Try this. In simulated trading, write a trading system (yes I am suggesting algorithmic trading) that uses the random function to enter. Make it so that you are trying to enter as often as you would normally to fit your psychology, commissions tolerances, etc. Now, concentrate on the exit only using systematic scientific problem solving (indicators, tech analysis, moon cycles, neural nets, whatever). Don't try to make money, just try to break even after comission. As a control, enter randomly and exit randomly. How good or bad does that do on ANY symbol? What is its expectancy over many many trades? How does that compare to a slot machine? Did you do the math? Were you very exact in your analysis? Were you surprised? You will learn a ton. Write it down. When you have done that go live trying to break even in EXACTLY the same way, random entries, throw everything you got at exits. Come back after that. How did it compare to your simulation? Can you generate alpha this way?
I look at the chart and determine the weekly trend and then try to take up the best possible entry level in the shorter time frame charts.