Artificial intelligence (AI) is now poised to be a disruptive technology, much like the internet was in the 2000s. In recent years, massive improvements in hardware and the flood of data on the internet have led to AI breaking barriers. It will only be a matter of time before AI becomes a significant presence in the financial markets.
Let’s explore AI in trading and see whether it can bring value to traders like you and me.
AI in trading means applying artificial intelligence techniques to predict future market movements and help make trading decisions. Let’s look at exactly what AI is and its key technologies.
MIT research scientist Boris Katz says, “The goal of AI is to create computer models that exhibit ‘intelligent behaviours’ like humans.” These behaviours include perceiving, reasoning, learning, interacting with an environment, problem-solving, and exercising creativity.
You’ve probably interacted with AI even if you didn’t realize it—voice assistants like Apple’s Siri and Google’s Alexa are founded on AI technology, and some customer service chatbots that pop up to help users navigate websites use AI.
Remember, the term “Algorithm” is not the same as “AI.” A traditional algorithm without AI is a static set of instructions, like a recipe, and will not change except through human input.
Much of artificial intelligence today rests on a technique called machine learning, or ML, and the terms AI and ML are often used interchangeably. As Professor Thomas W. Malone from MIT Sloan says, “Machine learning has become arguably the most important way most parts of AI are done.”
So, what exactly is machine learning?
AI pioneer Arthur Samuel defined it in the 1950s as “the field of study that gives computers the ability to learn without explicitly being programmed.” Traditional programming, or “Software 1.0,” requires creating detailed instructions for the computer—like a baker needing a recipe with precise ingredients and instructions. These instructions won’t change unless a human alters them. In contrast, machine learning lets computers learn to program themselves through experience.
Machine learning starts with gathering data to train the software.
Programmers let the computer model train on data to find patterns or make predictions. The human programmer can also tweak parameters to help it produce more accurate results. Some data is held out from the training data to use as evaluation data, like a control set, which tests the accuracy of the machine learning model when it is shown new data. The result is a model that can be used in the future with different data sets.
The more data, the better the program. One of the primary reasons for the recent successes in AI models is that the internet has provided lots of data for practice and improvement. Google Translate, for example, was made possible because it “trained” on the vast amount of online data in different languages. Conversely, if there is insufficient data, AI is useless.
Deep learning is a specialized form of machine learning that uses artificial neural networks—a method that teaches computers to process and learn from data in a way inspired by the human brain. Deep learning has shown remarkable success in tasks such as image recognition, natural language processing, and speech recognition.
The AI technique known as natural language processing (NLP) is crucial to real-time market sentiment analysis. NLP is about having computers understand text and spoken words in much the same way as humans can. Besides speech and text recognition, NLP can extract subjective qualities—attitudes, emotions, confusion, suspicion, and even sarcasm.
Two milestones in the evolution of AI in trading are worth mentioning.
One of the first examples of Artificial Intelligence impacting the world of trading was in the 1980s when two university researchers, K. C. Chen (California State University) and Ting-pen Liang (University of Illinois), designed “PROTRADER.” Their 1989 research paper describes it as “a learning mechanism that allows the system to adapt to the changes in the market it is presented.” Some sources suggest that they successfully predicted the famous 1986 87-point drop in the Dow Jones Industrial Average using the system.
2017: Launch of an AI-powered ETF
In 2017, a team of AI and investment management professionals (Chida Khatua, Art Amador, Chris Natividad) launched an AI-powered exchange-traded fund (ETF), “AIEQ.” The ETS is powered by IBM’s artificial intelligence platform called Watson. Today, AIEQ manages over $100m in assets.
One of the foundational technologies of artificial intelligence is machine learning (ML)—most AI applications today use machine learning at their core. It enables AI systems to learn from historical data and improve their performance over time—much like how human brains learn. ML can be applied across many aspects of trading, including pattern recognition, risk assessment, and predictions about future market movements. The following AI technologies in this list all use machine learning at their hearts.
AI can hunt the internet and analyze news articles, social media feeds, and other sources to gauge market sentiment. Sentiment analysis helps traders understand market participants’ collective feelings and emotions, which can influence asset prices.
Today, financial institutions, hedge funds, and individual traders employ real-time market sentiment analysis to complement their existing strategies and gain an edge.
