This document discusses automated approaches for selecting robust systematic trading strategies. It presents several techniques: system parameter permutation to evaluate all possible outcomes and address overfitting; domain regression to select strategies where train and test performance are correlated; step-down correlation minimization to choose uncorrelated strategies; and bootstrap sampling with random test sets instead of fixed periods. The goal is to estimate out-of-sample performance using all available backtest data through techniques like system parameter permutation, domain regression, and estimating profit and loss shortfall.