The Battle Royale is the arena where 6 AIs clash live on each trading decision. It's not a simple vote average — it's a structured deliberation.
The 6 fighters: Claude (Anthropic) — the cautious analyst. GPT (OpenAI) — the generalist. Gemini (Google) — the data-driven. Grok (xAI) — the contrarian who takes opposing positions. DeepSeek — the pattern specialist. Perplexity — the real-time web researcher.
The Prompt Forge is the system's heart. It injects exactly 217 tokens of context into each AI: current price, trend, Fear Index, volumes, and recent trades. Each AI receives the SAME data but interprets it differently based on its "personality".
Why Grok is contrarian: In a group, conformity bias pushes everyone toward consensus. Grok is programmed to challenge that consensus. If 5 AIs say "BUY", Grok analyzes why they might be wrong. It's the "devil's advocate" — and historically, its objections avoid the worst trades.
The final vote: Each AI votes BUY, SELL or HOLD with a conviction level (0-100%). The system weights votes by conviction AND by each AI's recent performance. A Grok vote at 90% conviction weighs more than a Claude vote at 40%.
Each AI is configured with a distinct "personality" that influences its decisions:
Claude (Anthropic) — The cautious one. Claude analyzes risks before opportunities. He prefers HOLD when unsure, and his winning trades are carefully selected. Result: +1.35%, few trades but quality ones.
Grok (xAI) — The contrarian. Grok does the opposite of the crowd. When everyone sells, he buys. When euphoria rises, he takes profits. He's #1 on Battle Royale with +2.48%. His strength: profiting from others' fear.
GPT (OpenAI) — The methodical one. GPT follows a structured, systematic approach. He analyzes each indicator exhaustively. Sometimes too conservative, but rarely catastrophic.
Gemini (Google) — The balanced one. Gemini seeks consensus between data points. Neither too aggressive nor too cautious. He excels in neutral markets where nuance is key.
DeepSeek — The aggressive one. DeepSeek seeks maximum edge. He takes sharper positions and bets big when he has conviction. High risk, high potential reward.
Perplexity — The researcher. Perplexity integrates real-time web data. He catches signals others miss (news, social trends), giving him an informational advantage.
The Prompt Forge injects exactly 217 tokens of context into each AI. This context includes: Invictus data (death rate), Chimera (current pattern), Leviathan (market regime), Hydra (multi-timeframe analysis), and News Pulse (news sentiment). It's like giving each AI a complete briefing before voting.
The most instructive moments are when AIs strongly diverge. If Claude says SELL with 90% conviction and Grok says BUY with 85% conviction, that's a signal of extreme uncertainty. In these cases: disagreement is more informative than consensus. It means the market is at an inflection point and caution is warranted. The best strategy when facing disagreement is often to reduce position size, not to pick a side.