Layered market analysis
A unified terminal, market regimes, ICT, support/resistance, indicators and news provide context for each signal.
ARG Trading Platform
AutoTrade / ArgTradeAI — Intelligent Gold Trading
An end-to-end AI gold (XAUUSD) trading platform: 17+ ML models, institutional-grade ICT analysis and fully automated execution on MetaTrader 5.
ArgTradeAI
AI Algorithmic Trading
It automates the full cycle: analysis → signal → risk check → execution → position management → learning. 15+ market features processed every 5 seconds.
FastAPIPyTorchReact 19GMM + XGBoostMT5 / EA
Open platform17+
ML Models
12
Regimes · 95.9%
342
API Endpoints
108K+
Lines of Code
6
Production Systems
11
Active Strategies
Platform in action
What makes it different
After reviewing 20+ global platforms, none combine these eight capabilities — in ArgTradeAI they work together in one closed loop.
Self-Improving Loop
Outcome records, PSI drift monitoring and retraining paths form the model-evaluation loop. Model authority depends on version and configuration; Comp Entry ML is observation-only in the catalog baseline.
LLM gate in the signal path
Enabled paths use configured LLM providers (Groq / Claude / Gemini) to return CONFIRM, SOFT REJECT or HARD REJECT. Gate activation and authority follow the policy of that path.
12-state regime (95.9% acc.)
A GMM + XGBoost model; risk, trailing and strategy adapt per regime. Competitors use only 3–4 states.
ICT / Smart Money + ML
6 institutional concepts (Order Block, FVG, Liquidity Sweep, …) scored by XGBoost + LightGBM — the only automated ICT+ML system.
Shadow Trading
Every strategy runs as paper trades on the live market first; only after statistical validation (PF / win-rate / drawdown) does it receive real capital.
On-device ML in MT5
The on-device inference path embeds regime classification in supporting EA versions, avoiding a network round trip for each prediction. Availability depends on the installed EA version.
5-layer risk + Kill Switch
Per-trade limits, a daily loss cap, a max-drawdown circuit breaker, and a global emergency Kill Switch (≤30s propagation).
“Most-protective-wins” trailing
Every tick, three stages (Break-even → Lock → Trail) are evaluated independently and the tightest SL wins; SL only moves forward.
Capabilities & Sub-systems
Market Regime Classifier
Multi-Strategy Engine
AI Signal Validator
Execution Bridge
Fundamental Analysis
Production-clone Backtest
Multi-Timeframe Dashboard
Live Market Data
Smart Stop-Loss
Telegram & Alerts
No-code Strategy Builder
Role-based Access
Revenue Model — Three Escalating Tiers
Value per customer grows from tens of dollars (retail) to millions (banking) — each tier modeled on a proven global benchmark.
Retail
Benchmark: TradingView
Revenue potential (at scale)
~$50M / yr
$15–$149 per user / month
e.g. 100K paying users × ~$40
Freemium cloud subscription → monthly Pro/Premium plans.
Customer: Individual traders
Monthly subscriptionPremium tierAdd-ons
Broker & Pro
Benchmark: White-label + Data API
Revenue potential (at scale)
~$100M / yr
$50K–$500K per broker / year
e.g. 200 brokers × ~$500K
Signal/data licensing, API and white-label for brokers and fintechs.
Customer: Brokers & fintechs
Data/signal licenseAPIWhite-label
Enterprise & Banking
Benchmark: Bloomberg Terminal
Revenue potential (at scale)
~$250M+ / yr
~$25K per terminal seat / year
e.g. 10K institutional seats × ~$25K
Institutional terminal + data feed + FATF/AML compliance and Core-Banking integration (B2B for banks).
Customer: Banks & financial institutions
Terminal seatsData feedCompliance/AMLEnterprise contract
Ref: Bloomberg with ~325K terminals ≈ $10B+/yr
Figures are model potential based on per-unit economics and global benchmarks (not current revenue) and may be higher or lower depending on market adoption. The banking tier depends on the required FATF and Core-Banking access and licensing.
ARG / PRODUCT & TECHNOLOGY
AutoTrade technical guide & execution workflow
ARG’s platform for market analysis, signal generation and control, and coordinated trade execution in MetaTrader 5. Web application, backend engines, desktop client and Expert Advisor connect one complete workflow.
Catalog v2.0 · 80 pages per language
FROM INFORMATION TO OPERATION
A shared market picture. Traceable decisions.
