Pro courseIntermediate12 minAI, Backtests and Bots
Understanding Backtesting
A backtest shows how a strategy would have done in the past. It is a useful research tool, but it is also the easiest place to fool yourself. This module teaches you how backtests are built, the traps that make them look too good, and why they must never be confused with a live track record.
- 1What is a backtest? In-sample, out-of-sample and walk-forwardA backtest replays a strategy on past prices. Learn why testing on unseen data and walk-forward testing make results more honest.3 min
- 2Overfitting: when a backtest is too good to be trueTrying many parameter combinations can make a useless strategy look brilliant. Learn how overfitting happens and how to spot it.3 min
- 3Lookahead bias, survivorship bias and hidden trading costsThree common reasons backtests look better than reality: using future data, testing only surviving coins, and ignoring fees, spread and slippage.3 min
- 4Backtest, paper and live: why the records must stay separateWhy results drop from backtest to paper to live trading, and why the three records must never be merged into one chart.3 min
Part of DUKOIN Pro
See plansEvery lesson's introduction is free to read. The full lessons, quiz and challenges come with Pro.
Education, not investment advice. Crypto prices can fall sharply and most retail traders lose money. Practice in DUKOIN uses virtual USD with no real value.