Train your agent with multi-layer context, create strategies, backtest with context replay, and let an adaptive learning engine calibrate signals over time. Export to ElizaOS or FreqTrade when you're ready. Paper trading only — practice and refine your edge with zero capital at risk.
Install the CLI, subscribe, query context, and let your agent learn from every trade. The training loop runs continuously.
1
Install
One command, globally available. Works with any AI agent or directly from your terminal.
npm i -g @trading-boy/cli
2
Subscribe
Pay with Stripe (card) or USDC on Solana. API key is provisioned instantly and stored in your OS keychain.
trading-boy subscribe --plan pro -e you@email.com
3
Query & Trade
Pull full context for any Solana token, log trades with thesis and confidence. Context snapshots are captured automatically.
trading-boy context SOL
4
Train & Export
Your agent learns from outcomes via adaptive signal calibration. Create strategies, backtest with replay, export to ElizaOS or FreqTrade.
trading-boy strategy export --format elizaos
Headline Feature
Same data. Different intelligence. Shaped by who you are.
Same token. Different traders. Different intelligence. Trading Boy personalizes synthesis, risk framing, and signal weighting to each trader's profile, behavioral history, and regime preferences. No other context platform does this — 0 of 13 competitors.
Aggressive
$ trading-boy context SOL --trader aggressive-mike
Synthesis:
"Strong entry opportunity — breakout potential building. Funding deeply negative with whale accumulation. Scale in now, add on confirmation."
Risk framing: opportunity-first
Signal weight: whale activity prioritized
Position guidance: full Kelly sizing
Conservative
$ trading-boy context SOL --trader careful-sarah
Synthesis:
"Wait for confirmation — scale in slowly with tight stops. Mixed signals, unlock in 12 days. Risk/reward insufficient at current levels."
Risk framing: risk-first
Signal weight: unlock pressure elevated
Position guidance: half Kelly, scaled entry
Same SOL context. Same market data. Same signals. The synthesis layer adapts to each trader's risk tolerance, behavioral patterns, and historical edge profile.
Trading Journal
Every trade tells a story
Log decisions with full market context attached. Your journal isn't just notes — it's training data for your agent.
Decision-linked context snapshots
When you log a trade, Trading Boy captures the full market state at that moment — prices, funding, whale activity, narratives, DeFi risk. Every decision is permanently linked to the context that produced it.
Pre-trade planning — Kelly sizing, thesis documentation, confidence scoring
Outcome tracking — PnL calculation, win/loss classification, hold time analysis
Calibration scoring — How often your 80% confidence trades actually win?
Process-outcome matrix — Separate skill from luck across all trades
Daily summaries — Delivered via email and Telegram at 08:00 UTC
$ trading-boy journal log entry SOL --direction LONG --confidence 0.75
Thesis: Funding deeply negative while price holds support
Confidence: 0.75
Kelly size: 4.2% of portfolio
Context snapshot captured:
SOL $148.32 | Funding -0.028% | OI $2.1B
Whale: 3 buys > $500K in 2h
Regime: ACCUMULATION
Signals: DISBELIEF_RALLY, CAPITULATION_BOTTOM
Decision logged. ID: dec_7f3a...
Linked to 2 active signals for learning
Learning Engine
Your agent gets smarter with every trade
A continuous adaptive feedback loop: log trades, observe outcomes, update beliefs, adapt signals. Four subsystems — signal calibration, regime adaptation, wallet predictiveness scoring, and drift detection — run automatically so each trade sharpens the next.
Context Layers
Six layers of intelligence, one package
Every context query assembles real-time data from multiple sources into a single structured object your agent can consume.
📈
Market Data
Real-time prices from Jupiter, funding rates and OI from Binance perps. 100 SPL tokens tracked every 30 seconds.
🐋
On-Chain + Whales
Whale detection via Helius. Exchange flows, smart money classification, and large transaction alerts with entity labeling.
Narrative lifecycle tracking (Emerging to Dead), sentiment aggregation, mindshare approximation from on-chain activity.
📅
Catalysts + Events
Token unlocks with materiality scoring, governance votes, protocol upgrades. Upcoming events filtered by impact.
🧠
Behavioral Coaching
Pre-trade intervention engine, tiltmeter, and trauma circuit breaker. Confidence calibration and process-outcome matrix to separate skill from luck.
Universal Integration
8 frameworks, 3 lines of config
Trading Boy speaks MCP — the open protocol for tool integration. Any agent framework that supports MCP gets full context assembly, signal calibration, and strategy management out of the box.
💬
Hermes
Nous Research
🔗
LangChain
LangChain Inc
🟣
Claude
Anthropic
🟢
GPT
OpenAI
🤖
ElizaOS
ai16z
🔵
Gemini
Google
👥
CrewAI
CrewAI
⚙
AutoGen
Microsoft
// .mcp.json — that's it
{
"mcpServers": {
"trading-boy": {
"command": "npx",
"args": ["@tradingboy/mcp-server"]
}
}
}
8 tools, 4 resources. Works with any MCP-compatible client.
Agent Intelligence
Train, test, and deploy agent strategies
The Agent Intelligence Platform turns your trading data into exportable strategies for any framework.
🤖
Agent Strategies
Create strategies from your trading data — signal weights, regime behavior, and risk limits derived from your actual edge profile.
⏪
Context Replay
Backtest strategies against historical context snapshots. Replay up to 90 days of full 6-layer context to validate before going live.
📤
Strategy Export
Export to ElizaOS character files, FreqTrade strategy.py templates, or raw JSON for custom integrations. One command.
🔌
ElizaOS Plugin
Native integration via @elizaos/plugin-tradingboy. Your ElizaOS agent gets full context assembly, signal calibration, and regime awareness.
⚡
WebSocket Streaming
Real-time updates across 7 channels: edge updates, belief snapshots, strategy performance, signal attribution, execution feedback, drift alerts, and regime changes.
🎯
Execution Feedback Loop
Track strategy performance with signal attribution and edge cell analysis. See which signals drive PnL and how execution quality evolves.
CLI Reference
Everything from your terminal
Full command reference — from querying context to journaling trades to reviewing your edge.
Context & Data
trading-boy context SOL — Full 6-layer ContextPackage
trading-boy query SOL — Quick price, funding, whale snapshot
trading-boy watch SOL -i 5 — Live watch with 5s refresh
trading-boy trader link-wallet— Link on-chain wallet
Setup & System
trading-boy subscribe --plan pro -e you@email.com — Subscribe (Stripe or USDC)
trading-boy login— Authenticate with API key
trading-boy logout— Remove stored credentials
trading-boy whoami— Show current identity & plan
trading-boy billing manage— Open Stripe billing portal
trading-boy config show— View configuration
trading-boy config set-llm-key <key> — Set BYOK LLM key
trading-boy infra status— Health check
All commands support --format json for machine-readable output. Full reference: docs
Pricing
All features, every tier
Every plan includes the full system — context assembly, learning engine, agent strategies, MCP server, WebSocket, and more. Tiers differ only by rate limits. Pay with Stripe or USDC on Solana.