Projects/AI & Finance/Intelligent Forex Trading System
Case Study

Intelligent Forex Trading System

Market signals, decoded.

An AI trading assistant for the forex market that analyzes economic news, price data and technical indicators to produce buy and sell signals. By combining fundamental, technical and quantitative analysis, it helps traders decide on data, not guesswork and emotion.

Industry
Finance & Investment (FinTech)
Status
Completed
Timeline
1401–1402 (SH)
Intelligent Forex Trading System
01 / The Problem

Deciding in a volatile, news-driven market.

Forex traders must follow economic news from several countries, price trends and the relationships between markets at the same time. A constant question is how much, and for how long, a piece of economic news moves the price. That is why many traders stay out of the market when news is released and miss profitable opportunities.

Sharp, unpredictable moves when news is released
Fear and greed shaping trading decisions
Deep macroeconomic knowledge needed for fundamental analysis
No human can monitor the market around the clock
High volatility emotional decisions
High volatility
emotional decisions
02 / Our Solution

A suite of intelligent analysis and trading tools.

The system is built from several independent modules. Each analyzes one angle of the market, and together they give a complete picture for decision-making.

Technical details
News Trading
News Trading
Estimates the strength and duration of Investing economic-calendar news on currency pairs and returns buy and sell positions with entry, take-profit and stop-loss
Quantitative Trading
Quantitative Trading
Predicts each pair's price with signal-processing methods and executes trades automatically with a bot
Currency Strength Prediction
Currency Strength Prediction
Daily estimate of each country's currency strength using fundamental and quantitative approaches
Technical Pair Prediction
Technical Pair Prediction
Predicts market direction with indicators and Heikin-Ashi labelling, and computes support and resistance levels by clustering
Best-Pair Suggestions
Best-Pair Suggestions
Daily ranking of pairs in which one currency is getting stronger and the other weaker
Inter-market Correlation
Inter-market Correlation
Measures how currencies relate to each other and to commodities such as oil
03 / System Architecture

From raw data to a trading position.

News, price and inter-market data are collected, analytical models run on them, and the output reaches the user or a bot as a trading signal.

Data collection
(Economic-calendar news, pair and commodity prices)
Analysis & modelling
(Transformer, signal processing, statistical modelling of the news-impact wave)
Signal generation
(Direction, entry, take-profit, stop-loss and approximate entry time)
Execution
(Dashboard display and opening positions directly in a MetaTrader account)
04 / Technical Details

Technical details.

Analysis modules
  • News trading
  • Quantitative trading
  • Currency strength prediction
  • Technical pair prediction
  • Best-pair suggestions
  • Inter-market correlation analysis
Models & methods
  • Transformer models
  • Signal processing
  • Statistical modelling of news-impact waves
  • Signal decomposition (Fourier transform and variational mode decomposition)
  • Clustering for support and resistance
Signal decomposition pipeline
  • Estimate the linear trend and seasonality with a Fourier transform
  • Decompose the non-linear term with variational mode decomposition
  • Select the best lags of each decomposed mode
  • Feed the result to a machine-learning model
Fundamental analysis inputs
  • Risk-on / risk-off indicator
  • Inflation / depression indicator
  • US, UK, Euro-area, German and French government bonds (1, 5 and 10-year US)
  • S&P 500, Dow Jones and major stocks (Tesla, Meta, Google, Amazon, Apple)
  • VIX, CRB and the dollar index (DXY)
  • Correlation of these features with currencies
Backend & delivery
  • Crawling of various financial news sources
  • Temporary web dashboard
  • Telegram signal bots
  • Trader bots that execute trades
Planned improvements
  • Complete currency-index prediction
  • Daily fx-pair selection
  • Ordering of fx-pair trends
  • Complete backend and frontend
  • Testing and tuning
Specifications
News sourceInvesting economic calendar (1–3 star news)
Prediction modelsTransformer, signal processing, statistical modelling
IndicatorsMA, RSI, OBV, Bollinger Bands, MACD
Support & resistanceClustering
Data labellingHeikin-Ashi
Execution linkMetaTrader
05 / Results

Performance on historical-data tests.

The technical prediction model was evaluated on past data of 12 currency pairs. Results are from historical data, not live trading, and past results do not guarantee future profit.

12
Currency pairs tested
80%
Average direction accuracy (Heikin-Ashi labelling, historical data)
6
Economies covered by news trading (US, Canada, UK, Europe, New Zealand, Australia)
6
Independent analysis modules
1000+
Economic news items analysed
25
Symbols predicted by news trading
8
Timeframes
16+
Technical indicators analysed
06 / The Team

The people behind it.

A multidisciplinary team of engineers, researchers and domain experts.

Amirreza Zibafar
Amirreza Zibafar
Machine Learning Engineer
Rozhina Taghdimi
Rozhina Taghdimi
Machine Learning Engineer
Najmeh Vasfinezhad
Najmeh Vasfinezhad
Product Designer
07 / Gallery

More from this project.

Visuals of the system and its components.

Trading system pipeline
High-impact news and average pip movement
Signal decomposition of price series
Economic data screen
Currency-pair trend ordering
Signal-processing core
08 / What's Next

Bring AI to your financial decisions.

If you are looking for a data-analysis or forecasting solution for financial markets, get in touch and let's discuss what you need.

Discuss a similar project
ZLabZLAB

Part of the FANAP ecosystem

FANAP Advanced Technologies Laboratory

Company
About UsOur TeamContact
Projects
All ProjectsIndustriesTechnologies
Follow Us
© 2026 ZLab. FANAP Advanced Technologies Laboratory.