By the numbers. How would you apply this information?

[Duke]

Hello, I have been playing with different algorithms this season to try to predict NFL. My best one has the following characteristics from week 2 to week 6. It is only an interesting game of numbers to me, as I do not wager. The Vegas line is not even considered in my answers. I do all the handicapping from scratch. The facts below relate to the actual game results
Out of 72 games total
17 wrong teams were picked to win. When I am wrong I am really wrong, by about 15 points average.
There were 2 "no picks" because a predicted difference <1
In 53 games the winner was picked of which,
26 predictions were optimistic by avg. of 3.5 points 27 predictions were pessimistic by avg. of 3.25 points.
My question is how would you apply this information to a wager, assuming it is consistent? So far it has been more consistent than I would have expected. I appreciate constructive criticism. Thanks, Bill (Duke )

[Walter]

Can you simplify that, I'm not exactly sure what you're getting at.

[Duke]

Ok, let's look at it this way. Assume I am preparing to make a committment predicting the outcome of week 7 performance. Based on the data gathered from the previous weeks, the "Duke" system says that I can expect the following: 1. Approximately 24% of my picks will be completely wrong, with the opposite team winning. This is based on 17 out of 72 being wrong previously. 2. 76% of my picks will be the correct winning team, but that doesn't mean they can all generate a cash win. 3. Almost exactly half of the winning team picks will be by more than expected (pessimistic prediction) and almost half will be less than expected (optimistic prediction). 4. Finally I look at the Vegas line. If I select only teams that have a "Duke prediction" of more than 3.5 points over the line the half that fall into the pessimistic group are all winners. The half that fall in the optimistic group will only have half of them winners since they are averaged.
To sum it up 24% of my picks will be losers off the top. Half of the remaining 76% (38%) will be winners due to the pessimistic results. Half of the other 38% (19%) will be winners from the optimistic results. The total real winners therefore are 38 plus 19 = 57% cash winners of games played. From what I have read 55% is the magic number of making it worth doing. Of course all games cannot be played because they must meet the requirement of having the Duke prediction 3.5 points over the line. Does this make sense? Bill Duke