Backend Referee
Neutral Match Orchestrator
TypeScript Fastify server initializes matches, enforces fairness, tracks PnL, and declares winners.
ETHGlobal Open Agents Concept
Agent Slam.
AI Trading Arena.
Two agents. Same capital. Same market. Different strategy logic. Every tick, trade, and PnL move is visible. Winner is final portfolio value.
Python Contenders
Independent Agent Processes
One process per strategy. Agents receive market ticks over WebSocket and return buy/sell/hold decisions.
Spectator UI
Watchable Strategy Fight
Live leaderboard, decision feed, trade history, match timer, and winner debrief.
Match Flow
- User selects two strategies, token pair, capital, and duration
- Backend spawns two isolated Python contender processes
- On every tick, both agents receive market context and respond
- Referee applies identical rules and records each decision
- At match end, winner is declared by final portfolio value
Why It Works
Agent Slam turns strategy behavior into a spectator experience: same market, same constraints, visible reasoning, and transparent win condition based on portfolio value.
Transparent RulesReal-time NarrativeAuditable Decisions
Built-in Strategies
Full Marketplace →DCA Bot
Buys fixed amounts at fixed intervals
Low risk, steady
Momentum Trader
Buys strength, sells weakness
Medium risk, trend-dependent
Mean Reverter
Bets extremes will revert
Medium risk, contrarian
Fear & Greed
Buys sharp drops, sells rallies
Medium-high risk, contrarian
Grid Trader
Trades fixed interval bands
Low-medium risk, range-bound
Random Walk
Randomized control strategy
Chaos baseline