Wednesday, January 11, 2012

Superbowl 46 Playoff Roster predicted winners after Wild Card week

Mostly the expected happened during NFL Wild Card week, HOU beat CIN, NO beat DET (but was interesting at the start), and NYG crushingly beat ATL.  Less expected was DEN beat PIT in an extremely exciting game.  This sets the outcomes for the RKB Fantasy Playoff roster predictions a little more firmly, though four of the teams haven't played yet. 

Like previous years, we can still model the performance of the list of 52 roster using the current points scored and plugging in the average points/game from the season multiplied by the number of games the simulations says a given team will play.  Add it up and compare to the other rosters to see which rosters make it to 1st or 2nd place, which is in the money.  There are obvious problems with the model, since it uses averages and there can be a wide variation in the performance of a given player in a given game.  But it points the way toward a prediction of the outcome.

Firstly, below is a plot of the outcomes of rosters in 1st and second place from 10,000 simulations of the rest of the games of the playoffs.

Each column represents the frequency of that given result.  The numbers are the number of the roster.  Green rosters have me in the money.  The darker green rosters are results in which I actually come in 1st and second place! As it stands I have a 54% chance of being in the money (1st or 2nd), a 47% chance of being in 1st, a 20% chance of being in 2nd, and 13% chance of being in 1st and 2nd.  Most of these centered around NE or GB making it to the Superbowl, with a little help from Arian Foster of HOU along the way.

We can take the bars above and break up the results by the final Superbowl team matchups.  Thus we can determine which combos result in wins for me and which do not.

The sizes of the entire pies above represent the relative probability of the various matchups in the Superbowl.  For the purposes of this model prediction the winner doesn't matter as much as the number of times a team plays.  Each pie is then broken down into slices for each of the 1st and 2nd place result combinations.  Light green pies are when I end in first place, dark green are when I end in 1st and 2nd place, and the lighter yellowish green are when a roster of mine ends in 2nd place.  The orange and blue colors represent other rosters of interest, for my sister and brother-in-law respectively.

It is in my best interest that NE and GB meet in the Superbowl, and my roster somehow overcome the all GB and all NE rosters of some of my competitors, I think this is due to spreading the players across teams, and the hopefully luck addition of Arian Foster to the lineup.  I can also be successful if NE plays SF and HOU plays GB, while NO vs. NE or BAL vs. GB might yield a first and SF  vs, any three yield 2nd place finishes.  Under no circumstances do I want to see NYG or DEN make it to the Superbowl.

I know who I need to root for this weekend, how about you?


Howard said...

Sorry, the Giants are beating Green Bay this weekend

Richard said...

That will completely screw up my fantasy playoff chances! Also if it was up to me the NFC East would have forfeited their chance to have any team in the playoffs, since they seemed to want to see who could lose the least instead of who could win the most.

Howard said...

Sorry (kinda)

Richard said...

Brutal game. I went with the high probability. I take solace in the fact that it doesn't mean that my model didn't predict the results, it's just I failed to cover what I felt were the less likely possibilities.

Howard said...

I should have bet. I picked Giants v Patriots early on :)

Superbowl2012 said...

Excellent pieces. Keep posting such kind of information on your blog. I really impressed by your blog.
NFL Schedule| Super Bowl Commercials 2012