My games today....

You are essentially correct.

What I practice is somewhere between your post and doryj's response.  First thing I do is make a line based on perceived results - not betting patterns.  Then I bet games that I can get a price that is beyond certain thresholds.

Then, I will bet more if the line gets better, and sometimes sell if it goes the other way.  The selling is based on thresholds as well - and those thresholds change upon how I am doing and how my statistical confidence level is at that time.

So sometimes - it is as Doryj states.  If the line I determined on the Phillies was -130, I would buy at +144 or better and sell if it hit -130.  This is if my line was highly correlated to results for a certain time period.  If my lines had been crappy (not highly correlated with actual results), my thresholds for selling become higher because of the implied variance in my line making ability.

So there are times when I would sell that bet I made at +144 at -136, even if my line was -130.

Your position is correct though - assuming that my line making had zero variance.  As the variance on my ability to set those line increases, it becomes a better longterm profitability model to take the money and sell the bet.

I think it comes down to a little bit of math, and a little bit of personal preference.  I have a relatively low tolerance for risk overall believe it or not.  I do understand your point - and it cant be argued - except the one factor of the accuracy of the pregame line to the actual results - that is a key determinant of this decision.

Price alone.

Nothing about the game itself.