Quote from sf631:
At a high level, sure.
I look for structural flaws and irrationality in the market (or with other market participants) and take the other side of their trade. I find it easier to identify others' mistakes than to come up with brilliance in a vacuum. The ideas come from lots of reading of sites like ET or Seeking Alpha, or others. You probably already know that there are many, many bad ideas out there, some of which can be traded against.
I then build a system that models that behavior, keeping it very simple and always building a system around a hypothesis, not randomly changing parameters until some combination works well. Curve fitting is a huge, huge danger that must be avoided.
If you can't explain a plausible case of (1) why it works (2) who is taking the other side of your trade, and (3) what they don't understand you're just heading for trouble IMHO. Try explaining these things to your dad (or anyone else who will hear you out). If you can do this (and really believe it) there's a better chance that you'll stick with it.
Taking your Yahoo seasonality example, start with a theory about others' behavior, like a tax effect, or portfolio manager window dressing at the end of quarters, or whatever, *then* build a model to test that belief and trade the other side of it. If you "data mine", you're more likely to find a spurious pattern.
Keep in mind that I differ from most others on this thread in that I'm not day trading, not trading futures, etc...
Feel free to PM if you want to discuss further
When are you going to do the high level?
