume an Oakland win, then the Chiefs loss (33 people - 9% of title pool) was the only damage in Week 11.
Aside from the obvious effect of everyone's perfect slate probability going up (because Week 11 is now "100%" instead of, say, 82%), that means not much changed in the overall pecking order.
Let's look at the updated standings, followed by who made the biggest jump up based on their Week 11 pick:
And here are the biggest risers. Detroit and Kansas City each have some high win-likelihood games still to play, so by virtue of picking a team with a lower Week 11 win probability than the Lions or Chiefs, plus not having very many good options besides the Lions or Chiefs in those future weeks, these people achieved outsized gains relative to their fellow survivors who might have more available options.
In a sense, presuming everyone is now looking ahead at future potential picks, NOT taking the Chiefs or Lions was probably obvious to these people. Still, their Week 11 wins gave them a big boost:
Finally, there's the flip side to the above chart - the people who won but whose perfect slate probability improved the least. Basically, who took a big favorite now at the expense of their future options. We should just call this "The Patriots Dilemma," because every year it's always a question of when to use your Patriots pick.
Some people who picked New England this week dropped a future week's win likelihood by as much as 14 percent. That's not objectively "bad," though. The Giants almost lost today, and several other games were close. If the underdogs pull those out, you look at Patriots picks as smart moves.
This chart simply says that, even with their win today, their perfect 17-0 slate probability barely moved:
These "perfect slate" probabilities seem to suggest we're headed for a Title Belt Tiebreaker. At least two of the 300+ remaining have to cash in on that ~20-30 percent likelihood, right?
Just remember, all it takes is a few ELI FACES and Cutler "your turn to cry(baby)" performances to see title pool numbers come crashing down.
Commish note: What other stats and analysis do you want to see? Let Commish know by writing me at yo@theszn.com. Thanks, Haiming, for this killer work!