Matrix Validation Server
44 rules. One AI. No bad trades.
Every trading system has the same problem. The setup looks good. The chart agrees. And then the trade opens at exactly the wrong moment — wrong session, correlated pair already running, spread too wide, trend actually pointing the other way.
The Matrix Validation Server exists to catch that moment before it costs you.
How it works
Every signal from your cBot goes through the server before a single order is placed. The server doesn't guess. It runs the signal through 44 individual filter blocks, one by one, each checking a specific condition.
Session filters — is this a valid trading window for this pair, or is liquidity too thin?
Trend filters — is price actually moving in the direction the bot wants to trade, or just retracing?
Strength filters — is the currency behind the signal showing real momentum, or is it flat?
PSAR filters — has the parabolic SAR confirmed the move on the relevant timeframes?
Risk filters — is there already too much exposure on correlated pairs? Is drawdown within limits?
If the signal passes all 44 blocks, it moves to the final stage.
The AI layer
The last check isn't a rule. It's a question.
The server sends the full signal context — pair, direction, score, zone data, session, ATR, RSI, currency strength readings — to an LLM running on Groq. Groq uses the LLaMA 3.3 70B model, one of the most capable open models available.
The AI reads everything and returns one of two answers: ALLOW or BLOCK.
It also explains why. Every decision is logged — what the AI saw, what it concluded, and how long it took. Response time is typically under two seconds.
This isn't AI for the sake of AI. The 44 filters handle the mechanical rules. The AI handles the context — the things that are hard to express as a rule but obvious when you see the full picture.
What it runs on
A basic Ubuntu VPS. €5–10 a month from any provider. The server is a Python application — no heavy dependencies, no database beyond SQLite, no cloud services required beyond the Groq API key.
You get a live web dashboard showing every signal check in real time — what came in, which blocks it passed, which it failed, and what the AI decided. You can watch your system think.
The result
Signals that would have opened losing trades during dead sessions — blocked.
Signals that looked fine on one pair but were correlated with three others already running — blocked.
Signals that passed every mechanical rule but arrived during a high-impact news window — blocked.
Signals that were genuinely good — allowed, with SL and TP already calculated, position size confirmed, Telegram notification sent.
The server runs in the background. You don't watch it. That's the point.
Matrix Validation Server is part of the ZaliMatrix ecosystem. It connects to GapTrader, FutureTrader, and MatrixBasket Hybrid — or runs standalone as a filter layer for any cTrader cBot that can make an HTTP request.