10,000 agent genomes trading live multi-market data. Genetic tournament selection. 115-dim feature space. Realistic paper trading with Kraken fee models (0.26% taker).
Real-time dashboard served from the laptop engine (C data_server + nginx, auth + rate-limited).
The full engine UI with live agent rankings, market prices, pipeline health, cycle metrics, and API key management — served from the laptop via secure nginx reverse proxy with basic auth and strict rate limiting.
Multi-source data pipeline feeding a real-time genetic evolution engine.
10,000 agent genomes with crossover, mutation, and tournament selection. Genomes are initialized with Xavier weights — evolution kicks in as training cycles accumulate.
10K AgentsData sources across crypto, equities, forex, on-chain metrics, sentiment (GDELT), prediction markets, and economic indicators — all in C.
12+ SourcesCross-asset training across crypto, equities, forex, commodities, and prediction markets with 16 differentiated room strategies.
16 RoomsSub-millisecond inference per agent. Feature extraction, signal generation, and capital allocation all in C with zero Python overhead.
~5ms/cycleVolume profile, order book imbalance, options flow, funding rates, liquidation clusters, correlation regimes, and volatility surface.
80+ FeaturesRealistic simulation with Kraken spot fee model (0.26% taker), slippage, and liquidity constraints. No real capital at risk.
LiveReal-time prices from our C collector pipeline.
| Cycle Time | — |
| Votes Cast | — |
| Top Agent PnL | — |
| Sharpe (avg) | — |
| Max Drawdown | — |
| Paper Age | — |
| Disk | — |
| Memory | — |
| Uptime | — |
| Active Crons | — |
| Failed Crons | — |
| Deployed | — |