Machine Learning Applications: The Machine Learning Revolution is Ready to Disrupt Asset Management

Linda Luo is a Financial Analyst at I Know First.

The Machine Learning Revolution is Ready to Disrupt Asset Management


  • Machine Learning Starting to Take Root in Asset Management
  • Why Asset Managers Are Still Reluctant to Adopt Machine Learning
  • The Cyborg Model – A Hybrid Approach for Investment Decisions
  • How I Know First Algorithm Can Help In Investment Decision-Making

Machine learning applications

Machine learning systems excel at observing patterns and making predictions. Insurance firm and investment banks have recognized this technology’s resourcefulness. They are applying machine learning capabilities to streamline processes such as claims processing, fraud prevention, and contract validation. Other industries in financial services are receptive to utilizing AI and machine learning technology. However, machine learning-based systems have yet to gain traction in the asset management industry. 

Machine learning is not widely adopted across the asset management industry. However, this technology is poised to have a significant impact in the field. Asset managers are starting to realize that machine learning will be a key differentiator. This is especially relevant amid the struggle of traditional industry practices to keep up with absorbing and processing the flood of real-time data.

Analytics incorporating machine learning can traverse into territories untapped by traditional financial modeling. For instance, text-based big data sets accumulated from social media and global digitalization, though considered unorthodox and highly unstructured, can be conquered and dissected by machine learning to gain unique insights useful for forming investment strategies. This can also be applied to streams of data coming from corporate press releases, call transcripts, and regulatory filings.

Machine Learning Starting to Take Root in Asset Management

machine learningAsset managers are starting to see the untapped potential of big data as an opportunity to invest in advanced data analytics and machine learning capabilities. For example, Merrill Lynch Wealth Management and BlackRock are among some leading asset management firms that are experimenting with reorganizing their operations around incorporating AI technology to optimize returns and reduce operating costs.

Merrill Lynch is testing an AI stock-picking tool to offer guidance on stock selection to its team of human analysts. In particular, the tool has been useful in identifying small-cap stocks that might have easily been overlooked by Wall Street analysts. BlackRock has also recognized the reality that technology is a crucial tool to investment selection. The company laid out an initiative to restructure its actively managed mutual funds. It plans to rely more on big data and artificial intelligence. It is steering towards using machine learning within quant and traditional investment strategies. The company hopes to use the technology to adopt a more rules-based approach to investing. There are different levels of AI implementation across the asset management industry. But at the very least, machine learning can complement traditional financial analysis.

Why Are Asset Managers Still Reluctant to Adopt Machine Learning?

With the exception of those leading firms, the majority of the asset management industry are still hesitant to completely hand over their portfolios and funds to the discretion of machine learning-based robo-advisors. Research from the Wharton School of the University of Pennsylvania attributed this behavior to the phenomenon of algorithm aversion. The researchers demonstrated that despite algorithmic forecasts consistently outperforming human forecasts, users still refrained from depending on algorithmic decisions. The researchers attributed to users’ distrust of the machine learning systems to the unexplained answers yielded by machine learning algorithms. Even if the machine provided accuracy, users find it hard to trust the system. This is because they do not see an understandable explanation behind the machine’s process in reaching the answer. However, if users feel that they have some kind of control in the decision process, they are more likely to accept the help of machine learning.

The Cyborg Model – A Hybrid Approach for Investment Decision-Making

machine learningIn his work Thinking Fast and Slow, psychologist Daniel Kahneman specified that the human decision-making reasoning is a dual system of thinking fast and thinking slow. Namely, “thinking fast” is automatic and effortless. “Thinking slow” involves deliberation in taking the freewheeling impulses from “thinking fast” and constructing these thoughts in an orderly way.

Thinking fast is more natural and instinctive for humans. However, machine learning systems can outperform us in complex thinking with their ability to perform billions of calculations quickly. Machine learning’s ability to quickly solve complex calculations gives them an advantage in “thinking” more thoroughly in a faster way than humans can.

Thus, this presents a solid opportunity for a natural partnership to emerge between humans and machine learning-based AI assistance. Human and machine can work in synergy when making investment decisions. Asset managers in particular are facing the problem of multi-objective optimization, amid growing competition and faster machine-based trading frequencies. In their quest to optimize alpha, asset managers have to simultaneously manage several objectives. These goals relate to profitability, earnings, risk, investment, and etc., all affected by multiple factors changing in real time. With the assistance of machine learning, asset managers can navigate through these objectives and variables. They can produce complex predictive models that improve every time with feedback. The analytics models, given machine learning capabilities, can adapt to changing conditions and objectives on its own with self-learning algorithms.

The asset management industry is at the cusp of disruption by machine learning and AI technology. However, this does not mean replacement of the human aspect in this field. Investors still desire human financial guidance. According to a survey by Capital One Investing in March, more American investors want digital-human hybrid solutions (69%) compared to strictly robo-advisors (56%).

How I Know First Algorithm Can Help In Investment Decision-Making

Investors may think that a stock’s trend seems random at a glance, but that is a common fallacy about markets being unpredictable.  Markets are chaotic systems with complex dynamics, but they behavior still contains a systemic component. This makes it possible for AI and self-learning algorithms, together with insights of chaos theory, to observe patterns from big data of past stock market activities to forecast future trends in the stock market to a certain extent. Chaos theory is a branch of mathematics that studies the behavior of nonlinear and dynamic systems, such as markets. A previous I Know First article delves deeper into the insights of chaos theory and how algorithms can navigate through the elements of randomness and uncertainty in a dynamic system. 

The I Know First predictive algorithm is a successful attempt at making accurate stock market forecasts through a rules-based approach. The algorithm takes advantage of elements of artificial neural networks and genetic algorithms to analyze non-linear interactions of factors that affect a stock’s price. At first an analysis of inputs is performed, ranking them according to their significance in predicting the target stock price. Subsequently, multiple models are created and tested on 15 years of historical data. Models are refined everyday as the algorithm only keeps the best performing models and rejects the rest, adapting to new conditions everyday as it learns from new data.

The I Know First predictive algorithm is an example of machine learning as a machine-human hybrid solution in generating investment strategies. Using the algorithm as a complementary tool that offers unique insights from the analysis of 15 years of historical data can enhance investment decisions.