Key steps involved in real-time market sentiment analysis include:
- Data collection: Gathering data from multiple sources, including financial news websites, social media platforms (Twitter, Reddit, etc.), online forums, and blogs.
- Text processing: Cleaning the gathered data to remove irrelevant information and convert it to a format suitable for analysis.
- Sentiment analysis: AI technology analyses the data and can express the sentiment in different metrics, for example, as an index or a scale from positive, negative, and neutral.
- Sentiment Impact Analysis: Next, the AI assesses how the sentiment scores align with market movements and identifies patterns or correlations between sentiment and price/volume changes.
- Real-Time Updates: Continuously update the sentiment analysis to reflect the latest information and market conditions.
Don’t use market sentiment analysis by itself! Traders rarely use the information in isolation. Instead, most market participants combine it with fundamental or technical analysis for a comprehensive approach to trading.
A trading algorithm is a set of instructions to analyze the markets and sometimes place trades. MetaTrader expert advisors (EAs) are a type of trading algorithm. A traditional algorithm that does not use AI has a static set of instructions that will only change if a human changes its parameters.
An AI-powered algorithm can learn, reprogram itself, and adapt to changing market conditions. One of the significant advantages of using AI-powered algorithms is that they can analyze vast amounts of data and often respond to changes much faster than human traders. The underlying AI technology cases are usually machine learning and deep learning.
With the advantages of AI outlined previously, applying AI to Forex trading strategies can increase performance. Let’s look at the main ways it can do this.
- AI can adapt to market conditions. For example, trend-following strategies perform poorly when the market is ranging. Still, AI can have the power over time as it learns to consider this, for example, by adapting indicator settings.
- Improved Data Processing and Analysis: AI systems can analyze large sets of historical and real-time financial data, including price movements, trading volumes, news, and macroeconomic indicators. These data can identify trends, correlations, and anomalies that human traders might miss or not spot quickly enough.
- Pattern Recognition: AI algorithms can identify and analyze recurring patterns in financial data. This information can make trading decisions based on historical market behaviour.
- Identifying Emerging Market Trends. With the adaptability of machine learning and the ability to rapidly apply learned experience, AI has the potential to more easily alert traders to new conditions and emerging trends.
This is an emerging field, but several companies today offer AI-trading bots. When evaluating these, here are several questions to ask:
- What’s the AI technology driving it? AI is resource-heavy and requires access to high levels of computing power to implement. Make sure the AI trading bot has access to this.
- Where is it getting its data? The quality and quantity of data will determine an AI-trading bot’s effectiveness.
- Does it have a track record? I want to see a real-time track record that has not been manipulated. The longer the track record, the better. This is challenging to find because AI in Forex trading is a new field.
AI is not a magic bullet. Here are some of the challenges:
- AI depends on the quality and quantity of the data it’s trained on. Without sufficient data, AI will give poor results.
- AI can still reflect the biases and shortfalls of the programmers that initially set up the strategy.
- Some companies use the term “AI” loosely, but their technology is not genuinely AI-powered—it may just be a regular algorithm.
I compare AI-powered trading to autonomous vehicles or facial recognition—fabulous technologies that can make certain aspects of our lives quicker and more accurate. Still, they are not in a place yet where they can be used entirely by themselves. Big market moves, especially over long periods, are driven by fundamental macroeconomics or significant shifts in companies’ earnings, and this requires a general understanding of the world and economics that AI does not currently have.
AI is an excellent tool for enhancing decision-making but may require human oversight in many scenarios, just like autonomous vehicles today.
Is there any AI for Forex trading?
Yes, some companies have begun incorporating AI tools that adapt to market conditions in their trading systems and trading bots.
Do AI bots work for Forex?
Sometimes – there are AI bots which have profitable track records in Forex.
Which AI is best for Forex trading?
The best AI for Forex trading is real-time market sentiment analysis and AI bots that use machine learning to adapt to market conditions.
What is the role of AI in Forex trading?
AI can enhance traditional forex algorithms by learning and adapting to market conditions. AI can be a tool or part of a fully automated strategy.
Is AI Forex trading profitable?
Sometimes – some AI tools have profitability in Forex markets.