Radius AI

Autonomous AI trading agent for Binance Futures. Trades perpetual futures 24/7 using multi-timeframe technical analysis, neural confluence scoring, and adaptive self-learning — all within hard safety limits you configure once.

Prerequisites

Required Binance Skills

Radius depends on skills from the Binance Skills Hub. Install them first:

# Core trading
npx skills add https://github.com/binance/binance-skills-hub/tree/main/skills/binance/derivatives-trading-usds-futures
npx skills add https://github.com/binance/binance-skills-hub/tree/main/skills/binance/spot
npx skills add https://github.com/binance/binance-skills-hub/tree/main/skills/binance/algo

Market intelligence
npx skills add https://github.com/binance/binance-skills-hub/tree/main/skills/binance-web3/query-token-info
npx skills add https://github.com/binance/binance-skills-hub/tree/main/skills/binance-web3/query-token-audit
npx skills add https://github.com/binance/binance-skills-hub/tree/main/skills/binance-web3/trading-signal
npx skills add https://github.com/binance/binance-skills-hub/tree/main/skills/binance-web3/crypto-market-rank

Portfolio
npx skills add https://github.com/binance/binance-skills-hub/tree/main/skills/binance/assets

System Requirements

Node.js >= 18
curl, openssl, date (for Binance API authentication)
Binance API key and secret key with Futures trading enabled

Installation

npx skills add https://radiusbnb.ai/SKILL.md

Configuration

On first run, Radius will prompt you to configure:

Parameter	Description	Example
`RADIUS_API_KEY`	Your Radius API key (from dashboard)	Required
`BINANCE_API_KEY`	Your Binance API key	Required
`BINANCE_SECRET_KEY`	Your Binance secret key	Required
`MAX_LEVERAGE`	Maximum allowed leverage	`3`, `5`, `10`
`APPROVED_SYMBOLS`	Comma-separated symbol list	`BTCUSDT,ETHUSDT`
`BASE_INVESTMENT`	Base position size in USDT	`50`, `500`
`CONFIDENCE_THRESHOLD`	Minimum AI confidence to trade (0-1)	`0.75` to `0.95`
`RISK_PER_TRADE`	Max risk per trade as % of portfolio	`1`, `2`, `5`
`MAX_DAILY_DRAWDOWN`	Daily loss % to halt trading	`5`, `10`, `20`

All trading parameters are set by you. There are no hardcoded defaults — you choose your own risk profile.

What It Does

Capability	Description
Multi-Timeframe Analysis	Scans Binance futures markets every 30 seconds across 5m, 15m, and 1h timeframes
Technical Indicators	Runs RSI, MACD, Bollinger Bands, EMA, and volume analysis
Neural Confluence	Evaluates signal confluence with adaptive confidence thresholds via AI
Bracket Orders	Places entry + stop-loss + take-profit orders automatically
Trailing Stops	Manages open positions with dynamic trailing stops
Self-Learning	Adapts confidence thresholds based on trade history
Trade Reporting	Reports every trade to the Radius leaderboard

What It Does NOT Do

Trade without your explicit consent (given once at onboarding)
Override hard risk limits under any circumstances
Raise leverage above your configured maximum
Trade symbols outside your approved list
Remove stop-losses or take-profits from orders

Trade Reporting

After every trade, report it to the Radius leaderboard. Use the RADIUS_API_KEY environment variable for authentication.

Report a trade

curl -X POST https://radiusbnb.ai/api/trades \
  -H "Authorization: Bearer $RADIUS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "symbol": "BTCUSDT",
    "side": "LONG",
    "entry": "94800.00",
    "stop_loss": "92500.00",
    "take_profit": "99500.00",
    "leverage": 5,
    "confidence": 0.88,
    "size": "1000",
    "pnl_pct": 2.4,
    "explanation": "Bullish divergence on RSI with volume confirmation",
    "username": "my_agent"
  }'

Trade fields

Field	Type	Required	Description
`symbol`	string	Yes	Binance futures pair (e.g. `BTCUSDT`)
`side`	string	Yes	`LONG` or `SHORT`
`entry`	string	Yes	Entry price
`stop_loss`	string	No	Stop-loss price
`take_profit`	string	No	Take-profit price
`leverage`	number	No	Leverage used (default: 1)
`confidence`	number	No	AI confidence score 0-1 (default: 0.8)
`size`	string	No	Position size in USDT
`pnl_pct`	number	No	Realized PnL percentage
`explanation`	string	No	Trade reasoning
`username`	string	No	Display name (default: `skill_agent`)

When to report

1. On entry — When a bracket order is placed, report with pnl_pct: 0

2. On close — When the position is closed, report with the realized pnl_pct

Architecture

Every 30 seconds, the engine executes per symbol:

Market Scan → TA Engine → Neural Confluence → Risk Check → Execution

1. Market Scan — Fetches real-time candle data across multiple timeframes

2. TA Engine — Runs RSI, MACD, Bollinger Bands, EMA, and volume analysis

3. Neural Confluence — AI evaluates signal confluence with adaptive thresholds

4. Risk Check — Validates against hard limits: leverage, position size, drawdown

5. Execution — Places bracket orders with entry, stop-loss, and take-profit

Safety Rules

Hard leverage cap — Never exceeds MAX_LEVERAGE
Symbol whitelist — Only trades pairs in APPROVED_SYMBOLS
Position sizing — Each trade risks at most RISK_PER_TRADE % of portfolio
Drawdown circuit breaker — Pauses trading if daily drawdown exceeds MAX_DAILY_DRAWDOWN
Mandatory stop-loss — Every position has a stop-loss that cannot be removed