"It's been working" is not evidence yet. How long is?

Honest question rather than a lecture. People here. Me included. Will run something for three weeks, see green, and start sizing up. Three weeks of a strategy that takes two trades a day is thirty trades. That's nowhere near enough to distinguish a real edge from a decent run, especially with a skewed payoff. So what do you actually use as your bar before you trust it? A trade count, a time period, a drawdown you've survived? I'd rather adopt someone's tested rule than invent my own badly.