Buddy did a good job of answering the first question. The confusing part is why the recommendation on the pick goes away even if the line is moving in your favor. That's entirely because of the "betting window" in every formula as Buddy noted - if the betting window was opened wider the recommendation wouldn't go away, but the formula would recommend more picks than before and might become less accurate.
One thing worth noting is that the line on a game is often fairly close to the actual outcome. Computer software that beats the line has to generate point spreads that are fairly close to the line. For example, let's say that you were capping the Colts game on Thursday night and you came up with a computer formula that favored Jacksonville by 40 points, even though the Colts are unbeaten and favored by 3. The tendency would be to distrust that formula because it is unrealistic. You instantly know that there is no way that Jacksonville is going to win by 40 points. Thus, instead of thinking that the formula has uncovered some hidden piece of information that that you need to bet the house, you're are probably going to the think that the formula is broken and that you shouldn't trust it at all. However, if the formula picks the Colts by 1, then you know that it's calculations are doing a pretty good job of approximating the actual line. In that case, Jacksonville +3 might seem like a safe bet. All you need is a slight edge to win.
Regarding your second question, if the home team is favored, enter a positive number for the line. If the visiting team is favored, enter a negative number for the line. Keep in mind that you don't have to enter lines unless you are keeping your own line folders (http://www.microbro.com/manual/sec6-1.htm), you can simply download lines by clicking on the globe icon.