A strong backtest is interesting, but it is not proof of a strong trading strategy. Working on systematic equity research has made me pay more attention to the assumptions behind a result than to the headline return itself.
A good-looking equity curve can mislead
When I first began testing mean-reversion and momentum ideas, the exciting part was seeing how a simple rule could turn price data into decisions. The harder part was learning that seemingly small modeling choices could change the answer.
Execution timing, transaction costs, security selection, and capital allocation all matter. A model that buys at a price it could not actually obtain, for instance, is not describing an investable result. It is describing an optimistic simulation.
Treat the research process as part of the result
One of my projects tested hundreds of parameter combinations. That exercise made overfitting feel much more concrete: after enough experiments, some configuration will look excellent by chance. A test that survives data it never saw is more persuasive.
My checklist now starts with a clear hypothesis, realistic execution assumptions, out-of-sample testing, and a closer look at drawdown and turnover. I also compare the strategy against a simple benchmark instead of assuming that a positive return is enough.
What I would tell a new researcher
Start with a rule you can explain in one sentence. Document every important assumption before tuning the parameters. If the findings change dramatically when one number moves a little, treat that sensitivity as a research question rather than a victory.
The CFA Institute’s overview of backtesting and simulation reinforces why historical tests, stress scenarios, and sensitivity analysis should be used together. For me, the central lesson is simple: good research is designed to find weaknesses, not merely attractive returns.
