A backtest is the easiest thing in the world to make look great. Add a few rules, tune a couple of parameters, and any random noise starts to look like a money machine. Then you go live and it dies. This is overfitting, and it kills the vast majority of strategies. Here are five signs yours is fooling you.
Real edges are lumpy — they win and lose in streaks. A backtest that goes up in a near-straight line usually means the rules were bent to fit every wiggle of the historical data. Smoothness is not skill; it's a red flag.
Change your moving-average length from 20 to 22, or your stop from 2% to 2.3%. If the returns fall apart, the strategy was balanced on a knife's edge of one specific past. A robust edge survives small parameter changes.
Split your data into chunks and test each separately (this is the idea behind cross-validation). If the strategy is brilliant in 2019–2021 and flat everywhere else, you didn't find an edge — you found a period.
If you tested 50 variations and shipped the top one, its great numbers are partly luck. The more you try, the higher the bar its Sharpe must clear to be real. This is the multiple-testing penalty — and almost nobody applies it.
Spreads, slippage, and fees quietly eat returns. A strategy with a paper Sharpe of 1.5 can drop below zero once realistic costs are subtracted. If your backtest ignores costs, it isn't measuring anything tradeable.
You don't need to reveal your logic, code, or parameters to test whether an edge is real — you only need the per-period returns it produced (e.g. 0.012, -0.004, 0.008 …). From those numbers alone you can apply purged cross-validation, a deflated Sharpe ratio, a multiple-testing penalty, and cost/tail adjustments — and get a straight verdict: real edge, self-deception, or dead.
That's exactly what isitalpha does. You paste the returns, we do the statistics, and you keep your strategy to yourself.
Free verdict. You share only the numbers — never your logic, code, or account.
Get an honest verdict →We ran this on thousands of strategies. Most of them didn't survive. The honest ones are worth trading — but you have to know which is which before you risk capital, not after.