I use standard deviation to measure how far my calculated line is away from my book's line. I try to build models that have a mean of 0 in terms of price difference. I will then calculate the standard deviation of the population. Anything above or below a certain STD level should give you your value plays.
Ex. Line diff = 4.5 points. Standard dev of population = 3% This play would be 1.5 deviations from the mean.
Read Conquering Risk to get you started on models. Don't know your background, but you need a solid skill set of statistics, programming, and work ethic to create these models. And most are going to suck.
The population refers to your entire set of entities where you draw your inferences from. In this case, every piece of data is a game. I do a lot of my modeling in Excel. It's easy to calculate standard deviation using the formula functions.
The STD rate is just how many deviations away from the mean you are in any given game. Theoretically, a working model will compute better plays the farther away you from your mean.
If your mean is 0, you should have a higher expected edge on games where your pricing difference is 5 points compared to 3 points.