Every setup LORENTZIAN considers goes through the same pipeline before an order is ever sent: an indicator z-score, a Lorentzian-distance KNN comparison against historical patterns, a neural-net second opinion, a regime check, and finally a veto check against what we call the Black Book.
The Black Book is a per-symbol memory of the feature snapshots that preceded past losing trades. Before the AI lets a new setup through, it checks whether the current pattern sits too close to something that has already burned it on that specific symbol. If it does, the trade is skipped — no matter how promising the earlier stages of the pipeline scored it.
Why weight losses more heavily than wins
It would be simpler to just reinforce whatever setups tended to win historically. But forex pairs don’t behave like a stable, single-regime system — the same chart pattern that worked cleanly in a trending market can fail repeatedly once volatility or correlation regimes shift. A system that only chases « what worked before » has no mechanism for recognizing when a specific setup has already stopped working for a specific symbol.
Explicitly remembering losses gives the system a targeted way to say « not this again » on a per-symbol basis, without needing to retrain or manually intervene. It’s a narrower, more conservative signal than trying to predict the next winner — and narrower signals are easier to trust.
What this doesn’t mean
The Black Book is a veto layer, not a guarantee. It reduces the odds of repeating a specific, previously-observed mistake on a specific symbol — it does not predict future price movement, and it doesn’t override the account-level drawdown breaker or the other risk controls running underneath it. Those exist independently, for the failure modes that pattern memory alone can’t catch.
Trading forex carries real risk of loss regardless of how a system is built. This describes how one part of LORENTZIAN’s decision pipeline works — it is not investment advice or a performance claim.
admin · Back to all posts