It is estimated that between 60 to 75 percent of trading on all major stock markets around the world is algorithmic, dominated by high frequency trading, bots, algo trading, algo portfolio management, and increasingly, the use of Artificial Intelligence (AI) in algorithmic trading and finance.
While it is difficult to put a precise figure on the percentage of algorithmic trading that uses AI, in recent years, Machine Learning (ML) has dominated the industry using big data trained on learning networks to statistically risk adjust individual trades to entire portfolios.
While much of the use of this technology is in the hands of market professionals, knowns as “quant traders” or “quants”, it is now increasingly becoming available to “retail traders”, and solutions are starting to emerge for crypto trading.
ChatGPT has heralded the next era of heuristic technology using Large Language Models (LLMs), generative text models that are optimized for a variety of natural language generation and specialized tasks. These models are characterized by their large size and enabled by AI accelerators which process vast amounts of data through data scraping from the Internet.
LLMs such as ChatGPT help interpret charts, trends, and financial indicators and deliver results in an understandable form. The analysis can support traders with their understanding of market dynamics, risk factors, and investment opportunities.
Cryptocurrency exchange Coinbase Global now uses ChatGPT for risk analysis and screens any new digital asset being added to its platform with the help of ChatGPT. There is crypto bot Omni for the Solana blockchain that can support passive income strategies such as crypto staking and claims to be an “expert” on DeFi. 3SingularityNET offers a range of AI bots which can be used for market and data analysis.
It’s even possible to create your own crypto trading bot using ChatGPT, should you choose. Crypto trading is changing and a major factor behind that change is the application of AI and language learning models emerging in the market.
GNY.io’s machine learning tool is designed to forecast the volatility of the top 12 cryptocurrencies by using multiple data points and advanced algorithms. GNY’s Range Report uses ChatGPT and Meta’s Large Language Model Meta AI 2 (LLaMa 2) and Long Short-Term Memory (LSTM), models that specialize in analyzing sequential data to better understand time-series financial information.
The models support market and technical analysis and recognize price (chart) patterns and indicators to generate trading signals and provide price predictions. The models can also forecast volatility and price trends of assets including cryptocurrencies, stocks, and commodities.
GNY.io has developed a model to learn from three years of trading patterns in data across 25 charts and indicators, which it uses to forecast the next seven days of crypto volatility and claims the algo averages over 95 percent accuracy.
Crypto Is A Volatile Asset Class
It is from volatility that many professional traders make much of their money and volatile assets and markets are a big attraction factor to traders. Cryptocurrencies have been through spectacular booms and busts in the past few years with prices surging to all-time highs and then falling back. Many traders look at most cryptocurrencies as speculative, and the volatility with momentum offers traders many opportunities.
Cryptocurrency volatility is currently not what it was, with major currencies like bitcoin and Ethereum up 50 and 30 percent respectively year to date, trading volumes remain historically low. Last month saw bitcoin exchange volumes hit a near five-year low.
Alissa Ostrove at CCDATA says, “Bitcoin and Ethereum have marked a historic low in volatility, despite seeing the largest liquidation event since FTX on August the 17th.”
Periods of low trading volume in crypto have often been followed by bull runs. The fall in trading volumes is usually seen when traders are sitting the market out as prices drop. Many are of the view that traders are looking for a reason to return to the market after a strong start to the year.
Ostrove adds, “As low volatility starves investors of short-term opportunities, it also encourages accumulation of fundamentally-strong assets in anticipation of the next market cycle.”
Market pundits are looking at the impending approval of bitcoin ETF applications from large asset managers like Blockrock, Fidelity, WisdomTree, Invesco Galaxy, and others to drive the next cryptocurrency bull market. Grayscale’s recent victory in a countersuit to convert its Grayscale Bitcoin Trust (GBTC) into a listed bitcoin ETF had been previously rejected by the SEC and has market watchers predicting we are months away from bitcoin ETF approvals.
Robots To Overcome Human Emotion
GNY.io has commissioned a global research study with retail traders in the U.S., U.K., Germany, Brazil, Hong Kong, Singapore, the UAE, and South Africa that indicates there is strong demand for AI and machine learning tools along with a growing acceptance of their potential in trading in the retail market.
