A backtest is a tool for honesty, but it is just as easily a tool for self-deception. The myths below are persuasive because they all produce a chart that looks ready to trade. Each one is a way to be fooled by your own work.
Myth 1: a clean equity curve means a good strategy
A smooth, rising curve is the most seductive image in trading, and the easiest to manufacture by accident. Tune enough parameters against one stretch of history and you can fit almost any curve. The curve is not evidence; the way it was produced is. A clean line on tuned data proves nothing until it holds on data the strategy never saw.
Myth 2: costs are small enough to ignore
A backtest run without fees, spread, and slippage measures a market that does not exist. Friction is paid on every side of every trade, and for high-frequency or marginal strategies it is often the difference between a winner and a loser. A test that does not charge the full cost model is a flattering fiction.
Myth 3: one good run is proof
A single strong backtest is one sample. Markets are noisy, and a strategy can look excellent on one window by luck alone. The honest question is whether it holds up across many rolling out-of-sample windows — not whether it had one good day on paper.
Myth 4: more parameters make it smarter
Every extra tunable knob is another way to memorise noise rather than capture a real edge. More parameters usually mean a better fit to the past and a worse fit to the future. Simplicity is not a limitation here; it is a defence.
The cure is method, not optimism
The antidote to all four myths is the same: validate on unseen data, with full costs, across multiple folds, and treat a pass as permission to paper trade rather than to go live. Read look-ahead bias, the silent killer and how to spot overfitting in a strategy to harden your own testing. Backtests do not guarantee future results, and nothing here promises profit.