FOREX Trading Decision Support Tool
The foreign exchange (FX) market is a global decentralised market that facilitates the trading of currencies between market participants. The FX market is the largest global financial market, with a daily trading volume of approximately $6.6 trillion in 2019. The modern FX market can trace its inception to the disbandment of the Bretton Woods accords in 1970.
The Bretton Woods system governed inter-national monetary management and fixed exchange rates relative to the US dollar. The critical switch from a fixed rate scheme to a floating exchange rate catalysed the formation of the FX market. The FX market operates on a 24/5 schedule, and market participants include central banks, commercial banks,hedge funds, multinational corporations and individual investors (speculators).
Currencies are traded as exchange rate or currency pairs. A currency pair (Base/Quote) is the quotation of two currencies, forming a ratio of the value of the base currency over the quote currency, termed exchange rate. Therefore, EUR/USD is the exchange rate for trading the euro against the United States (US) Dollar.
The exchange rate, determined by supply and demand and other market factors, reflects the quote currency amount required to purchase a single unit of the base currency. The EUR/USD pair is the most liquid (exchanged) currency pair, accounting for approximately a quarter of daily trading volume.
The FX market has high volatility due to multiple erratic economic and geopolitical factors. Therefore, there is inherent complexity involved in trading currency pairs, especially for speculative purposes. Speculative trades are profit-motivated, seeking to take advantage of the increase or decrease in the relative value of a currency, often over short time-frames.
Trading currencies, therefore, carries a high risk to capital for market participants. This risk disproportionately affects individual investors, who over-leverage limited capital or incur debt to trade in the FX market. Without an adequate understanding of the market and relevant economic information, an investor is at risk of rapid insolvency.
The FX market is a dynamic and chaotic system that is difficult to accurately model. The existing literature investigating FX market prediction has focused primarily on Machine and Deep Learning techniques that are black-box and provide minimal decision support. With black-box approaches, the decision-making rationale is opaque and cannot be explained to the user. Further, the problem is formulated as exchange rate prediction in the recent studies rather than providing an explicit Yes/No decision.
Our proposed Bayesian network model is an explainable or glass-box Artificial Intelligence technique in which the model inference can be extracted and explained to the user. Our model is focused specifically on the EUR/USD currency pair, given its liquidity and global demand. The tool will be used to determine whether to buy, sell, or not trade the EUR/USD currency pair to improve the long-term investment decisions of different classes of market participants
Installation
python3 -m venv venv
source ./venv/bin/activate
pip3 install -Ur requirements.txt
FXNet Tests
- Bayesian Network Inference Test
- Decision Network Profit/Loss Backtest
- Decision Network Use Case Simulation
python3 app.py --test
EUR/USD Trading Decision Support
python3 app.py --predict
EUR/USD Trading Decision Support w/ Available Evidence
python3 app.py --predict --evidence
FXNet Interactive Mode
python3 app.py --interactive