I think you are being typically obtuse.
80% from 1 book is not "80% of bets" So why didn't you entitle the thread "80% of bets at William Hill"?
Anyway, I went back and looked at the RLM bowl games:
Colorado State +4.5 -> +3 (L)
Duke +8 -> +7 (W)
Minnesota +5.5 -> +3 (L)
Boise St. +3.5 (some places 4)-> +2.5 (W)
Note some of the biggest line move favorites: Pitt, Washington, LSU, and Oklahoma (all games moved at least 2 points) lost outright. The people moving the lines got killed during this bowl season.
But by all means, hitch your wagon to this.
There is a difference between line movement and reverse line movement. The games you mentioned with regular line movement, I agree means nothing. Its reverse line movement thats significant.
Reverse line movement is a solid handicapping factor, as well as betting against the public. There are 2 "systems" regarding bet percentages and line movement that are significant. they are:
1) Play ON dogs of 3-14 that receive less that 35% of public action. 62-27 since 05/06, and 5-2 this year. Wins were OK ST, ND, Clemson, Nebraska, CMU. Losses were Texas and Col St.
2) Play ON dogs with Less than 35% of public action, AND the line moves 1 point or more (ex +5 to +4). 22-7 since 05/06, 2-0 this year. Boise St and Illinois St (includes playoffs from other divisions).
You can dispute the numbers, question the accuracy of the wagers, whatever. The numbers aren't locked up in the Pentagon. The books contribute the data to companies such as Sports Insights, and I'm sure there is some sort of compensation for the data. Its bet data, not the launch codes.