Thanks. Yeah doing it by the quarter gives a misleading idea.
Suppose we look for traders that are profitable in every quarter. Say 33% are profitable there, or one-third in that quarter's data. That means (assuming independence - yes I realise it's not really independent), 1/9 will be profitable for two quarters, 1/27 for for 3 quarters, 1/81, i.e. 1.2% will be profitabl in all 4 quarters. Things are not independent, so the figure will be higher than that, but this figure seems much closer to the reality than the 33% figure. And if we talk about profitability over a few years, then 99% failure might not be unreasonable
The way I see it, being profitable for over a 3 month period is enough time to exclude\filter most winning streaks. The data speaks for itself really, any assumptions about it without further evidence is hearsay.