In a phrase, getting into a trade early. It's what I've devoted heaven knows how many hours figuring out NL derivatives for, and I'm still refining. As James Goldsmith said, "If you see a bandwagon, it's too late".
Does the time frame used also affect variance? For example, is someone who only trades ACD intraday and always exits prior to the close, at a significant disadvantage to say the longer term trader?
The shorter the time frame, the lower you need your variance to be. Unfortunately, shorter time frames have the most variance.
Do you know anyone who has successfully modeled the volatility of day trading? I'm discounting HFT models that rely on front running or analysing order flow. I'm talking of the volatility/noise, and how it can be traded as a direct model.
So does that mean that an intraday ACD trader should look for ways s/he can reduce variance by doing such things as: - using a 5 min OR rather than 30 min OR - trading ETH hours rather then RTH hours - avoiding London open if trading Forex
No, it means you better be spot on with whatever the hell you are doing. If I'm trading crude oil for a 3 week hold I can be off on my entry by 200 or 300 ticks. If I'm daytrading oil I have no such luxury. I better nail that entry and hope it moves NOW.
The better your quant skills and the more granular the data the better. There are signals out there but there are billions of bits of false data you have to sift through to extract that signal. HFT firms are very good at going through the rubble and finding literally that one tiny needle in a haystack. Everyone else is just finding hay.