BACKTEST LAB

Strategy research without fake data depth

Run backtests across multiple data sources, keep saved runs, and compare ideas without pretending weak feeds have more history than they do.

What this app is good at

Market-agnostic engine Saved runs Compare mode Dataset history

Yahoo is fine for daily bars. Local files or real exchange feeds are the right path for deep intraday research.

Backtest for execution and charts. Research for sweep and walk-forward. History for saved runs, compare, and data provenance.

Run backtest

Pick a provider, symbol, interval, and strategy. Advanced controls stay available, but the app now shows what history actually came back.

Idle

Data provider Symbol Interval Years Market tag Timezone Session tag Initial cash Fee bps Strategy

Force fresh data

Strategy params JSON Custom Python strategy

Results, saved runs, and dataset history refresh automatically.

Headline metrics first. Provenance and raw details stay available below when needed.

Show raw source and request details

Close and equity curves share the same time axis so you can see whether the strategy actually tracked the market move cleanly.

Drawdown stays separate so the pain profile is obvious instead of hiding behind the equity line.

Most recent completed trades first. Missing values are cleaned up instead of dumping raw nulls.

Entry	Exit	Side	Entry Px	Exit Px	PnL %	Equity After

Parameter sweep Rank parameter combinations without assuming the best curve is automatically real. Idle

Grid every combo, rank best-first, and show significance stats beside the curve so the leader is not just the luckiest overfit.

Sweep grid JSON (param → list of values) Sort by

Params	Return %	Sharpe	Max DD %	Trades	Win %	t (HAC)	p-value	Verdict

Walk-forward Check whether the idea survives on unseen data instead of dying right after optimization. Idle

Run the same setup on an in-sample slice and an unseen out-of-sample tail. If the edge only lived in-sample, the OOS leg exposes it fast.

OOS fraction

Saved research runs with quick reload and side-by-side comparison.

Idle

Pick	Created	Type	Symbol	Strategy	Setup	Return %	Sharpe	Verdict	Load

Overlay normalized equity and compare the saved runs you selected on the left.

Idle

Created	Type	Symbol	Strategy	Interval	Return %	Sharpe	Max DD %	Trades	Verdict
Pick saved runs and compare them here.

Dataset history Shows what data was requested versus what coverage actually came back. Idle

Fetched	Provider	Symbol	Request	Rows	Coverage returned	Note