AI That Improves Itself
Most trading systems are static. Ours runs experiments 24/7, keeps what works, and discards what doesn't. The system you use today is smarter than the one from last week.
Your AI Trading Desk
Each agent has one job. Together, they cover every angle β from scanning markets to managing risk to tracking performance.
Scanner
Scans 38 instruments across forex, crypto, indices, and commodities during active kill zones. Finds setups before you wake up.
Analyst
Scores every setup through 3 layers of institutional data β COT positioning, retail sentiment, and H4 trend alignment.
Executor
Places and manages live trades with dynamic position sizing, partial profit splits, and progressive trailing stops.
Guardian
Enforces risk rules 24/5. Circuit breakers, daily loss limits, consecutive loss cooldowns. Protects your capital first.
Intel
Monitors news events, session transitions, and market regime changes. Blocks trades before high-impact releases.
Tracker
Tracks every trade to outcome. Daily report cards, weekly recaps, equity curve snapshots. Full transparency.
Copier
Mirrors trades from the master account to your MT5, MT4, or other platforms. Your risk settings, our execution.
Amun
AI assistant powered by GPT-4 and Claude. Answers questions, explains setups, and coaches you through market conditions.
Self-Improving Strategy Engine
Inspired by Karpathy's autoresearch. An autonomous AI agent experiments with strategy parameters, backtests against real historical signals, and keeps only what makes more money.
Read Current Strategy
The agent loads every tunable parameter β confidence thresholds, volume tiers, risk limits, trailing stops, session filters.
Modify Parameters
AI proposes a change β tweak a confidence boost, adjust volume scaling, shift a kill zone window. One experiment at a time.
Backtest Against Real Data
Runs the modified strategy against 90+ days of real historical signals from the production database. No fake data.
Compare Results
Net profit is king. If the change makes more money, it stays. If not, it gets reverted immediately.
Log Everything
Every experiment is tracked β commit hash, net profit, win rate, profit factor, max drawdown. Full audit trail.
Repeat Forever
The agent loops 24/7 with cost controls. 50 experiments per session, early stop after 10 no-improvement runs. Always searching.
Recent System Upgrades
We ship improvements every week. Here's what changed recently.
Autoresearch Agent Deployed
Mar 2026Autonomous AI agent now runs 24/7, testing strategy parameter changes against real historical data. First session: 695% improvement in net profit.
MRA-Only Architecture
Mar 2026Simplified to single-strategy execution. Market Reversal Alerts EA provides 24/7 signal generation. WEAK signals auto-blocked. MEDIUM+ pass to execution.
Risk Engine v8.0
Mar 2026Added MRA grade boost scoring, OFF_HOURS hard block, direction retry limit, and tighter ATR-based stop losses. Circuit breaker with $500 warn / $1000 kill thresholds.
Dynamic Volume Scaling
Feb 2026Lot sizes now scale automatically with account balance. 2% risk target maintained whether trading $1K or $100K. No manual adjustment needed.
3-Layer Confluence Gate
Feb 2026Every signal must pass through COT institutional positioning, retail sentiment (contrarian), and H4 trend alignment. Requires 2+ layers aligned before execution.
Built Different
Most trading tools are black boxes. You pay, you get signals, you have no idea why. We show you the scoring, the confluence data, the risk calculation, and the outcome β every single time.
And while other systems stay frozen after launch, ours evolves. The autoresearch agent tests parameter changes against real data and applies only what works. The system you trade with today is measurably better than last month's version.
Your Next Prop Challenge Doesn't Have to Fail.
8 agents. 38 instruments. 6 platforms. 3 layers of institutional validation. The system runs whether you're watching or not.