another way to approach this is to find out what the chance is that a random stock gets delisted (P). Then you can make an assumption about the loss in case of a delisting (L). Now your total profit becomes a function of P, L, the initial % of capital you're investing (C) and the expected profit (R) from a simple backtest. The math should be straightforward.
Yes. If your system is short term, ignore it. Making sure your test data includes all market conditions is way more important.