The idea that systems developed on historical data of a portfolio of securities have better chance of being non-random as compared to systems that are developed for a single security is based on the assumption that the size of the rule set used in the data-mining process is minimal . . .
Trader education, Trading Strategies
