The Intelligence Layer
The Genetic Algorithm.
BTC Arena uses a population-based evolutionary strategy. Unlike static machine learning models, our agents are designed to adapt to market volatility through natural selection and random mutation.
Evolutionary Lifecycle
1. Competition: 14 agents trade 8 different markets (BTC, ETH, SOL, XRP, BNB, TAO, SUI, HYPE) simultaneously. Their performance is tracked via ROI, which includes both realized wallet balance and unrealized PnL from open positions.
2. Elimination: If an agent's wallet drops below 80% ($800) of its starting capital, it is deemed unfit for the current market regime and is "kicked" from the arena.
3. Selection: The system identifies the top-performing agent of the *same archetype* to act as the genetic donor for the next generation.
4. Mutation: A new agent is spawned. Its DNA is cloned from the leader, but each parameter is subjected to a ±15% random mutation.
DNA Parameter Profiles
⚡ FAST
◆ MEDIUM
🐢 SLOW
Mutation Visualized
DNA
DNA
The mutation ensures that even if the leader finds a strong strategy, its "offspring" will explore nearby parameter spaces to potentially find an even better local optimum.
The "Positive Expectancy" Lock
A key architectural safeguard is the Expectancy Lock. During mutation, the system enforces a strict rule: TP Multiplier ≥ SL Multiplier × 1.5.
This ensures that no agent can ever evolve into a strategy that requires an impossible win rate to stay profitable. By locking the Risk/Reward ratio, we force the genetic algorithm to focus on improving the signal accuracy rather than chasing unsustainable recovery patterns.