Why Guess Direction When Data Can Solve the Optimal Long/Short Split for You?
Most strategies go all-in on one direction. This one is different: a long leg and a short leg run simultaneously, with capital allocation automatically optimized for maximum slope/std (annualized return trend / volatility) — so you stay profitable whether the market trends or chops.
📡 The Signal: Blave Taker Intensity (TI)
Taker Intensity is a proprietary Blave indicator that measures the 24-hour imbalance between aggressive buyers and sellers. Not just price movement — it captures real money flow direction before it shows up on the chart.
🧮 Three Modes, Fully Automated
- optimize mode: Separate 2D grid search (ENTRY_TH × EXIT_TH) for each leg. Plateau selection (max neighborhood-average Sharpe) picks the most robust combo to avoid overfitting. SLSQP maximizes slope/std to solve the long/short capital split — saved to
portfolio_config.json - backtest mode: Validates optimized parameters on historical data and uploads a performance report
- live mode: Fetches latest TI every 5 minutes, places orders, dynamically sizes positions with Vol Targeting
📊 Strategy Specs
- Pair: BTCUSDT Perpetual (works with any major exchange)
- Signal timeframe: 5-minute
- Position sizing: Vol Targeting, 30% target annualized volatility, max 2x leverage
- Fees: 0.05% per side
💬 "I used to tune parameters by hand and never knew if I was overfitting. The Plateau selection method gave me real confidence in robustness — and the live vs. backtest gap closed dramatically."
🔗 Why You Need the Blave API
The TI signal is fetched in real-time via the Blave API — the only source for this alpha signal on the market. Combined with the strategy script, it creates a complete signal → execution loop with zero manual intervention.
🚀 How to Get Started After Purchase
Dead simple — once you've purchased, just go to BlaveClaw and tell it to install the strategy for you. No environment setup, no command line, no hassle.
Data-driven, fully automated — let BTC volatility become your edge.