95 percent of survey participants would trade more than the $5,000 month minimum floor they were currently trading if they had access to AI and machine learning tools for trading. On average, the study found traders would increase trading by 16 percent if they had confidence in AI tools which could detect patterns in trading and predict price movements. Nearly three out of four traders believe they would benefit from using AI and machine trading tools to detect patterns and predict price movements.
The big issue identified by the research highlighted the quality of and access to trading data. Only 29 percent of traders rated the data sources they currently use as excellent with crypto platform and exchange news services the most used. Most traders are not impressed with the data sources they currently have access to – there is a lot of noise and a lot of information to sift through.
Reading the market and being able to identify trends is crucial for any trader hoping to have success in the crypto markets but tuning out the noise of competing data and being able to control your emotions when trading is difficult. This is one of the reasons why AI is increasingly playing a bigger role in crypto trading just as it does across financial services as a whole.
Cosmas Wong, CEO GNY, says, “Although we see AI’s role in trading to be inevitable, they can never replace a human’s intuition and oversight. We’ve been working with these models now for a long time and we know the strengths and weaknesses associated with them.
“Like humans, we’ve learnt how to work with and accommodate their weaknesses, and we bank on their strengths. AI tools will need to be supervised – humans are good at certain things, machines at others. We see a collaboration to achieve the best outcomes.”
One important factor about crypto trading that AI and machine learning addresses is the impact of emotion, general sentiment, and bias when trading. The research found nine out of ten retail traders admit emotion has some impact on their crypto trading with a quarter of traders admitting it is the driving force.
Emotion loses money for traders – the study found that on average traders estimate one in five, or twenty percent, of their loosing trades are driven by emotion rather than rational decisions based on data analysis.
More than half of the traders say that breaking news on companies involved in the sector, such as the collapse of FTX or Genesis, is the biggest factor driving trading while 60 percent believe social content has a huge impact on crypto prices. Three quarters of traders surveyed say they look to trade during periods of increased volatility and liquidity such as when the U.S. and U.K. markets open.
Emotion can be eliminated when traders switch to automated “programmatic” trading if the program has been tested for performance and a predictable outcome. An AI-driven algo trading bot can make trades based on a predefined program strategy and react quickly to changes in the market.
In the 1970s, the legendary Richard Dennis with his Turtle Trading System, promised to turn anyone with no finance or trading experience into a world class commodities trader in two weeks as long as they could follow the rules of the trading program. The turtle traders that followed the rules were successful, those that let emotions, sentiment and bias get in the way were not.
It is, however, important to bear in mind that programs like ChatGPT are in beta and still being tested, so longer-term performance remains unanswered. It has only been trained on data up until September 2021 and can provide inaccurate results.
Also, relying on one AI program to do everything in the crypto market has a risk – programs have limits to what they can predict. For example, the FTX collapse, which had nothing to do with crypto prices or volatility, was a significant governance failure, and additionally, there are always potential data security and privacy concerns with AI.
AI offers a suite of tools that are growing in importance, but human traders are most often the architects of “program strategies” which need research and back testing. Traders will always remain pivotal in the decision-making process, but need to extend their capabilities with AI’s quantitative power to deliver consistent risk adjusted strategies and returns – man and machine.
As U.S. technology heavyweight CEOs meet senators in Washington this week, legislation to regulate the AI industry is high on the agenda and is something that both industry and policymakers agree needs to be implemented expediently. The A-list of tech CEOs includes Telsa’s Elon Musk, Facebook’s Mark Zuckerberg, with CEO’s from Google, Microsoft, and IBM, joined by union leaders and civil rights activists, convened by Senator Chuck Schumer (D., NY.).
While big dogmatic discussions will revolve around “Open AI”, which is “open” to manipulation and hacking by state sponsored terrorists seeking to disrupt democratic capitalism, versus the “closed” development of AI by corporations and NGOs, one thing is for sure – the AI trading revolution in financial services is just getting started, and retail crypto traders are its latest beneficiary.