indian-algo-trading

Indian Algo Trading

Production-quality Python trading strategies for Indian markets — AI-assisted, from backtest to live.

Overview

This plugin helps you build algorithmic trading strategies for Indian markets (NSE, BSE, MCX) with best practices baked in. It doesn't ship pre-built strategies — instead, it teaches AI how to help you design safe, realistic, and compliant strategies from scratch.

Covers the full lifecycle: backtesting → optimization → paper trading → live deployment, across equity, F&O, currency derivatives, and MCX commodities.

---

Installation

Claude Code Plugin

# Register the marketplace (one-time)
claude plugin marketplace add RupeezyTech/algo_ai_skill

# Install
claude plugin install indian-algo-trading@rupeezy

Standalone Skill

Download from Releases and extract:

unzip indian-algo-trading-*.skill -d ~/.claude/skills/indian-algo-trading/

Local dev / testing

git clone https://github.com/RupeezyTech/algo_ai_skill.git
claude --plugin-dir ./algo_ai_skill

---

Usage

Once installed, the skill activates automatically when you ask about:

• Writing a strategy ("write a moving average crossover for Nifty")
• Backtesting ("backtest this on 2022-2024 data with realistic costs")
• Live trading ("make this strategy production-ready for live deployment")
• F&O automation ("iron condor strategy for weekly expiry")
• Risk management ("add position sizing and daily loss limits")

The skill will ask clarifying questions (asset class, live vs backtest, broker, risk tolerance) before generating code.

---

What Gets Generated

Every strategy follows a strict separation of concerns:

my_strategy/
├── main.py          # Entry point, scheduling, SIGTERM handler
├── strategy.py      # Signal generation only
├── execution.py     # Order placement, fill tracking
├── risk_manager.py  # Position sizing, exposure checks, drawdown limits
├── config.py        # All parameters — no hardcoded values
└── requirements.txt

Every strategy includes: stop-losses, margin checks, tick size rounding, IST timezone, structured logging, and graceful shutdown. No exceptions.

---

Reference Library (16 files)

File	Covers
strategy-patterns.md	Momentum, mean reversion, options, pairs trading
risk-management.md	Position sizing, drawdown controls, margin monitoring
indian-market.md	Timings, expiry calendar, STT, circuit limits, auction risk
backtesting.md	Library selection, realistic costs, parameter optimization
error-handling.md	Order state machine, partial fills, graceful shutdown
code-quality.md	Project structure, logging, testing, type hints
options-greeks.md	Delta-neutral, gamma scalping, theta harvesting, IV vs RV
regime-detection.md	HMM for trending/volatile/sideways, strategy decay
india-data-edge.md	FII/DII flows, OI analysis, PCR, max pain, delivery %
execution-alpha.md	TWAP, VWAP, iceberg, impact cost, intraday timing
robustness-testing.md	Walk-forward, Monte Carlo, sensitivity analysis
portfolio-construction.md	Multi-strategy allocation, correlation-aware sizing
psychological-guardrails.md	Daily loss breaker, consecutive loss pause, killswitch
tax-optimization.md	STCG vs LTCG, tax-loss harvesting, F&O business income
python-performance.md	Vectorization, Numba, Polars, async, profiling
brokers/rupeezy-vortex.md	Full Vortex SDK reference for live trading

---

Repository Structure

algo_ai_skill/
├── .claude-plugin/
│   └── marketplace.json              # Marketplace catalog (GitHub sync)
│
├── plugins/
│   └── indian-algo-trading/
│       ├── .claude-plugin/
│       │   └── plugin.json           # Plugin manifest
│       ├── .mcp.json                 # Rupeezy MCP server config
│       └── skills/
│           └── indian-algo-trading/
│               ├── SKILL.md          # Skill instructions + routing logic
│               ├── references/       # 16 reference files
│               └── scripts/
│                   ├── scaffold_strategy.py      # Generate strategy skeleton
│                   └── validate_strategy.py      # AST linter for common mistakes
│
├── evals/
│   └── evals.json                    # 10 skill evaluation test cases
│
├── build/                            # Generated — gitignored
│   ├── *.skill                       # Standalone skill zip (GitHub release)
│   └── *.plugin                      # Full plugin zip (Anthropic marketplace)
│
└── Makefile                          # Build, validate, release

---

Developer Commands

# Build both artifacts
make all

# Build standalone skill zip only
make skill

# Build full plugin zip only
make plugin

# Validate JSON manifests and SKILL.md frontmatter
make validate

# Test scaffold script generates valid output
make test-scaffold