Truly. There's some standard approaches like beta-adjusting that can be done to put things on equal footing in different market regimes but it's only one part of what determines a stock's performance.

Testing for the significance of someone's average returns for example often can be rules of thumb. For example, given a return stream of a trading strategy, the Sharpe ratio (mean return/vol) is directly related to the t-stat of the mean return being different than 0 (mean ret / (vol/sqrt(n)). Several simplifying assumptions must be made about the distributions of returns that don't line up with reality to do the above but it's mostly a heuristic that will let you say "ah, the Sharpe is X over the last 5 years? That's significantly different enough than 0 for it to be tradeable."