macrocephalopod

But also, do not backtest with open and close prices since you cannot reliably transact at these levels.

In particular because the only reasonable explanation for the intraday/overnight anomaly in SPY is a systemic one-way distortion of either open or closing prices (or both!)

The first one seems plausible (and testable — look at direction of trade flow in overnight sessions). The second one would not persistently move the price in one direction, would it?

I would still point to open/close distortions as the most likely factor. The close to close return on S&P 500 futures is more volatile than literally any other 24 hour return you could choose (by a *lot* - 17% vs 15.5% in the sample I looked at) which indicates big distortions.