Join the Community

22,042
Expert opinions
43,974
Total members
375
New members (last 30 days)
176
New opinions (last 30 days)
28,689
Total comments

Contextualizing AI & ML in the Investment Accounting Space

In this article, I aim to shine a light on the application of artificial intelligence and machine learning to enhance investment accounting capabilities.

How is Artificial Intelligence Used in Investment Accounting? 

Let me just start by defining AI and ML in a way that is specific to investment accounting.

Within any investment accounting system, there are volumes of data that need to be managed and made fit for purpose for multiple business views. When we think about the role AI and ML play in investment accounting, this role largely has to do with how we store and interpret data within the system. These technologies are also vital for improving investment accounting operations and workflows.

Before diving deeper into this concept, it is important to note the distinctions between AI and ML. 

AI is the broader field that encompasses all intelligent machines, systems, and programs that can simulate human intelligence. ML, on the other hand, is a subset of AI that is more specifically focused on developing algorithms and learning models that enable AI systems to automatically learn from data ingested and improve future performance based on new information. ML algorithms can be designed to learn and evolve from user feedback, for example, highlighting exceptions and recognizing trends in how different types of exceptions are handled. When we consider this in the context of investment accounting, AI and ML can help create smart investment accounting systems that quickly ingest new data, learn from it, and report key findings and insights back to the users.. 

To contextualize this within a specific investment accounting scenario, let’s think about how AI and ML can be mutually beneficial to a system that uses data lakes. 

A data lake is a repository that can store very large amounts of structured, semi-structured, and unstructured data without requiring pre-defined schemas. Data lakes are advantageous for investment accounting systems due to the flexibility and scalability they offer. 

When paired with AI and ML algorithms, a data lake can not only provide a vast amount of centrally stored data but also enable users to make faster and more accurate decisions based on advanced data analytics capabilities that can produce real-time alerts and recommendations.  

As for how this benefits AI and ML, these technologies work best when given access to a massive supply of data. Given that data lakes are specifically designed to store vast quantities of data, this provides the ideal data environment for learning and evolution to take place. 

What Makes an Investment Accounting System “Smart”?

A smart investment accounting system leverages AI and ML to enhance its processing capabilities. Ideally, a smart system should integrate both traditional statistics and machine learning models. This dual integration enables far greater efficiency and accuracy through  a low- to no-touch process that reduces the amount of tedious, error-prone work that has traditionally been a hallmark of the very manual investment accounting process. For example:

  • Reduced manual labor required in the data mapping process; 

  • Automatic price checks and analysis of factors impacting a specific security in real-time (e.g. trend, history, market conditions, etc.); 

  • Automatic identification of anomalies and exceptions that trigger a designated review and approval workflow.

Not only could AI and ML be applied and trusted to intelligently auto-investigate and auto-validate today's common valuation anomalies such as big price fluctuations, but when the AI engine is an inherent part of a modern investment accounting system, then the concept of “smart accounting” truly starts to evolve, creating preemptive accounting controls, such as:

  • Security and Compliance, e.g., quickly restricting user access by tracking large deviations to routine user menu to menu patterns; 

  • Management decision-making: e.g., creating portfolio management alerts when projected or existing net new fund capital values do not tightly correlate to net new trading values in core asset types.

Today’s advancements in AI enable smart systems to not only automatically perform statistical analyses but to also learn from the results and generate tailored recommendations. And where accounting controls have historically been reactive, in the new era of smart investment accounting, it is now possible for those controls to be preemptive.

Building Confidence in AI

The lack of human intervention is a very fair and understandable concern to bring up when it comes to using AI in the investment accounting process. 

What is most important in our use of AI within our platform capabilities is to ensure technologies actively gain an understanding of specific factors impacting a change in the market, such as a price change for a security. 

For example, let’s imagine two different securities each experienced a 50% price movement in a same-day period. On a surface level, these price movements could trigger an alert within the investment accounting system to inform the investment accounting team of the movement and prompt an action in response to the alert.

Let’s say that one security experiences a 50% price movement due to good news, such as a pharmaceutical company that just received FDA-approval for a new product. Meanwhile, the other security experienced a 50% price movement due to bad news regarding a company sale.

An investment accounting system needs the ability to understand the difference between these two scenarios and provide different prompts based on the specific circumstances. 

Likewise, for an investment accounting team to gain confidence in this system, there needs to be the added layer of approval, also known as a “four-eye check” by designated users before the system can act on a recommendation. 

What is also incredibly helpful for building trust in an AI-powered system is the flexibility to set custom rules and parameters that determine how the system operates. For instance, one could set a rule to receive alerts every time a security experiences a 5% price movement, enabling them to give feedback and final approval on even minor events.

This confidence only builds from here, as each response to each alert becomes a new piece of information for an AI-powered system and ML algorithm to learn from. 

Over time, this algorithm will become more and more tailored, learning specific behaviors and preferences along the way. Eventually, this can allow you to automate certain activities while still requiring a final human approval on others. 

External

This content is provided by an external author without editing by Finextra. It expresses the views and opinions of the author.

Join the Community

22,042
Expert opinions
43,974
Total members
375
New members (last 30 days)
176
New opinions (last 30 days)
28,689
Total comments

Trending

Now Hiring