I used to track the lines and rate the teams according to how the line changed from week to week. And I noticed something interesting. Just because a good team lost wasn't necessarily a reason for them to change a team's rating. That got my gears turning. A few years later I designed this system using that theory. It is based on head to head competition. There is no exaggerated margin of victory factor and there is no schedule strength. They are natural in this system.
The actual rating is the top end of their expected ability. It is in general about twice the margin of the line. Therefore the expected range of the final outcome is 50-100% of the posted margin.