If you read my post yesterday, forget it. I slept on it and decided there was a less arbitrary, simpler, and more statistically reliable way to predict the probable outcomes of NCAA tournament games. I call it "Neutral Site Point Value over Average".
I went way back in NBA history to the days when they played true "neutral site" games (for example, the Lakers would play the Celtics in Michigan. I guess it was the days before TV contracts) and discovered that team's average neutral site performances usually ended up splitting the difference between their overall performance and their road performance. Relying on that, here is my method for predicting neutral site outcomes:
1. Average each team's Overall and Road offensive and defensive efficiencies. The product is the team's likely "Neutral Site" Offensive and Defensive efficiencies.
2. Get each team's "Opponent" Offensive and Defensive efficiencies. This is your schedule neutralizer.
3. Determine each team's "Point Value over Average" by subtracting the team's neutral site efficiencies from their opponent's opposite efficiency. (For instance, Wisconsin's neutral site offensive efficiency is 105.8. Their Opponent's defensive efficiency for the season is 95.4. Therefore, Bucky's offensive "Point Value over Average" for neutral site games is +10.4.)
3. Add both teams average number of possessions per game and divide by 2. The outcome is the likely number of possessions when the teams play.
4. Now use that possession number to play out each team's "average game". In other words, multiply each team's neutral site efficiency by the possession number Now you have an average "score" each team would expect on a neutral site given the number of possessions.
5. Take each team's "Point Value over Average" and apply it to the other team's "average game". (For instance, comparing Wisconsin to Georgetown. Wisconsin's neutral site PVOAs are +10.4 on offense and -19.3 on defense. Georgetown's are +12.2 and -14.1. Georgetown's average neutral site game score is projected to be Gtown 67.3, Opponent 58.4, and Wisconsin's is Wisco 65.8, Opponent 54.6. Therefore, if the two teams met, from Gtown's perspective the score would be Gtown 66.8, Wisco 51.7, and from Wisco's perspective the score would be Wisco 68.8, Gtown 48.)
6. Take each team's PVOA game score and average the results. That is your projected outcome. In the above example, the predicted outcome would be Wisconsin 60, Georgetown 57.
What it tells us
All this is telling us is the likely result of the game if each team plays their average game. And, isn't that about the best you can hope for? If you use this method, you are playing the probabilites.
I tested the method on tonight's meaningless "play-in" game between Coppin State and Mount Saint Mary's. The predicted score was Mount Saint Mary's 65, Coppin State 58. The actual score was MSM 69, C State 60, so it was close and it got the victor right.
New Tournament Picks
First Round (Picks in Bold; notes where appropriate)
1. North Carolina-MSM
2. Indiana 72 Arkansas 67
3. Notre Dame-George Mason
4. Washington State - Winthrop
4. St. Joes 70 Oklahoma 65
6. Louisville-Boise St
7. Butler-South Alabama
10. Kent St-UNLV
13. Kansas St. 64 USC 63
18. Texas AM
21. Purdue (barely)
26. Miss St
31 St Mary's (just barely)
Its getting late, so I'm just going to list the teams and explain them later. Some are very close calls.
Memphis...PItt...Stanford...Texas...UCLA...Drake...Xavier...Duke...NC...Wash St....Louisville... Tenn...Kansas...Clemson...Wisconsin...Gtown.
Final Four (Really Bold Picks...hey, I'm just going to let the numbers play out)
UCLA...Texas (only because I analyzed Texas' Elite Eight game as a home game)...Kansas...NC
Kansas 70, UCLA 66
Sorry about the sloppy post... I'm really out of gas with this stuff.