As a general rule, are simpler strategies better than complicated strategies? My reasoning is that the more complicated your strategy is, the more likely your returns are the result of data mining (or whatever it's called when you unrealistically perfect strategies for backtesting). Also, are strategies that work over long periods better than ones that only work over short periods? You could rationalize that the longer you test, the more different the markets were back then and therefore the results aren't germane to today's market. So...would the ideal system be simple and work over long periods of time?