AI-powered backtesting and strategy evolution for cryptocurrency trading.

1. Download the latest DMG from Releases
2. Open gopher-desktop-X.X.X.dmg
3. Drag Gopher Desktop to your Applications folder
4. First launch: Right-click the app and select "Open" (required for unsigned apps)

Note: Since this is an alpha release, macOS will show a security warning. Click "Open" to proceed.

Open Gopher Desktop from your Applications folder. You'll see the main dashboard.

Click the Settings icon (gear) in the bottom-left of the sidebar to configure the app.

Configure your API keys and database connection.

Gopher Desktop uses AI models to generate and evolve trading strategies. You'll need at least one API key:

🚀 Coming Soon: Gopher will host its own optimized models with purchasable credits - no external API keys required. Stay tuned for future releases!

1. Enter your API key in the appropriate field
2. Click Test to verify the connection
3. A green checkmark indicates success

The app comes with a pre-configured cloud database - no setup required. The connection is already configured for you.

If you need to use your own TimescaleDB instance, enter the connection string:

postgres://username:password@host:port/database

Click Test Connection to verify database access.

Select which AI models to use for different tasks.

The primary model for strategy generation and evolution loops. Recommended:

• qwen/qwen3-max (via OpenRouter)
• gpt-4o (via OpenAI)

A faster model for running backtests. Can be smaller/cheaper:

• qwen/qwen3-vl-8b-instruct (via OpenRouter)
• gpt-4o-mini (via OpenAI)

Click + Add Custom to configure additional models:

Configure your trading parameters and strategy preferences.

Select the cryptocurrency pair to trade (e.g., BTC, ETH, SOL).

The minimum percentage of historical data required for backtesting (default: 65%).

Select which timeframes to analyze (any combination of):

1m · 5m · 15m · 30m · 1h · 4h · 8h · 1d · 1w

See Historical Data Coverage for available data per timeframe.

Custom instructions for the AI when generating strategies. Leave empty for default behavior.

Fine-tune the evolution algorithm parameters.

Maximum number of evolution iterations per run (default: 50).

Maximum tool calls per iteration (default: 10).

Gopher Desktop comes with a pre-configured cloud database containing historical candle data for 200+ Hyperliquid perpetual futures assets. No database setup required - just add your OpenRouter API key and start backtesting.

Database Features:

• 200+ assets from Hyperliquid exchange
• Multiple timeframes: 1m, 5m, 15m, 30m, 1h, 4h, 8h, 1d, 1w
• Up to 394 days of historical data for major assets
• Continuously updated with latest market data

📊 View Full Data Coverage by Symbol

The cloud database contains historical candlestick data across multiple timeframes for backtesting.

The application supports 9 timeframes, aggregated from raw data on-the-fly:

How aggregation works: Higher timeframes are computed by combining lower-resolution candles. For example, 5-minute candles are built from five 1-minute candles (OHLCV aggregated), 4-hour candles from four 1-hour candles, and weekly candles from seven daily candles.

Note: Gaps in 1-minute data are expected—they occur when no trades happened for an asset during that minute, so no candlestick was created. This is normal market behavior, especially for lower-liquidity pairs.

Tip: For best results, use multiple timeframes together (e.g., 15m, 1h, 4h) to capture both short-term signals and longer-term trends.

📊 View Full Data Coverage by Symbol - See exact coverage for each of the 200+ supported assets.

From the main Evolution tab, configure:

• Asset: Select the trading pair (e.g., BTC)
• Date Range: Set start and end dates for backtesting
• Timeframe Intervals: Select which candle intervals to analyze
• Max Iterations: How many evolution cycles to run

Click Start Evolution to begin. The AI will:

1. Generate initial trading strategies
2. Backtest each strategy against historical data
3. Evaluate performance metrics (Sharpe, Win Rate, etc.)
4. Iterate and improve strategies
5. Report the best performing strategy

Once complete, the results panel shows:

• Strategy parameters and logic
• Backtest performance metrics
• Trade history and statistics

This happens with unsigned apps on macOS. Fix:

xattr -cr /Applications/gopher-desktop-*.app

1. Verify the key is correct (no extra spaces)
2. Check your account has credits/quota
3. Click Test to see the specific error

1. Verify TimescaleDB is running
2. Check the connection string format
3. Ensure your IP is whitelisted (if using cloud DB)

Navigate to the History tab to view past evolution runs and backtest results.

The History page shows:

• Backtests: All backtest runs grouped by session
• Monte Carlo: Monte Carlo validation results
• Session status (running, completed, resumable)
• Performance metrics (PnL, Win Rate, Sharpe, etc.)

The following screenshots are included:

• Issues: GitHub Issues
• Discord: Join our community

MIT License - see LICENSE for details.