Simulating the winner of the 2015 Superbowl

It's an annual tradition for me to simulate the potential winner of the Superbowl by simulating the NFL playoffs tens of thousands of times.  As always I have used the Sagarin ratings pure points to predict the outcome of each game.  The difference in the ratings yields the predicted spread but we must also remember that over the history of the NFL the standard deviation in the actual results vs. the predicted spread is almost two touchdowns, 13.86.

So on this last day before the wildcard playoff games begin, here are the predictions.  The bar chart below shows outcomes sorted by the most frequent results of 50,000 simulations.

Seattle wins the Superbowl 32% of the time in the simulations, followed by New England at 24%, Denver at 18%, and Green Bay at 13%.  I think that jibes with the expectations of sports analysts.  Those four teams dominate the simulations, every other teams chance is less than 4%.

The pie chart above shows the chance of the teams meeting in the Superbowl.  Blue represents the NFC team winning (on the Y-axis, names of each row), while red represents the AFC team winning (X-axis, names of each column).  Again, SEA, NE, DEN and GB dominate the results, they appear in the final game often, especially playing against leaders in the other conference.  The block diagram below shows the results in a different format colored by the teams who win the Superbowl.  Seattle dominates, and the chart shows Seattle's opponents and the frequency of that matchup out of the 50,000 simulations.

These simulations used to serve the purpose of letting me pick players for a playoff fantasy football game that unfortunately doesn't run anymore.  As part of building those teams I needed to know how many games a player would play based on the progress of their team through the NFL playoffs. 

 The chart above shows the number of games each team plays.  The number 1 and 2 seeds with first playoff week byes can only play a total of three games if they make it to the playoffs, other teams can play four.  You can risk a four game player, perhaps from IND, BAL or DAL, but this year those teams are dominated by the top two in each conference.  Seattle really sticks out as a team that gets to play three games more than 50% of all simulations.

Tomorrow starts the playoffs, I will update the simulations as the teams get knocked out.