I remember a system on C2 that was designed in Wealth Lab Developer. Basically the vendor had significantly overestimated profitability and underestimated drawdown because the time frame allowed a buy and a sell and a buy in the same bar. This can't happen, and only on the close will you even be sure within 1 or 2 ticks what the real curve looks like. I think he (op) got a 10 minute ascii dat file export, popped it into excel, and didn't know how to differentiate between buy/sellatstop and buy/sellatlimit/sell/buymoc, which he didn't use MOC so I'd be highly suspicious of the backtest anyway. The 3.5 avg win loss/ratio is probably also too high, but would put the profit factor around 5-6.