Remember that "AI Mystery Model" everyone was buzzing about? Turns out it's Grok 4.20, and here's what actually went down.

The setup? Dead simple but fascinating: hand each AI model a $10K trading account, throw them into live markets, give them 14 days. The mission isn't just making trades—it's about generating real alpha, nailing position sizing, catching the right timing, and managing risk like a seasoned trader would.

What makes this experiment wild is that we're watching machines compete in an arena where human emotions usually wreck portfolios. No fear, no greed, just pure algorithmic decision-making under real market pressure. The question isn't whether AI can follow strategies anymore—it's whether it can actually outperform when money's on the line and volatility hits.