From your posts, it appears that you are comfortable in your skin. If you aren't already profitable you are just 3 clicks of your heels away.
(1) You can try to imagine them (2) You can try to read academic literature if you can make any sense of it (3) Make a list of indicators and throw darts on them (often works) (4) Write code to have the computer do it for you (hedge funds/quants) (1) works for people with charisma. (2) works for people who know math and statistics at the graduate level. (3) works for the lucky ones. (4) is about data-mining and it is the best way (according to Aronson) but the toughest to setup correctly and analyze the results. There are some expensive, some less expensive and even free (with expensive support) programs to do the job. Look for Trading System Lab, Price Action Lab, TSSB, Strategy Quant, for example. Just beware of the pitfalls of data-mining that are not limited to data-mining software only. Even if you are looking for academic edges your search is subject to data-mining bias because someone already tested the edge. Your question does not have an easy answer. It's like asking for a way to find happiness. Too abstract and too general.
I get most of my ideas from Gurus, though they never really spell out their recipe so I have to reverse engineer their trades. I've even got an idea or two from some ET educators, though they didn't work in live testing. My best trading system came about by accident after several ideas were tried together by accident ("not following the rules" of one of the other ideas) :eek:
Another important rule is to make sure you do not "choke" your algo. Do not make too tight of conditions otherwise it will not work in most market conditions.
Lets be fair and say, your reversed engineered assumption of their "recipe" "didn't work in live testing".
Thanks for the link. More than anything it helped organize what i already knew. From your posts you come off as purely a price action trader. Just out of curiosity: Are you a discretionary trader or do you use some automation (finding your patterns, placing orders)? And since you say you got done with testing a long time ago, how have your patterns worked for you in their current state?