It would be better to trade what you test and test what you trade. Thus, if you could find a way to match your backtest with your execution or to match your execution with your backtest, that would be a better solution, and there’s two ways to do that. The first way is to automate so that it can run when you’re asleep, and it’ll just happen, that way there’s no slippage because what you’re talking about here is a drift in the price of one of the assets from when you get the signal to when you place the trade.
It’s the same in stocks. If you get the signal and you’re supposed to place the trade at the open, but you’re asleep and you place the trade in the afternoon, the price can move. In mean reversion, that could make a big difference. In trend following, it could make a difference, but it’s probably likely to be less than in mean reversion, which is the same logic. Therefore, you need to either automate so that your trades are placed at the appropriate time or you set up a database that time shifts so that you get your signals when you wake up. They’re not eight hours old, they’re now. That would be different to my signals because I’m trading the regular trading hours, the regular daily bar tick over. However, you could shift your entire database forward eight hours so that the daily bar ticks over whenever it is that you do your analysis, or just before that.