Expected Playoff Games Played - Footballguys

Photo: Dale Zanine, USA Today Sports

In playoff fantasy football, predicting the correct number of games played for each team is paramount. This article, now in its fifth year, uses statistics to do just that. If you're unfamiliar with how my system works, click on any (or all) of the following links to the previous installments:

• 2014
• 2015
• 2016
• 2017

The probability-based process underlying how I project expected playoff games remains the same this postseason, but the sources I used this year are slightly different than the ones I've used in years past (mainly because certain outlets don't seem to care about providing the same data from one year to the next).

For win probabilities related to this week's Wild Card games, I used data from FiveThirtyEight, Vegas Insider, Pro Football Reference, ESPN, and Pro Football Focus. To determine each team's probability of advancing beyond the Divisional Round, I used additional data from FiveThirtyEight, Vegas Insider, and ESPN, as well as data from Football Outsiders.

The end result of all this data collection and applied mathematics is an estimate of each playoff team's probability of playing exactly one game, exactly two games, exactly three games, or exactly four games. As an example of one such calculation, here's how it worked for Kansas City:

• They have a first-round bye, so they can only play three games at most. Therefore, their probability of playing exactly four games is 0.0%.
• Based on data from Football Outsiders and FiveThirtyEight, they have a 64.7% chance of reaching the AFC Championship game, which means they have a 35.3% probability (i.e., 100% minus 64.7%) of playing exactly one game (i.e., losing their first game in the Divisional Round).
• Similarly, based on data from Football Outsiders, FiveThirtyEight, and Vegas Insider, they have a 38.9% probability of making it to the Super Bowl, i.e., playing exactly three games.
• With the above probabilities of playing exactly four games, one game, or three games, we can easily calculate their probability of playing exactly two games via simple subtraction: 100% minus 0% minus 35.3% minus 38.9% equals 25.8%.

These probabilities, (i.e., 35.3%, 25.8%, 38.9%, and 0.0%) are what's displayed in the table below.

EXPECTED PLAYOFF GAMES PLAYED

Based on the above-detailed process, here's how many games each playoff team is expected to play, both in terms of normal rules (EXP G) and "Super Bowl counts double" rules (ADJ EXP G):

These data suggest there are something like five teams to target in playoff fantasy leagues. Four of these five teams have a first-round bye, which has been of considerable import of late. That means the outlier fifth team, Chicago, is the clear value in playoff fantasy contests. A glance at the table's percentages reveals why. The first data point in their favor is that, among teams playing on Wild Card weekend, they have the third-highest probability of playing exactly two games. But far more important is that they have -- by far -- the highest probability of playing four games.

PROJECTED PLAYOFF FANTASY POINTS

As always, I've used the expected game totals above to calculate the expected FFPC points for both standard (Points) and "Super Bowl counts double" (Adjusted Points) contests.

What's different this season is that, rather than exclusively relying on scoring averages for Weeks 13-16 (with tweaks for unsustainable per-game averages), I did some statistical modeling instead. Specifically, I used regression analyses to determine which groups of regular season games best-predicted playoff scoring averages.

For each position, I split player game logs for 2014-2017 into six groups:

• Weeks 1-4
• Weeks 5-8
• Weeks 9-12
• Weeks 13-16
• Weeks 9-16
• Weeks 1-16

For instance, in 2017, Tom Brady averaged the following:

• 27.9 points in Weeks 1-4
• 20.1 points in Weeks 5-8
• 26.7 points in Weeks 9-12
• 15.5 points in Weeks 13-16
• 20.3 points in Weeks 9-16
• 22.3 points in Weeks 1-16

Which one (or more) of these scoring average groups best-predicted playoff points per game from 2014-2017? Well, the results of my regression analyses revealed that regular season scoring average in Weeks 9-16 best predicted playoff points per game for quarterbacks and wide receivers. Meanwhile, for running backs and tight ends, regular season scoring average in Weeks 13-16 best predicted playoff points per game. Therefore, rather than multiplying expected games -- as well as adjusted expected games -- by raw Week 13-16 scoring averages as I have in the past, I instead multiplied by my regression-based expected scoring averages.

As an example of what I mean, let's refer back to Tom Brady. This season, he averaged 18.4 points for Weeks 9-16. The regression model for quarterbacks says to multiply that by 0.72; doing so yields an expectation of 13.2 points per game in the 2018 playoffs. From there, I multiplied 13.2 times New England's 1.9 expected games from the table above to get 25.0 total playoff points, which is the value shown in the table below:

Follow @IR_DannyT