I will tell you this because I am currently doing this and so far the results are good. Go out and find yourself an EA that has a lot of features and settings, it doesn't have to be the best forex robot out there. Once you have your free robot, load it onto your rig and start testing it again.
This is where a robot with a lot of features pays off. Chances are it won't make a good profit right off the bat, so you'll have to change some settings and see what happens. Do some back testing to see how the robot reacts to the various changed parameters.
Once you understand your bot, you will want to optimize it. You might have already made a profit with simple back testing and if you are doing that, it's great. Go to settings and check all the settings you want to optimize. You are going to have to define a start and stop for the range of anything you want to be tested.
If you are trying to optimize a lot at once it may take more than a few days to run all scenarios. It's like that on my computer anyway. It seems like the more scenarios you give, the better the end results will be, which sucks because it's the slow lane.
Once you have a robot with a low circulation and are making steady profits from back testing, it's time to put it on a demo account and do some forward testing. I am currently working on a free forex robot that uses RSI, Heiken-Ashi, and Stochastics to enter trades. It's been going well so far, but needs a lot more work.
Remember everyone, the best Forex robots are free Forex robots! Good trading.