Groww AI Swing Trader is a full-stack, read-only swing-trading analysis platform for the Indian markets: a FastAPI backend that scans, scores, and backtests NSE symbols through the Groww API, and a Next.js trading desk that turns the output into decisions — charts, factor radars, catalyst timelines, risk planning, and an Azure OpenAI analyst you can interrogate.

Read-only by design. The platform analyses live account and market data but exposes no order placement, modification, or cancellation — it is an analysis copilot, not an execution bot.

Highlights

• 12+ technical indicators — RSI, MACD, EMA, SMA, VWAP, ATR, Bollinger Bands, Supertrend, Donchian & Keltner channels, Stochastic, and momentum — computed via pandas-ta with dependency-free fallbacks, plus candlestick pattern detection.
• Multi-factor long/short signal engine — alpha factors, market-regime detection, and a strategy library with a declarative spec format (strategies can even be imported from GitHub) feed a multi-factor scanner that ranks long and short candidates.
• Transaction-cost-aware backtesting — realistic brokerage, STT, and slippage assumptions (all configurable) with Sharpe ratio, maximum drawdown, profit factor, and win rate reported per run, and paper trading to forward-test signals.
• Multi-provider market data — Groww SDK first with NSE India, Google Finance, and Screener fallbacks, cached in SQLite with per-class TTLs and a request budget so free-tier API limits survive a full scan.
• AI analyst mode — Azure OpenAI summaries, threaded Q&A, and streaming commentary grounded in the platform's own indicator, factor, and catalyst context.
• Catalysts & sentiment — earnings calendar, news catalyst ingestion, sentiment scoring, watchlists, and Telegram alert delivery.
• Live trading desk — a dockable Next.js 15 dashboard (dockview + lightweight-charts) with stock analyzer, account cockpit (holdings, positions, orders, margin, risk cards), live price streams, and an admin panel with health metrics, feature flags, and audit logs.

Architecture

flowchart LR
    subgraph Web["🖥️ Next.js 15 trading desk"]
        PAGES["Pages<br/>stock · live-desk · account · admin"] --> COMP["Dockable panels<br/>charts · factor radar · AI analyst"]
        COMP --> API_TS["Typed API client<br/>lib/api.ts"]
    end
    subgraph Backend["⚙️ FastAPI backend"]
        ROUTES["API routes"] --> SVC["Services<br/>scanner · indicators · backtests<br/>strategies · regime · factors"]
        SVC --> DATA["Market data layer<br/>SQLite cache + request budget"]
        ROUTES --> AI["AI analyst<br/>commentary · threads"]
    end
    subgraph Providers["🌐 Data providers"]
        GROWW["Groww SDK"]
        NSE["NSE India"]
        GF["Google Finance"]
        SCR["Screener"]
    end
    API_TS --> ROUTES
    DATA --> GROWW
    DATA --> NSE
    DATA --> GF
    DATA --> SCR
    AI --> AOAI["Azure OpenAI"]

Loading

• src/groww_trader/ — Python package: Groww client, CLI, and ~35 service modules (indicators, patterns, scanner, strategies, backtests, paper trades, catalysts, sentiment, alerts, admin, AI).
• app/ + components/ — Next.js App Router pages and 50+ React panels for the trading desk.
• lib/ + hooks/ — typed API client, AI tool context, and live-stream React hooks.
• tests/ — pytest suites for indicators, strategies, analytics, market data, and session/AI gating, plus vitest for the web client.

Getting started

Prerequisites

• Python ≥ 3.9 and Node.js ≥ 20
• A Groww trading API key (or TOTP credentials)
• Optional: an Azure OpenAI deployment for analyst mode, a Telegram bot for alerts

Setup

# 1. Backend
python3 -m venv .venv && source .venv/bin/activate
pip install -e .

# 2. Frontend
npm install

# 3. Configure
cp .env.example .env   # fill in Groww credentials (+ optional Azure OpenAI / Telegram)

# 4. Run everything (FastAPI on :8000, Next.js on :3000)
groww-trader dashboard

CLI

groww-trader profile
groww-trader quote RELIANCE --exchange NSE --segment CASH
groww-trader candles RELIANCE --start-time "2026-05-20 09:15:00" --end-time "2026-05-20 15:30:00" --interval 5
groww-trader dashboard --api-only

Testing

pytest      # backend: indicators, strategies, analytics, market data, AI gating
npm test    # web client (vitest)
npm run lint

CI runs both suites plus ESLint on every push and pull request.

Disclaimer

This project is for research and education. It is not investment advice, and it deliberately cannot place orders. Trade at your own risk.

License

MIT © Lalit Kumar