(Updated w/scenarios Nov 9) Last year, I put together playoff probabilities for my fantasy football league (link).

This year, I’m starting the forecast even earlier! With 4 weeks left, there are almost 17 million possible outcomes just in terms of W/L (2^6=64 outcomes per week, for four weeks = 16,777,216) and I looked at them all (and the approximate likelihood they occur). The table below shows each team’s expected wins and rank, the probability that they make the playoffs, and the probability of earning that important first-round bye.

icon |
Team (record) |
E(Wins) |
E(Rank) |
Playoffs(%) |
Bye(%) |
---|---|---|---|---|---|

Pack (8-1) | 9.89 | 1.62 | 100.00 | 89.45 | |

Vikings (7-2) | 9.25 | 1.91 | 99.47 | 80.03 | |

Jets (6-3) | 8.39 | 3.60 | 96.13 | 18.98 | |

Economists (5-4) | 7.50 | 4.63 | 85.53 | 6.04 | |

Wonders (5-4) | 7.49 | 5.52 | 70.00 | 1.45 | |

Giants (6-3) | 7.34 | 6.02 | 54.67 | 2.83 | |

IncdtlPun (5-4) | 7.31 | 6.33 | 48.16 | 1.20 | |

NotGonnaLie (4-5) | 6.60 | 6.49 | 45.69 | 0.03 | |

Wentz (3-6) | 4.11 | 10.21 | 0.00 | 0.00 | |

Fourth (2-7) | 3.78 | 9.88 | 0.00 | 0.00 | |

Broncos (2-7) | 3.78 | 10.35 | 0.01 | 0.00 | |

Blue (1-8) | 2.56 | 11.39 | 0.00 | 0.00 |

The Pack, the Vikings and the Jets are all basically in the playoffs. I have a very good shot (85%), but still have room to mess it up.

There are three clear tiers: The top four (just above mentioned teams) are almost all going to make the playoffs. The middle four (Wonders, Giants, IncidentalPunishment and NotGonnaLie) are battling for two slots (and potentially taking mine). The bottom four (Wentz, Fourth, Broncos and GoBlue) are essentially out of playoff contention already. Even though the Giants are 6-3, due to schedules and expected points my alogorithm is fairly bearish on their playoff chances. Their low tie breaker (points scored) is probably a major factor.

The three graphs below show the distribution of final rank by each team (grouped by tier). You can see the clear groupings, and the fact that the Pack and Vikings are almost surely going to go 1-2. Erik has a very small chance of coming in first (2%) and Erik and I have a chance of stealing the bye through grabbing 2nd (16% and 6% respectively). I have the ability to run scenarios and see how probabilities change given a certain outcome, so let me know if there is something you’d like to see. Some scenarios are shown after the distributions.

**Scenario: IncidentalPunishment vs Giants**

- If IncidentalPunishment wins, their probability of making the playoffs rises to 68.49%

(from 49.46%) but if they lose it falls to 21.02%. - If the Giants win, their probability of making the playoffs rises to 81.53% (from 56.47%) but if they lose it falls to 38.59%. They also would have a decent (relatively speaking) probability of getting a first round bye with the win: 6.37%, which decreases everyone else’s.
- Interestingly, this game does not have that much significance outside of these two. Both NotGonnaLie and my team (Economists) has a slightly better chance (~1.25%) of making the playoffs if the Giants win. The Vikings, Jets and Economists chances of a first round bye increase with a Giants loss (1%, 0.2% 0.2% respectively).

**Scenario: Economists (me) vs Fourth **

- With a win, I would improve my playoff chances to 91%. If Fourth won, their chances would moves to just above 0% (0.10%)
- With a loss, my playoff chances drop 64%. My distribution (not shown) would resemble the middle crew.
- My loss is the gain mainly of the middle crew. Each of them increases their playoff probability by around 4%, but IncidentalPunishment

increases their probability by 6%. I’m not sure why they get an extra big boost.

P.S. Short version of methodology: used a version of Pythagorean expected wins to compute win probabilities (exponent of 6) – this seems to be similar to what Yahoo uses in their projections. This time I looked at the optimal lineup for future weeks to get predicted points. Then I computed the probabilities for each of the 16m+ possible win / loss outcomes (2^6=64 outcomes per week for four weeks). Also, had to make assumptions about points scored for the tiebreakers (gave the winner the greater of the two expectations).