It's possible to do, however once you click that Attack button to start searching, your troop number is committed, which means whatever number of troops you had at that point will be used in the subsequent battle once you find a good opponent. I agree though that most of the time I don't need the entire army to make profits. One practical suggestion that works for me is put some bulky troops in the camp, troops you don't use for the bot. Like a couple-three of Healers, a couple of dragons or a bunch of giants. Then change your camp capacity number to the number of remaining available slots, and voila! - you have a decreased camp, which decreases your time between loots. I have not ran projections yet plotting the trade-off in number of troops versus the frequency of raids versus profit, but it's underway. In fact, I have been working on getting all that data saved programmatically and using as metrics to determine baselines and make intelligent tweaks to methods. In a few words: the numbers of the search attempts, G/E/DE/Tr gained/lost, number and type of troops used are inserted into a database (local CSV file, MS Access, or a remote SQL server) and the metrics developed on any frontend reporting services. Could just dump stuff into an XML file to be displayed on a web page (local IIS or apache, or remote webserver) that you can access remotely to check on what's up with your hardworking bot. 
Possibilities are endless...
-- D
Possibilities are endless...-- D