Guides
Backtesting guides: build, validate and correct a strategy
Plain-English guides to building a trading strategy, proving the edge is real and fixing what breaks — Walk-Forward, Monte-Carlo, overfitting and more.
What is overfitting?
Overfitting is why most backtests lie. Learn what overfitting is, how to spot it, and the four tests that prove an edge is real: out-of-sample, Walk-Forward, Monte-Carlo and the Deflated Sharpe ratio.
How to backtest a strategy
A step-by-step guide to backtesting a trading strategy honestly — with no code. Define rules, use realistic costs and slippage, read the right metrics, and validate out-of-sample so the result isn't overfit.
Strategy walkthrough
A full end-to-end walkthrough of the Strategy workbench: brief your AI research desk in plain English, edit the block graph, run an honest backtest and get an overfitting verdict — no code.
Portfolio walkthrough
A full end-to-end walkthrough of the Portfolio Lab: start from a preset or plain English, set contributions and rebalancing, simulate on shared capital and optimize the weights without overfitting.
Shared-capital portfolio
Most portfolio backtests blend independent equity curves — each holding behaves as if it had its own money. A shared-capital book runs one account where sleeves compete bar by bar, with cash redistribution and per-holding ceilings. Learn the difference and when to use it.
Multi-asset backtesting
A strategy that works on one ticker is an anecdote — you found the ticker where it worked. Run the same rules across up to fifty instruments in one pass, read the aggregate and the per-asset spread, and learn why profit factor is a median and per-asset tuning is overfitting.
Walk-Forward analysis
Walk-Forward analysis re-optimizes on a rolling in-sample window and trades the next unseen segment, stitching a 100% out-of-sample equity curve. Learn how it works, anchored vs rolling, and how to read the result.
Monte-Carlo simulation
Monte-Carlo simulation reshuffles your trades hundreds or thousands of times to estimate how much of a backtest was luck — the probability of profit, plausible drawdowns and risk of ruin. Learn how to read it.
Deflated Sharpe ratio
Deflated Sharpe ratio and PSR explained: how testing many strategies inflates the best Sharpe, and how deflating for the number of trials, skew, kurtosis and sample length gives the honest number.
Adaptive position sizing
Adaptive position sizing scales your trade size with a signal — VIX, volatility, trend strength — instead of a fixed bet. Learn the four response-curve shapes, why it usually overfits, and how to test that the curve holds out-of-sample.
Custom indicators
Build a custom indicator once and reuse it in any strategy. AI Import reconstructs non-standard indicators exactly (Linear Regression / LSMA, LazyBear's Squeeze Momentum) and saves them to your library — and each one shows which strategies use it.
The Strategy Score
The Strategy Score explained: a free 0–100 blend of a strategy's realized metrics for fast triage — and why you must pair it with the Overfitting Polygraph verdict before trusting it.