# Panther MCP Server

**Backtest any trading strategy by describing it to Claude, ChatGPT, or Gemini.**

Panther gives your AI assistant the tools to fetch market data, run backtests, and show you the results — no code required. Just describe your strategy in plain English.

## Quick Start

### 1. Get your API key

Sign up free at [panther.watch](https://panther.watch) and copy your API key.

### 2. Install

**Claude Desktop** — add to your `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "panther": {
      "command": "uvx",
      "args": ["panther-mcp"],
      "env": {
        "PANTHER_API_KEY": "pthr_your_key_here"
      }
    }
  }
}
```

**Claude Code:**

```bash
claude mcp add panther -e PANTHER_API_KEY=pthr_your_key -- uvx panther-mcp
```

**ChatGPT** (Pro/Team/Enterprise — Developer Mode):

Add as custom MCP server: `https://panther.watch/mcp/`

**Gemini CLI:**

```json
{"mcpServers": {"panther": {"command": "uvx", "args": ["panther-mcp"], "env": {"PANTHER_API_KEY": "pthr_..."}}}}
```

**Any MCP client** (Cline, Cursor, VS Code Copilot, etc.):

```bash
uvx panther-mcp
```

### 3. Start backtesting

Just tell your AI what strategy you want to test:

> "Backtest a golden cross strategy (SMA 50/200) on BTC/USDT over 2024"

You'll get results including return, Sharpe ratio, drawdown, every trade, and a shareable link to the full report.

## Available Tools

| Tool | Description |
|------|-------------|
| `list_available_assets` | Browse 30+ tradeable assets (crypto, forex, commodities) |
| `get_price_data` | Fetch historical OHLCV price data with summary stats |
| `run_backtest` | Define strategy with indicators + conditions, execute backtest |
| `get_backtest_status` | Poll running backtest status + progress |
| `get_backtest_results` | Get performance metrics + shareable panther.watch link |
| `list_backtests` | View your backtest history |
| `optimize_strategy` | Run parameter sweeps across combinations |
| `get_optimization_status` | Poll optimization progress |
| `get_optimization_results` | Get ranked parameter results |
| `list_optimizations` | View optimization history |
| `run_portfolio_backtest` | Test a strategy across multiple weighted assets |
| `get_portfolio_backtest_status` | Poll portfolio backtest progress |
| `get_portfolio_backtest_results` | Get portfolio + per-asset results |

## Supported Assets

| Type | Source | Examples |
|------|--------|----------|
| Crypto | Binance (CCXT) | BTC/USDT, ETH/USDT, SOL/USDT + 17 more |
| Forex | IC Markets (cTrader) | EUR/USD, GBP/USD, USD/JPY + 7 more |
| Commodities | IC Markets (cTrader) | XAU/USD (Gold), XAG/USD (Silver) |

## Features

- **25 indicators** — SMA, EMA, RSI, MACD, BB, VWAP, ATR, Stochastic, ADX, OBV, Supertrend, Ichimoku, and 13 more
- **Short selling** — Long, short, or bidirectional strategies
- **Strategy optimization** — Parameter sweeps with constraints and ranking by any metric
- **Portfolio backtesting** — Test across multiple weighted assets with portfolio-level metrics

## Example Strategies

**Golden Cross:**
> "Backtest SMA 50/200 crossover on BTC/USDT daily for 2024"

**RSI Mean Reversion:**
> "Buy when RSI(14) drops below 30, sell above 70 on EUR/USD daily with 5% stop loss"

**MACD Crossover:**
> "Enter when MACD line crosses above signal line, exit on reverse, on ETH/USDT daily"

**Portfolio:**
> "Backtest RSI strategy on a portfolio of 60% BTC and 40% ETH over 2024"

**Optimization:**
> "Optimize the SMA crossover on BTC — try fast period 10-50 step 10, slow period 50-200 step 50"

## Shareable Results

Every backtest gets a public URL you can share on Twitter, Discord, or with your trading community:

```
https://panther.watch/backtests/<backtest-id>
```

Results include performance metrics, equity curve chart, and full trade list.

## Development

```bash
git clone https://github.com/PantherWatch/panther-mcp.git
cd panther-mcp
uv sync
PANTHER_API_KEY=pthr_... uv run panther-mcp
```

## License

MIT