Why your backtest is probably lying to you

A backtest that shows +400% will get you to fund the strategy. It will also, very often, lose you money. The gap between those two facts is where most retail traders blow up.

Here are the three ways a backtest lies — and how the assistant on this site is built to not.

1. Look-ahead bias

The strategy "knew" today's close when it decided to buy today. Tiny bug, enormous fake returns. The fix is boring: decide on yesterday's data, execute on the next bar. (Every backtest here uses next-day execution — no peeking.)

2. Survivorship & cherry-picking

Test on NVDA since 2022 and of course it looks great — NVDA went up. The strategy didn't do that; the stock did. The honest question isn't "did it make money," it's "did it beat just buying and holding?" — net of risk.

3. Hiding the drawdown

A 30% CAGR with a -60% max drawdown is not a 30% strategy — it's a strategy most humans would abandon at the bottom. Return without drawdown is marketing, not analysis.

The test that survives

A strategy is only interesting if, versus buy & hold, it does one of two things: comparable return at lower drawdown, or higher return at comparable risk. Anything else is noise dressed up as edge.

That's why every backtest here shows the strategy and buy & hold, side by side, with Sharpe and max drawdown — including the (frequent) cases where the simple thing wins. Honest beats impressive.

Go run one and see for yourself.

Educational, historical analysis only — not investment advice. Past performance does not predict future results.