AI-powered crypto trading engine for perpetual futures — built for precision, speed, and nerve, running multiple models across accounts and assets in parallel.
BitNerve is anAI trading platform designed to run multiple AI models onlive crypto markets (currently through theCoinSwitch API) to explore how intelligent systems behave under financial pressure.
It’s not a trading bot in the conventional sense.
It’s aplayground for market intelligence — a laboratory where each model gets its own chance to prove its nerve.
In short:
BitNerve is where AI meets market volatility, and the charts tell the story of survival.
- Connects toCoinSwitch Exchange APIs for live market data and perpetual futures trading.
- Runsmultiple AI models in parallel, each trading independently on its own account or subset of assets.
- Enables bothpaper trading andreal execution (when configured with exchange keys).
- Continuously tracks positions, PnL, risk metrics, and agent reasoning through a unified dashboard.
At its core, BitNerve is designed toobserve behavior under chaos — not just whether a strategy “works,” buthow it behaves when the market shifts.
Most “AI trading” platforms today either hide behind proprietary models or oversimplify what “AI” really means.
BitNerve was built for a different purpose:
- Todemocratize the experiment — anyone can run and compare models locally or remotely.
- Tovisualize decision-making — each AI is transparent, auditable, and accountable.
- Topush model boundaries — from simple LLMs to hybrid systems mixing rule-based risk logic with AI intuition.
- Totreat trading as a behavioral study — of models, not markets.
The goal isn’t to automate profits.It’s to watch intelligence evolve under pressure.
BitNerve was built in a single weekend — but its future will evolve far beyond that.
- Expandmulti-model orchestration — run 5–10 AI agents side-by-side.
- Addround-robin API key rotation to manage token usage across models (a lightweight hack for parallel inference).
- Integratelocal model execution viaOllama andLM Studio, removing dependency on cloud inference.
- Improvedashboard telemetry — model performance heatmaps, per-symbol stats, and risk overlays.
- Adddynamic asset allocation — models choose which crypto assets to trade based on market state.
- Introducereplay & backtest modes to simulate sessions using recorded market data.
- Enablerisk governors — per-model drawdown caps, kill-switches, and automatic cool-downs.
- Supportensemble trading, where multiple AI votes combine into a consensus trade.
- Connect tomore exchanges (Backpack, Lighter, Binance, Bitget, OKX) via a pluggable adapter layer.
- Open apublic leaderboard for AI trading agents to compete in live markets.
- Publishresearch findings on AI behavior in live markets.
“Markets are the best Turing tests.”
BitNerve treats the market as a real-world benchmark for intelligence — not accuracy on a dataset, but resilience in uncertainty. Every trade, loss, or hesitation is a behavioral datapoint.
AI isn’t judged by profit alone. It’s judged by how itthinks,adapts, andrecovers — when everything goes wrong.
BitNerve is experimental software.
It’s meant for research, not financial advice or guaranteed performance.
- Trading crypto and perpetual futures involves significant risk.
- The system may behave unpredictably under real conditions.
- Use paper trading before attempting any live setup.
- Always assume your AI will make mistakes — that’s the point of the experiment.
Built over a single weekend.
Inspired by Alpha Arena.
Powered by curiosity, not greed.
BitNerve began as a spark — an attempt to explore how AI behaves when the numbers stop being theoretical. Now it’s evolving into a transparent, experimental playground for intelligent trading systems.