✅ My Telegram AI Assistant: — My AI assistant will help you about manual guide, setting, …
✅ How Our AI Trading Server Works: — Where you can observe how news data is processed on the server transparently…
1. Abstract
This document outlines the architectural framework of a proprietary Expert Advisor (EA) designed to bridge retail trading platforms with institutional-grade artificial intelligence. The system utilizes a two-stage processing model consisting of a local technical scan and a centralized AI verification layer. This paper explains the rationale behind the “No API Key” requirement, details the precise lifecycle of market data transmission, defines the privacy protocols governing data collection, and describes the built-in fail-safe mechanisms that ensure operational integrity. The objective is to provide absolute transparency regarding how the system interacts with user terminals and remote servers to generate trading signals.
2. System Architecture: The Two-Stage Model
The EA operates on a hybrid architecture that distributes processing loads to optimize for both speed and analytical depth. This design negates the need for individual users to supply personal API keys.
2.1 Stage 1: Local Technical Scan (Client-Side Processing)
This process executes directly within the user’s MetaTrader terminal (VPS or local PC).
-
Function: It analyzes raw price action and technical indicator values based on the current market structure.
-
Purpose: To identify immediate technical setups with zero latency, ensuring rapid detection of short-term shifts.
2.2 Stage 2: Centralized AI Verification (Server-Side Processing)
Data from the local scan is forwarded to a centralized server for validation.
-
Function: The server utilizes a hybrid model combining TensorFlow logic and GPT-4o inference to perform sentiment filtering and pattern validation against broader market contexts.
-
Economic Rationale: By operating a centralized enterprise-grade API subscription, the developer absorbs the variable token costs associated with AI inference. This model is more cost-efficient for the end-user than requiring individual API keys, which would incur per-tick analysis expenses.
3. Data Transmission and the Middleware Bridge
Communication between the client terminal and the AI server is facilitated by a dedicated middleware bridge. This ensures standardized data packaging and secure transmission.
3.1 The Data Lifecycle
-
Collection: At the close of every H1 candle, the EA collects Open, High, Low, Close (OHLC) data and local indicator values.
-
Packaging: This data is serialized into a lightweight JSON payload.
-
Transmission: The EA utilizes the standard MQL5 WebRequest function to POST the payload to the bridge server at the dedicated IP address: .
-
Inference: The server processes the data through the hybrid model to calculate a probability score for the anticipated market movement.
-
Response: The server returns the validated signal to the EA for execution.
3.2 Privacy Protocol: Data Scoping
To ensure user security, the system strictly limits the data transmitted:
-
Transmitted Data: The payload is restricted to market data (Symbol, Timeframe, Price Data) and a hashed license identifier for authentication purposes.
-
Excluded Data: The system explicitly does not collect Personally Identifiable Information (PII), including account name, balance, broker name, or IP address.
-
Stateless Operation: The inference server is stateless regarding user identity. It processes the mathematical input, generates an output, and does not retain user-specific session data.
4. Transparency and Performance Safeguards
The architecture includes multiple layers of transparency and risk management to protect capital.
4.1 Dynamic Risk Management (News Filtering)
The EA connects to external time servers and news feeds to synchronize trading activity with global economic events. By integrating URLs such as and , the system identifies high-impact news events and halts trading to protect against slippage and spread widening.
4.2 The “Fail-Safe” Mechanism
The system is hard-coded with a dependency on the central server to prevent unverified execution.
-
Safety Mode Activation: If the connection to the AI Server is interrupted (due to internet outage, API timeout, or server error), the EA immediately enters a “Safety Mode.”
-
Default Action: The default command becomes “WAIT.”
-
Execution Logic: The EA is programmed to never open a trade without a confirmed validation response from the server. This ensures capital is never exposed to trades based on incomplete or unverified data.
5. Required Configuration for Connectivity
To enable the data flow described above, the user’s MetaTrader terminal must be configured to authorize outgoing web requests. This is achieved by navigating to Tools -> Options -> Expert Advisors and adding the following URLs to the “Allow WebRequest for listed URL” section:
6. Conclusion
This system is designed to provide retail traders with access to a centralized AI processing network. By distinguishing between a “Black Box” and a “Proprietary Algorithm,” this document aims to clarify the data flow architecture. While the specific weights of the neural network remain proprietary for intellectual property protection, the logic governing connectivity, data transmission, and fail-safe protocols is transparent. This ensures users understand precisely how their terminal interacts with the global intelligence network to generate trading signals.
