According to Deb Alterman, Mathematical Genius:
This approach incorporates both Neil's conviction that total number of wins should count for something and Matt's belief that the accuracy of your point spread is what matters most. To these considerations we have also added a way to have correct predictions count more early in the season, when we have less information available to use in making predictions. It also improves the evenline standard to the Fat Tony line. Which will be addressed next.
Both Neil's and Matt's calculations are based on the notion that each team is equally likely to win, so they are based around the evenline of a tie. This is not really true, if you picture a MUS/TBS matchup of last season. So Fat Tony plays the role of oddsmaker. He will make a prediction, based on previous season stats and some magic from J. Robert, about who will win and with what point spread.
Say Fat Tony predicts TBS will beat the MUS by 3.5 points. You can bet either above or below Fat Tony. If you bet above Fat Tony, you believe TBS will win by at least 4 points. If you bet below Fat Tony, you believe the MUS will do no worse than a 3 point loss.
This takes the place of Neil's prediction of win or loss—that would be the correct approach if Fat Tony predicted a tie every game.
How is this bet scored? You "put up" a certain number of points. For the first 10 games of the season you put up 2.5 points, games 11-20 you put up 2 points, games 21-30 you put up 1.5 points, games 31-40 you put up 1 point, and games 41-44 you put up .5 of a point.
If you beat Fat Tony's line, you double your bet. If you lose to Fat Tony, you lose the points you wagered. (The astute gambler here will note that Fat Tony does not collect anything as the house. Let's assume that either he doesn't have any operating costs, or his revenue comes strictly from internet advertising.)
Suppose that for Game 1, Fat Tony picks TBS over MUS by 3.5 points. Neil picks TBS over MUS by 6 points and Tedd picks MUS over TBS by 1 point. The final score is TBS over MUS by 10 points.
Because this is Game 1, Neil and Tedd have each wagered 2.5 points. Because Neil was right, he gets 5 points. Because Tedd was wrong, he loses his 2.5 points.
Current score: Neil 5, Tedd -2.5
You can stop here, by simply guessing to one side or the other of Fat Tony's line using the simple form on the home page. Or you can increase your wager by predicting the points spread (using the detailed form). These points are added much like Matt's approach, except we replace the evenline with the Fat Tony line.
In the example above, Neil was off from the final score by 4 points. Fat Tony was off by 6.5 points. Tedd was off by 11 points (he went 11 points in the wrong direction. We crafty mathematicians like to use negative numbers here, but that isn't really necessary.)
So Neil's pick was 2.5 points better than Fat Tony's. Tedd's pick was 5.5 points worse than Fat Tony's. So Neil gets 2.5 points and Tedd loses 5.5 points.
Current score: Neil 7.5, Tedd -8.
(Note: in the event of a 3 point MUS win, Tedd is 2 points off, Fat Tony is 6.5 points off, and Neil is 9 points off. Then Tedd picks up 4.5 points, and Neil loses 2.5 points. A TBS-MUS tie is similarly taken into account.)
Now to factor in the LOCKS: Let's say that when you predict a lock, you are twice as sure of your pick. In other words, you are wagering twice as much. In the case of a lock pick, your point total doubles. If both Neil and Tedd had picked locks for this game (and we'd really have to mock Tedd for that pick), their scores would double.
Current score: Neil 15, Tedd -16
The lock pick can also be used without the additional point spread wager. In that case, Neil and Tedd's basic scores, 5 and -2.5 respectively, would be doubled, leaving
Neil 10, Tedd -5.
Still confused? Buy Deb a beer at the next draft, and she'll explain it in person!