$ panshi-lab --status
> Quant research workbench for code-first analysts
> Phase 0 invitation-only beta · 2026 Q2-Q3

A research workbench, not a trading terminal.

Walk-forward optimization · Real-friction backtesting · Private notebook sandbox · Multi-market data. Built for the engineers who write Python before they place orders.

Get beta access See pricing

Modules

Notebook

Private sandbox · Python · pandas · vbt Pro · isolated container per session.

Optimizer

Walk-forward + Optuna search across symbols, intervals and parameter spaces.

Friction Lab

Real fee tier, funding rate, slippage and latency baked into every backtest.

Datasets

A-share, HK, US and crypto K-lines via AkShare / Binance / yfinance — public sources only.

AI Copilot

Code explanation only. The model refuses every request for trade calls or price predictions.

Code preview

$ panshi-lab compute
┌─────────────────────────────────────┐
│ from panshi.engine import klines    │
│ df = klines("000001.SZ", "1d", ...) │
│ pf = vbt.Portfolio.from_signals(... │
│ pf.stats()                          │
└─────────────────────────────────────┘
                                CAGR: 12.34%
                                MDD : -23.45%
                                Sharpe: 1.21

This is a research tool. We never:

· recommend specific stocks, funds or crypto pairs
· execute trades on your behalf
· redistribute real-time market data
· evaluate any strategy as "profitable" or "ready"
· interpret macro or policy events for you