Xelium | X3 Lab

Recent finance research keeps pointing to the same pattern: full automation is not always the right answer; strong systems combine human expertise with AI support. One 2024 finance study argues that replacing human judgment entirely with automation is often flawed, and that augmented decision-support systems can reduce decision noise. A 2026 FMSB review says market-facing AI is increasingly integrated into automated, scalable, analytics-driven trading systems, but does not currently operate with full autonomy; it remains under human supervision and control frameworks. That's why I'm building an AI-native trading intelligence infrastructure that helps traders understand market state, structure decisions, and compound their own experience.

Last week I read one trader’s 90-day journal. 367 closed trades. 54% win rate. -9.16% avg PnL. Then I found the pattern… 119 of 137 losses came from re-entering within 30 minutes. Same brain. Same account. Different days.

I’ll try to compile all the recent trades here, well it’s just part of the process🤫

Reinforcement learning changed how I think about trading. Not because it “predicts” the market — but because it forces brutal honesty. Train → observe → adjust → repeat. That loop applies to both machines and traders.

It’s been a while since our last update. Over the past two weeks, we’ve been fine-tuning the model to address challenges arising from a regime shift that caused temporary drawdown. Adjustments are progressing well as we continue refining performance under the new market condition