Wednesday, August 31, 2016

2016 Season Simulation: The Effects, for Conferences, of Game Locations and Less Than Full Round Robins

One of the things the 2016 Season Simulation has turned out to be good for is to illustrate, for conference standings, how conference game locations and conferences playing less than a full round robin can affect conference standings.  They can influence the standings so that a team that is better than another nevertheless ends up poorer in the conference standings.

The simulated 2016 standings for Virginia Tech and North Carolina, in the Atlantic Coast Conference, will show what I mean.  The ACC has 14 teams, but each team plays only 10 conference opponents for purposes of the conference regular season standings.

According to the Simulation, here are the 2016 ranks of the teams in the ACC:

1.5 Duke
1.5 Clemson
3.0 Florida State
4.0 Boston College
5.0 Virginia Tech
6.0 Virginia
7.0 North Carolina
8.0 Wake Forest
9.0 Louisville
10.0 Notre Dame
11.5 Syracuse
11.5 Miami FL
13.5 North Carolina State
13.5 Pittsburgh

Each rank has a rating assigned to it, based on the historic rating for that rank within the ACC.  When I apply those ratings to ACC conference games, after (1) adjusting the ratings for home field advantage, which is worth .0120, and (2) treating any game with an adjusted rating difference of 0.0134 or less as a tie (consistent with historic data), this year's conference competition ends up with Virginia Tech finishing #7 in the conference with 17 points, Virginia as #5 with 20 points, and North Carolina as #6 with 19 points.  The effect of this is that the three teams end up with different ACC conference tournament seeds than they would have had if the conference standings reflected the teams' simulated strength.

The question is, How could this happen?  With Virginia Tech and North Carolina as my examples, here are their simulated results:

Virginia Tech

v Virginia #6, away, tie,  Game location doesn't matter  1 point
v North Carolina #7, away, tie,  Game location changes a win to a tie  1 point
v Duke #1.5, home, loss, Game location doesn't matter
v Clemson #1.5, home, loss, Game location doesn't matter
v Syracuse #11.5, away, win, Game location doesn't matter  3 points
v Boston College #4, away, loss, Game location changes a tie to a loss
v Louisville #9, home, win, Game location doesn't matter  3 points
v Notre Dame #10, home, win,  Game location doesn't matter  3 points
v Miami FL, #11.5, away, win,  Game location doesn't matter  3 points
v Wake Forest #8, home, win,  Game location doesn't matter  3 points

If you count down the simulation ranks list, you'll see that the teams Virginia Tech does not play are #3, 13.5, and 13.5.

North Carolina

v North Carolina State #13.5, home, win, Game location doesn't matter  3 points
v Virginia Tech #5, home, tie, Game location changes a loss to a tie  1 point
v Clemson #1.5, away, loss, Game location doesn't matter
v Virginia #6, away, loss, Game location changes a tie to a loss
v Miami FL #11.5, home, win, Game location doesn't matter,  3 points
v Wake Forest #8, home, win, Game location doesn't matter,  3 points
v Notre Dame #10, away, win, Game location doesn't matter,  3 points
v Pittsburgh #13.5, away, win, Game location doesn't matter,  3 points
v Syracuse #11.5, away, win, Game location doesn't matter,  3 points
v Florida State #3, home, loss, Game location doesn't matter

North Carolina does not play the #1.5, 4, and 9 teams in the conference.

Thus, in relation to North Carolina, Virginia Tech is hurt by both game locations and the assignment of opponents.  Virginia Tech loses 3 points due to game locations whereas North Carolina comes out even.  Virginia Tech doesn't play the #3, 13.5, and 13.5 teams in the conference, whereas North Carolina doesn't play the #1.5, 4, and 9 teams.

The bottom line for conferences is that game locations and playing less than a full round robin can skew conference standings so that they may not reflect the actual strength of the conference's teams.

No comments:

Post a Comment