User Guide
Arena — Live Feed
How the Arena live feed works and how to trigger a new analysis.
The Arena is the main feed of all recent council debates, sorted by the most urgent or interesting results first.
Sorting
The feed uses an urgency score to surface the most signal-rich debates:
- Split + high conviction debates float to the top — when agents strongly disagree, that's worth reading closely.
- You can manually re-sort by Urgency (default), Conviction (highest weighted score), or Disagreement (most internal conflict).
Analysing a Token
Use the search bar at the top of the page to find a token. Select it and the council will run a full 6-agent debate in real time.
What happens when you submit:
- Five agents (Bull, Bear, Quant, Macro-Regime, Narrative) run in parallel, each calling the Ryo MCP API for market data and using
gpt-4o-minifor reasoning. - The Contrarian runs last — it stress-tests the top-conviction peer verdict.
- A consensus is generated from all six verdicts: final call (long/short/neutral), agreement level, and conviction score.
- The result appears at the top of the Arena feed.
Rate Limiting
Manual analyses are rate-limited to 4 requests per minute per IP to prevent abuse. Scheduled rechecks via Triggers & Alerts bypass this limit.
Debate Cards
Each card in the feed shows:
- Token symbol, final call (BULLISH / BEARISH / NEUTRAL)
- Agreement level (Unanimous / Majority / Split)
- Conviction score (0–100)
- Time of analysis
- A summary of the outlier agent (if any dissented)
Click any card to open the full token detail page with the complete debate history and individual agent reasoning trails.