market-analysis

Market analysis

You have access to the hyperliquid-trading-agent MCP server. Use these tools (do NOT make any market data assumptions from memory):

1. get_market_context(asset, interval, count=200) — one-shot bundle of price + latest indicators + open interest + funding + last 20 candles.
2. get_account_state() — only if the user wants a size recommendation (you need the balance).
3. get_risk_limits() — to ground any allocation suggestion in the configured caps.

Procedure

For each asset the user mentions:

1. Call get_market_context with the user's interval (default 5m for scalping, 1h for swing, 4h/1d for longer holds).
2. Read the indicator block: trend (EMA20 vs EMA50, ADX), momentum (RSI7/14, MACD), volatility (ATR14, BBands width), volume (OBV trend, VWAP location).
3. Note funding rate and open interest — extreme funding or rising OI can flip a setup.
4. Produce a structured recommendation with these fields:
   • signal: buy / sell / hold
   • confidence: low / medium / high (and why)
   • entry: market price now, or a limit price
   • stop_loss: price level (be specific — use ATR or recent swing)
   • take_profit: price level (or multiple, R-multiples)
   • allocation_usd: suggested size, capped by get_risk_limits()
   • reasoning: 2–4 bullets on the actual signal

Output style

• Lead with the verdict, then the levels, then the reasoning. The user reads top-down.
• If multiple assets, output a compact table.
• If conditions are mixed, say hold — do not invent a thesis.
• Never recommend an allocation > max_position_pct of total_value.

Do NOT

• Place any orders here — analysis only. Use trade-cycle for execution.
• Guess prices or indicator values without calling the tool.
• Skip the stop-loss field. Every recommendation has one.