← ClaudeAtlas

walk-forward-checklisted

Check whether a trading rule's result is real or fitted to its own history with a walk-forward on QuanterLab. Use when the user asks whether a backtest is overfit, robust, luck or skill, whether it would hold out of sample, or right after a backtest that looked good.
quanterlab-ug/quanterlab-claude-plugin · ★ 0 · AI & Automation · score 72
Install: claude install-skill quanterlab-ug/quanterlab-claude-plugin
# Walk-forward check ## When QuanterLab is not connected yet This skill works through QuanterLab's connector (`https://quanterlab.com/mcp`). If none of its tools is available in this conversation (no `get_started`, no `search`, no `start_test`), the person has the plugin but has not signed in to QuanterLab yet. Do not answer from memory and do not make up a result. Say in one sentence that QuanterLab has to be connected once, and how: in Claude, open this plugin's page, press Connect on the QuanterLab connector and sign in; in Claude Code, run `/mcp`, choose quanterlab and sign in. Signing in makes a QuanterLab account; a new account starts with a free trial and no card is asked for. Then offer to run what they asked as soon as it is connected, and stop there. A backtest whose settings were picked by looking at the same history will look better than it ever trades. A walk-forward removes that advantage: it tunes the settings on one window, trades them on the next window it has not seen, steps forward and repeats. Only the unseen windows count. ## 1. Run first A request that names no stock and no rule, whatever its verb (make a backtest, test something, show me what it does), is never filled in with defaults: call `get_started`, which shows three setups from QuanterLab's Library, and play its first one at once with its `call`. It plays from its stored result and uses none of the user's runs. A walk-forward needs a rule: when there is none in the conversation and the user