From this paper: http://www.sciencedirect.com/science/article/pii/S1059056016301563
It appears to be possible to predict the overnight SPY and the first 30 minutes of trading (with a negative correlation with the overnight ETF), based on the last 30 minutes of trading of the day before.
This is my backtest in which I try to implement this strategy, what do you think? is there a way to improve it?
Thanks a lot!