COMING SOON – Siena at Georgia Tech (December 2)

Happy Thanksgiving to all!

The Siena at Georgia Tech prediction will be the next game predicted by the formula.

Delaware at Siena (November 24)

Predicted winner:
Siena by 31pts

(Raw score:  Siena 95.413 over Delaware 64.587)

Accuracy at 31pts plus: 100%

Notes:  Each team has only played 3 games thus far in the season, so the numbers used for calculation are only a very small sample. Also, the adjustments for each team for Last 10 Games and Road/Home Bonus take into account games from the previous season (as irrelevant as that may be, the adjustments call for a larger sample than 3 games).

Pittsburgh vs. Texas (November 24)

Predicted winner:
Texas by 11pts

(Raw score:  Texas 70.6 over Pittsburgh 59.4)

Accuracy at 11pts plus: 90.972%

Notes: Pittsburgh only has 4 games thus far in the season and Texas has only played 3 games, so the numbers used for calculation are only a very small sample. Also, the adjustments for each team for Last 10 Games take into account games from the previous season (as irrelevant as that may be, the adjustments call for a sample of 10 games). Also there is no adjustment for Road/Home Bonus as the game is on a neutral court.

Wichita State vs. Pittsburgh (November 23)

Predicted winner:
Pittsburgh by 10pts

(Raw score:  Pittsburgh 71.379 over Wichita State 61.621)

Accuracy at 10pts plus: 89.024%

Notes: Pittsburgh only has 3 games thus far in the season and Wichita State has only played 2 games, so the numbers used for calculation are only a very small sample. Also, the adjustments for each team for Last 10 Games take into account games from the previous season (as irrelevant as that may be, the adjustments call for a larger sample than 3 games). Also there is no adjustment for Road/Home Bonus as the game is on a neutral court.

Ohio State vs. California (November 20)

Predicted winner:
Ohio State by 4pts (Raw score; Ohio State 69.168 over California 64.832)

Accuracy at 4pts plus: 80.685%

Notes: Each team only has 3 games thus far in the season, so the numbers used for calculation are only a very small sample. Also, the adjustments for each team for Last 10 Games take into account the last 7 games from the previous season (as irrelevant as that may be, the adjustments call for a larger sample than 3 games). Also there is no adjustment for Road/Home Bonus as the game is on a neutral court.

Charlotte at Duke (November 17)

Predicted winner:
Duke by 3pts (Raw score; Duke 69.666 over Charlotte 66.334)

Accuracy at 3pts plus: 78.742%

Notes: Each team only has 2 games thus far in the season, so the numbers used for calculation are only a very small sample. Also, the adjustments for each team for Road/Home Bonus and Last 10 Games take into account the last 8 games from the previous season (as irrelevant as that may be, the adjustments call for a larger sample than 2 games).

Teams Added to Watch List

The following teams have been added to the early season watch-list for games for the formula to predict:

• Ball State
• California
• Duke
• Michigan
• Mississippi State
• Nevada
• Ohio State
• Oregon State
• Portland
• UNLV
• UTEP
• Western Kentucky
• West Virginia
• Xavier

Wichita State at Pitt (using 2008-2009 stats)

Predicted margin of victory:
Pitt by 17 points

Not that the teams are similar to last season as Pitt lost 3 great college basketball players (DeJuan Blair, Sam Young, and Levance Fields), but I was just curious to see a number.

First game prediction coming

Monday November 23, 2009

Wichita State at Pitt

Methodology

Using statistics pulled from http://kenpom.com/rate.php

My formula calculates various expected outcomes including;

• Possessions
• 3 point fg %
• 2 point fg %
• Turnovers
• Blocked shots
• 3 point fg attempted / Made
• Free throws attempted / Made
• Offensive rebounds

A raw point total is calculated for each team. From this, adjustments are made for the following;

• Home court advantage
• How each team has played in the past 5 games
• Bench play of each team
• Experience of each team
• Strength of schedule for each team

The formula predicts an expected margin of victory.

The higher the predicted margin of victory, the more accurate the formula has been.

When predicting a 1 point margin or higher, the formula has picked 74.2% of these games correct.

When predicting a 5 point margin or higher, the formula has picked 83.0% of these games correct.

When predicting a 10 point margin or higher, the formula has picked 89.0% of these games correct.

When predicting a 20 point margin or higher, the formula has never lost a game going 21/21.