Analysts inspect market context, operations teams manage accounts and risk, and strategy teams develop and evaluate logic. AutoTrade connects these roles through one workflow.
Independent strategies
Scalp, Swing and Position horizons in Comp, alongside the strategy library, reversal and momentum logic and final USL controls.
Visual strategy authoring
AND/OR conditions, sequences, lookback and wait-for rules use analytical variables and contribute evaluation evidence to the engine.
Risk controls across layers
Strategy geometry, account exposure and broker-symbol constraints have distinct scopes; actual volume is checked by the EA.
Account and operations visibility
EA readiness, feed health, signal delivery, execution events and component versions sit alongside settings and access controls.
AI assistance and evaluation
Market-grounded assistance, conditional alerts, research tools, replay and shadow paths distinguish observation from trade execution.
THE PRODUCT WORKFLOW
From market data to execution reporting
- 01
Trusted data
Quotes and candles checked for provenance, UTC contracts and feed freshness.
- 02
Market context
Structure, regimes, levels, indicators and news in a shared market hub.
- 03
Decision & strategy
Comp and StrategyEngine with independent state and policies.
- 04
Risk & delivery
Geometry and account eligibility, durable queues and signal expiry.
- 05
MetaTrader execution
The Bridge, file contract and EA handle volume, orders and positions.
- 06
Feedback & evaluation
Match delivery state and execution outcomes to account and signal identity.
INSIDE THE PRODUCT
Analysis and control in one workspace
Quotes, direction, signal horizons, levels and market state sit alongside analytical detail, supporting an overview followed by deeper inspection of each decision.
STRATEGY BUILDER
Make strategy logic visible
Author logical and temporal rules over analytical inputs, inspect risk and level settings, and follow evaluation. The Builder contributes evidence to the signal engine; drawing a graph alone does not submit a broker order.
EXPLICIT OWNERSHIP AT EVERY LAYER
Four layers. One decision and execution path.
Data & analysis
MarketDataHub · Comp · StrategyEngineValidated feeds and shared snapshots support structure, regimes and independent signal candidates.
Policy & persistence
USL · PostgreSQL · RedisGeometry and quality controls, account eligibility and durable delivery with identity and TTL.
Delivery & connectivity
REST / WebSocket · ParsGold BridgeAccount-scoped delivery and conversion of server messages into atomic terminal-local files.
Expert Advisor & broker
MQL5 · MT5 · BrokerVolume and order checks, position management and actual broker-outcome reporting.
Feedback: delivery state → execution event → account and signal reporting
A transport acknowledgement is distinct from an executed order. Account and signal identity, expiry and the broker response must be assessed together.
PARSGOLD / BRIDGE + EXPERT ADVISOR
Execution inside MetaTrader. Connectivity in the Bridge.
ParsGold Bridge manages network communication and file exchange. The MQL5 EA consumes the local signal and owns volume checks, order submission and position management.
- 01
Read & validate
The JSON contract, signal identity, freshness, direction and account state are checked.
- 02
Volume & order geometry
SL/TP, estimated loss at the stop, lot minimum and step, and symbol constraints are checked.
- 03
Order & position protection
Supported orders reach the broker; BE, Lock and Trailing run in the terminal under the associated strategy policy.
- 04
Outcome & active version
Execution events, position state and heartbeat are reported. Installing a file is distinct from proving EA-version activation.
Account policy and execution settings
Strategy settings, access and account risk have separate scopes. An online connection alone does not establish trading readiness; entitlement, versions and terminal configuration are checked during deployment evaluation.
Explore the full catalogCapability authority is explicit: Comp Entry ML is observation-only in the catalog baseline, position-management analysis is read-only, and daily-policy enforcement depends on activation and account scope. Enabled strategies and features depend on deployed versions and configuration.
PRODUCT & TECHNOLOGY BOOK / 2.0
Explore the product in depth
Each 80-page edition covers the product, backend mechanisms, frontend, Expert Advisors, protocols, 61 web-page views and 14 desktop views.
فارسی / PERSIAN
AutoTrade product & technology catalog — Persian
Version 2.0 · 80 pages · PDF · 14.6 MB
Download PDF
ENGLISH / انگلیسی
AutoTrade product & technology catalog — English
Version 2.0 · 80 pages · PDF · 13.0 MB
Download PDFThis introduction follows the catalog baseline. Software screenshots use synthetic data; the book contains no return guarantee or audited profitability report.