Testing has to be done with small percentages because we do modifications a lot during development phase. It takes minimum 6 months to finalize a strategy and also, people who are into algo trading, we all know that bakctesting results are only a estimation of how well a bot can do but not the real expected amount. We consider, sharpe ratio, win rate, drawdown and many other factors apart from returns alone. I have explained it in many of my previous articles also. Everything takes time and need patience to make good/decent profits. I have given an explanation of how the code is able to generate high returns. And during testing, sometimes bot doesn't give profits and make losses, based on the behaviour of the algo, we modify, keeping high stake amounts and running a testing phase is stupidity as far as I consider (after being in this industry for past 3 years and developed a lot of succesfull strategies, my articles and my patreon page are the proof for it) . I hope this answer helps you further discover related to algo trading psychology and development