Hedge funds utilising artificial intelligence capabilities have shown a competitive edge over investors that didn’t use AI, according to new research. The coronavirus pandemic has given partial proof of the effectiveness of the application of artificial intelligence as a predictive tool in fund management; reveals the latest issue of The Cerulli Edge―Global Edition.
An examination by Cerulli Associates of the assets under management (AUM) of various funds and net new flows of Europe-domiciled AI-enabled funds from 2013 to April this year reveals substantial AUM growth from 2016 to 2019. The aggregate return of AI-led hedge funds was almost three times higher than that of the overall hedge fund during this time: 33.9% compared to 12.1%.
Despite this, AI-powered hedge funds’ net new flows dropped somewhat last year, before dropping sharply mid-January and April. Nevertheless, Cerulli’s research tells that European AI-led active equity funds increased at a quicker rate than the other active equity funds from January to April this year and presented a less-pronounced slump in March.
AI Vs Humans For Managing Investments
Large investment funds around the world also have started using AI extensively to manage funds and augment their processes and capabilities. AI agents which utilise machine learning algorithms can process data in ways that humans fundamentally can’t. It shouldn’t surprise anyone that these machines will one day be ubiquitous in both portfolio management and industry.
But the challenge of using machine learning is that its predictive value is only as good as the historical data that you have. While it can find patterns and predictive insights in historical data, it might be a challenge in volatile situations such as the rare global pandemics.
Much of it is about identifying cyclical trends in the marketplace. If an event hasn’t happened in the recent past, the technology is not going to be as effective. There are millions of things happening in the marketplace every single day, and AI allows investment firms to gauge all the data using ML models and create useful insights for clients.
Quantitative hedge funds have been utilising algorithms for years to make trade decisions. The algorithms, however, were not able to perform with the volatility of financial markets as they were used in static models and thus yielding bad returns.
Most of the ML algorithms employed in the finance industry are supervised, and the model learns to recognise patterns by analysing provided historical data points. With the pandemic and the global lockdown being a novel and unforeseeable situation, it is tough for the models to adjust to the numerous scenarios dynamically, describes Cerulli.
Of course, until the turbulence caused by the pandemic is over, it is not likely to draw definitive judgments on the effectiveness of AI during the period. The report finds that with the training, the ability of models to predict dynamic novel changes remains challenging for two main reasons:
- The short and scattered history of financial data: Only the past few decades can work as a training example. A big part of human history (such as new pandemics) is unknown to the algorithms.
- The chaotic nature of financial markets: In such scenarios models may not work as expected.
New Hedge Funds Using Autonomous AI Models
This is where, according to Cerulli, ‘pure’ AI models — self-sufficient models that don’t require human programming — come into the picture. These models, unlike pre-AI quants, adapt to changing market conditions with significantly greater autonomy as they don’t rely on a human component.
According to Justina Deveikyte, associate director of Cerulli, AI has finally made an impact on the world of investing and outperformed other investments. “There have long been doubts about the capacity of AI to respond to unexpected events, like the COVID-19 pandemic. But there is now a sense that the technology has progressed to the point where it is properly able to accommodate unforeseen scenarios through the ever-expanding amount of market data available.”
New AI model can automatically update itself as it gathers new data, with no need for human oversight – this could be an especially pertinent benefit in an uncertain market. Many of these AI systems dig through various social media venues to gauge market sentiment on a particular asset or security. Some hedge funds are even using technologies that scan keywords in news articles to help predict any rise and fall in financial markets.
For instance, Project One hedge fund is created to generate performance and provide robustness toward volatile markets because of unforeseen events. According to Project One, their system uses an alpha-learning AI model which continues to adapt and update itself without human involvement for manual data collection and processing.
During the testing phase, Project One claims to use its proprietary “alpha-learning” AI market fund algorithm and achieve a return of 160 per cent, which was delivered over three months, reporting limited downsides. Unlike several other hedge fund AI systems, Project One also analyses, projects, and uses volumes of direct and peripheral market activity data and makes real-time orders without human interaction.
Subscribe to our NewsletterGet the latest updates and relevant offers by sharing your email.
Vishal Chawla is a senior tech journalist at Analytics India Magazine and writes about AI, data analytics, cybersecurity, cloud computing, and blockchain. Vishal also hosts AIM's video podcast called Simulated Reality- featuring tech leaders, AI experts, and innovative startups of India.