Machine Learning Applications In Financial Risk Management

Machine learning has become a very popular term. But what is it exactly and how does it apply to financial risk management?

Machine learning is actually a type of artificial intelligence (AI) that provides computers with the ability to learn without being explicitly programmed. Machine learning focuses on the development of computer programs that can change when exposed to new data. The process of machine learning is similar to that of data mining.

Given high volume, accurate historical records, and quantitative nature of the finance world, few industries are better suited for artificial intelligence. There are more uses cases of machine learning in finance than ever before, a trend perpetuated by more accessible computing power and more accessible machine learning tools.

Now, putting that machine learning in the context of financial risk management: external and internal pressures are nowadays requiring financial institutions to reevaluate the cost efficiency and sustainability of their risk management models and processes. Some of the pressure comes, directly or indirectly, from regulators; some from investors and new competitors; and some from the financial institutions’ own customers.

In other words, as financial institutions are feeling the pressure in terms of return on capital and profitability while at the same time being compelled into investing more in risk management, artificial intelligence and machine learning could bring a lot of relief for banks faced with these challenges. Indeed, the share of compliance and risk in total banking costs is bound to rise from an average of 10% to 15% while concomitantly RoE is in the single digits at around 8.5%.

We could summarize five areas that directly revamp the way how risk management is being conducted in financial institutions by making use of artificial intelligence (AI):

In practice, machine learning applications can be found in an array of risk management processes, such as the following eight:

1. Machine learning can help financial institutions determine the credit worthiness of potential customers, or in other words underwriting can be described as a perfect job for machine learning. By analyzing past spending behavior and patterns, a system could identify how much credit should be extended to a given customer. The technology would be especially useful in the case of new customers or those who lack a long credit history, i.e. millennials. Automating credit and risk scoring processes on a mass scale can help banks enhance their credit and risk scoring models across the board. Especially at large companies (big banks and publicly traded insurance firms), machine learning algorithms can be trained on millions of examples of consumer data (age, job, marital status, etc…) and financial lending or insurance results (did this person default, pay back the loan on time, get in a car accident, etc…?). The underlying trends that can be assessed with algorithms, and continuously analyzed to detect trends that might influence lending and insuring into the future (are more and more young people in a certain state getting in car accidents? Are there increasing rates of default among a specific demographic population over the last 15 years?). These results have a tremendous tangible yield for companies – but at present are primarily reserved for larger companies with the resources to hire data scientists and the massive volumes of past and present data to train their algorithms.
2. When talking about portfolio management, the term “robo-advisor” was essentially unheard-of just five years ago, but it is now commonplace in the financial landscape. The term is misleading and does not involve robots at all. Rather, robo-advisors (companies such as Betterment, Wealthfront, and others) are algorithms built to calibrate a financial portfolio to the goals and risk tolerance of the user. Users enter their goals (for example, retiring at age 65 with $250,000.00 in savings), age, income, and current financial assets. The advisor (which would more accurately be referred to as an “allocator”) then spreads investments across asset classes and financial instruments in order to reach the user’s goals. The system then calibrates to changes in the user’s goals and to real-time changes in the market, aiming always to find the best fit for the user’s original goals. Robo-advisors have gained significant traction with millennial consumers who don’t need a physical advisor to feel comfortable investing, and who are less able to validate the fees paid to human advisors.
3. It can be applied in early warning systems (EWS), for example. Here it can enable deeper insights to emerge from large, complex data sets, without the fixed limits of standardized statistical analysis. At one financial institution, a machine learning–enhanced EWS enabled automated reporting, portfolio monitoring, and recommendations for potential actions, including an optimal approach for each case in workout and recovery. Debtor finances and recovery approaches are evaluated, while qualitative factors are automatically assessed, based on the incorporation of large volumes of nontraditional (but legally obtained) data. Expert judgment is embedded using advanced-analytics algorithms. In the SME segment, this institution achieved an improvement of 70 to 90% in its model’s ability accurately to predict late payments six or more months prior to delinquency.
4. Hedge funds hold their cards tight to their chest, and we can expect to hear very little by way of how sentiment analysis is being used specifically. However, it is supposed that much of the future applications of machine learning will be in predictive analytics through understanding social media, news trends, and other data sources – not just stock prices and trades. The stock market moves in response to myriad human-related factors that have nothing to do with ticker symbols, and the hope is that machine learning will be able to replicate and enhance human “intuition” of financial activity by discovering new trends and telling signals.
5. Globally, more than a thirty financial institutions have replaced older statistical-modeling approaches with machine-learning techniques and, in some cases, experienced 10 percent increases in sales of new products, 20 percent savings in capital expenditures, 20 percent increases in cash collections, and 20 percent declines in churn. The banks have achieved these gains by devising new recommendation engines for clients in retailing and in small and medium-sized companies. They have also built microtargeted models that more accurately forecast who will cancel service or default on their loans, and how best to intervene.
6. In the field of fraud detection & compliance there is also a huge interest in incorporating artificial intelligence. Up until recently, compliance was conducted manually and by relying on a selected data sources. The old system was often inconclusive and often more useful for reconstructing incidents that were already detected. In contrast, the use of machine learning in compliance activities has essentially shifted the paradigm away from a risk-auditing methodology based on backward looking sampling to a more comprehensive and continuous monitoring. This provides several advantages. First, the approach is more efficient and allows the bank to do more with less manpower. Second, it is more effective. The fact that incidents can be detected earlier allows the bank to prevent them from spiralling out of control. Third, the system is adaptive. Humans have a great capacity to adapt to controls imposed on them. In contrast, policies adapt at a much slower rate to changes in practices and business conditions. The new system’s learning capability helps address this problem. This creates a positive impact on organisational culture by reducing the bureaucratic burden created by meaningless controls and by protecting social norms through the detection of early deviations.
7. With origins going back to the 1970’s, algorithmic trading (sometimes also know as Automated Trading Systems) involves the use of complex AI systems to make extremely fast trading decisions. Algorithmic systems often making thousands or millions of trades in a day, hence the term “high-frequency trading” (HFT), which is considered to be a subset of algorithmic trading. Most hedge funds and financial institutions do not openly disclose their AI approaches to trading (for good reason), but it is believed that machine learning and deep learning are playing an increasingly important role in calibrating trading decisions in real time. There some noted limitations to the exclusive use of machine learning in trading stocks and commodities, see this Quora thread for a good background on machine learning’s role in HFT today.
8. Another example is anti-money laundering (AML) compliance. Trade finance, one major area of AML monitoring, is traditionally supported by heavy documentation that is more or less manually reviewed for compliance. Big data analytics can similarly support the detection of trade anomalies through the monitoring of activities, networks and trends.

Generally speaking, there is an emerging recognition in the financial services sector that leveraging advanced technologies, such as artificial intelligence and machine learning, is the key to deriving real value from big data infrastructure.

Naturally, like any other innovation, the new approach is not a panacea. For example, although algorithms used to manage risks can be described in general terms, understanding and perhaps more importantly explaining exactly how they work is extremely challenging.

Regulators, executives, auditors or clients without a technical background may be wary of relying on these new oracles. Data scientists are currently in hot demand but their technical skills will gradually become a commodity. However, the capacity to mesh hard and soft skills will continue to carry a premium. Perhaps paradoxically, the technicity of the new tools has made the combination more valuable.

Indeed, the new technologies may have made the human element of risk management more important!