Quote from piezoe:
What i look for is the slope of the regression line on the equity curve, after costs, to be positive, and a standard deviation about the regression line ideally no larger than the magnitude of the slope. Positive slope required for profitability, low variance for consistency. "Edge" is directly proportional to slope and inversely to the standard deviation about the regression line, hence a quantitative measure of edge is: e = m/s.
I would add the number of trades to this equation:
e = sqrt(trades) * m / s
since (s / sqrt(trades)) is known as the "standard error", the original equation can be rewritten as:
e = m / standardError
This may also be recognized by some as Van Tharp's "System Quality Number".
