I wonder if anyone else has input on this. Assuming I'm using market orders, how large can they get before I experience more than the usual slippage? E.g. SPY traded approximately 15 billion USD today, how large percentage of that is reasonable for an intraday algo to issue a market order for at any given moment without moving market against us much more than usual?

Maybe it's best to test this case in IB's simulation by measuring slippage in their simulation versus my expected slippage? I have my doubts about the utility of this though as the market maker will not attempt to fill my order, nor would other market participants react.

Or should I simply not worry about this until the point where it actually starts to show off in the algorithm's equity curve? (At which presumably it would be wise to invest in an optimized execution platform.)