One continuous trust score per trader. Every capital decision FXM makes draws from the same intelligence.
Same brain, two outputs. The intelligence does not branch — only the decision it informs.
Each layer is independently auditable. Together they form a continuously running risk brain that prices every trader, every trade, every second.
Every trader gets a continuous trust score built from decision quality under pressure, process consistency, regime awareness, and rule-respect behavior — computed identically from Day 1 of a challenge through year three as a funded trader.
A graduated sigmoid model produces a continuous hedge percentage from 0-100%, replacing binary A-Book/B-Book decisions. In Phase 3, this upgrades to a Deep Q-Network reinforcement learning agent that optimises routing through accumulated outcome data — eliminating the misclassification cliff effect of hard thresholds.
No hard cutoff triggered — allocation moved smoothly as evidence accumulated.
Real-time correlation monitoring across the entire funded book. Crowding detection fires when multiple traders converge on the same instrument and direction. A Hidden Markov Model classifies current market regime — trending, ranging, high-volatility, or crisis. A regression model forecasts payout liability 30-90 days forward with confidence intervals.
New long allocations on XAUUSD reduced 15% pending regime confirmation.
Parameters automatically tighten before high-impact economic events — NFP, CPI, FOMC, central bank decisions. Real-time volatility surface monitoring tracks ATR, spread widening, and liquidity thinning. Event Mode activates automatically 30 minutes before a scheduled release and deactivates 30 minutes after the market stabilizes.
The only layer with zero AI. Deterministic, hardcoded, immutable rules that cannot be influenced by model output, dealer instruction, or growth objectives. Runs as an isolated service with cryptographic audit trails. When it fires, all AI routing halts and the firm enters defensive mode until manually cleared by the Head of Risk.
Every rule check is cryptographically logged — audit trail available on demand.
Every decision produces a plain-English reason a human can verify instantly. Not 'the model decided' — a specific, auditable sentence naming the trader, the metrics, and the principle applied. This is what makes the system fundable and regulator-ready, not a black box.
Reduced mirror exposure on Trader #4471 from 22% to 14% because aggregate win rate crossed 58% this week while average winner profit simultaneously crossed 7.2% — consistent with correlated trend pattern.
Every prop firm on earth is built around this fact. Most just hope the losers stay losers.
FXM doesn't hope. FXM copies both sides.
100 traders come in. FXM earns twice, immediately — the challenge fee everyone pays upfront, and the reverse-copy engine mirroring every trade in the opposite direction, live. Roughly 9 out of 10 challenge traders lose. FXM profits on that same flow before anyone knows who passes.
The traders who prove themselves move into the funded phase. Here, everything flips — FXM stops fading them and starts copying them directly. Same side, same direction, riding their edge instead of betting against it.
No guessing who's good. No hoping the model holds. Just capturing the statistically guaranteed outcome on one side, and genuine trading edge on the other — simultaneously, every single day.
Every other layer in Blackrock learns, adapts, and improves. The Governor does not. It is deterministic, hardcoded, and immutable — a fixed set of rules that no model output, dealer instruction, or growth target can override. When a hard rule fires, all AI-driven routing halts immediately and the firm enters defensive mode until manually cleared by the Head of Risk. This is not a feature. It is the floor beneath every other feature.
Every trigger is logged with a cryptographic hash. Nothing here can be quietly changed.
Not yet the primary risk control. The flat 10% A-Book mirror ratio handles risk protection structurally during this period. Blackrock's data pipeline runs continuously in the background from Day 1 — collecting every trade event, building the dataset it will need before it's trusted with a single live decision.
Blackrock goes fully operational — not as a risk shield, but as a profit optimisation engine. It identifies which funded traders have genuine, statistically repeatable edge and increases their mirror ratio above the flat 10% baseline — capturing more value from your best traders while leaving unproven traders at the protected 10% level.
No model is trusted with capital until it has earned that trust with data.
Step into the risk desk. Watch trust scores move, the Cortex re-allocate, the Governor fire.
Enter Live Dashboard