how many sit n gos do you need to play before you can see if you are a winning player i.e. the long run?ie. with cash games its debatable but around 500 000.
Agree you need a whole lot to have sample size that's relevant. I'd say if you are winning over 500 games you're a winning player, but your ROI will still fluctuate with variance.
Problem is if you're winning, you'll probably be moving up in stakes before you have a large enough sample size to have accurate numbers (of course having the bankroll to move up is a pretty nice problem to have). You really want a large sample size at a single stake to get an accurate picture.
I couldn't really tell you without guessing, but probably 5% would be a good figure to aim for. I don't play HU SNGs at all though, so don't take that figure as anything but a random guess .
From what i hear heads up has a great deal of variance involved and like a 60%win to 40%lose ratio is as good as you can realistically hope to see over a large sample. I have seen very good players at mid to high stakes with only a 56% win to 44% loss rate.