There is a book by Kevin Davey, “Building Winning Algorithmic Trading Systems” - it talks mostly about analyzing automated strategies.
The author states clearly that all automated strategies eventually stop working and need to be retired. Thus - a trader has to develop criteria on when to stop running a particular strategy.
Kevin Davey suggests using walkforward analysis, Monte Carlo analysis and monkey analysis.
According to him, walkforward analysis results can forecast the real trading results reasonably well.
Monte Carlo analysis can tell statistics about the expected behavior of the system (probabilities of various drawdowns, earnings, being profitable in a month, etc.)
Monte Carlo + monkey analyses are used to determine when a strategy does not work anymore (goes beyond its statistical boundaries) and needs to be retired.
Personally - I like how this testing process rips to shreds most (but not all) of my strategies. It’s better this way than the market does it. I have ‘retired’ one XAUUSD strategy last year but I am still hesitating if the boundaries I put were too restrictive.