After doing a lot of guessing, I finally figured out that rolling backtest results on screens calculate slippage by using the buy price without slippage and then subtracting the percent slippage from the selling price.
I don’t know if this is a good idea or not, but if anyone is wondering about how it’s done, there’s your answer. It’s not at all the way slippage is calculated in the regular backtests and the simulations, which subtract slippage from both buying and selling prices.