When you evaluate a non-discretionary trading system, performance data such as profit to loss ratio, average profit per trade, % of trades profitable etc. can be calculated with a math formula. But, these performance points are worthless if you cant determine maximum contract/share size before slippage cuts into profitability. For trading systems with results from a variety of securities, this isn't practical. With data from trading a single security, is it possible to use volume or market depth at the time of trade to create a scalability metric that provides size traded/average expected slippage? for example: up to 2k shares = X< slippage, each additional 1k add X slippage.
For measuring slippage, you had to place orders with your broker and then store the resulting slippage delay in a 2D matrix with elements orders by your trade size and by the overall trade volume. This slippage matrix gives you an estimate of the expected slippage dependent on those two parameters. Then there's the network latency that's independent on the trade size. It depends on your Internet connection and whether the broker uses an UDP protocol or a TCP/IP protocol. For UDP, mimimum slippage is in the 0.1 to 0.2 seconds range, for TCP/IP it can be up to 0.5 seconds.
It depends on your time-frame but when holding for minutes or more, one could also look at the volume on a specified (low period) time bar. Slippage would be a function of the volume on that bar. While your solution was more precise, you need to take into consideration that when doing size, HFT will trade around you - you're not invisible, that's why you can't theorize much.