Kojak-
Thanks for posting your tracking info. For a system that's supposed to hit above 70% of the time, I decided it was worth spending the time to research myself.
You said you were using Don Best's opening numbers, which as I'm sure you know are archived and available. I just downloaded the entire 2005 season to check your tracking and am not sure how you compiled that list of games as ones that Don Best shows as opening at 40.5.
For instance, you list the Seattle/Houston game... well that game actually opened at 46, not 40.5
http://www.donbest.com/website/html/freeodds/archives/nfl.shtml?20051016
Another example, you list the Tenn/SF game... that actually opened at 42.5, not 40.5
http://www.donbest.com/website/html/freeodds/archives/nfl.shtml?20051127
Of the 22 games you listed for 2005, I only see 2 that actually opened at 40.5, according to Don Best's archive.
I did find 9 other games in 2005 that opened at 40.5 that you did not mention, for a total of 11 games in 2005 that opened at exactly 40.5. In those 11 games, the over hit 5 times, and the under 6.
While I was at it, I checked for 37.5. In 2005 there were 10 games that opened at 37.5 and 4 games went over and 6 under.
Looking at all opening totals for all of 2005, for numbers where there were at least 10 possible plays, the best bet would have been to bet the Under when Don Best's opening line was 45. In these plays, the under hit 10 out of 15 times. That's 67%, but still with a pretty small sample set of only 15 games.. I'd want to check a lot more years to see if that was a real trend or not. The opening number you could have played the most was 44 and that went over 13 times, and under 11.
Again, I'm using the online archive from Don Best. If their archive is accurate, I believe your tracking to be inaccurate (at least for 2005... I haven't gone back further).
I'd be happy to post the details of my findings if anyone is interested and can also tell me how to insert an Excel table into a post.