February 11, 2019

Artificial Intelligence: Key Considerations for Success

By Arup Das, CEO & Founder of Alphaserve Technologies

Artificial Intelligence: Key Considerations for Success Intelligent Automation

In today’s digital world, investment and asset management firms are seeing increased pressure to incorporate artificial intelligence (AI) and data-driven tools into their business practices. For many, there is an immediate negative reaction when they hear the term AI, largely due to Hollywood-inspired images of the rise of the machines. When it comes to AI for investment and asset management firms, the reality is far from that Hollywood stigma, or even technology like Siri or Alexa. Institutional AI is far less flashy and is entirely driven by data. A common concern is that these technologies will replace jobs, when really, they are put in place to enhance business processes. Regardless of how many functions you’re able to automate, there will always need to be a human in the loop on every project or process you’re undertaking.

By utilizing the firm’s data, AI can create new avenues for growth and efficiency that previously hadn’t existed. The key is figuring out how best to make AI work for you and executing the plan successfully. In the world of investment and asset management, firms have typically been data-driven; meanwhile their data sources have been more quantitative driven. Recently, we have seen a rise in alternative data sources (data about a company published by sources other than the company itself). This data gives insight on investment opportunities, deriving the alpha of an investment. Alpha is defined as the active return on an investment, gauging the performance of an investment against a market index or benchmark that is considered to represent the market’s movement. Alternative data is considered to be big data, meaning that software is usually used to collect this data. Web scraping and acquisition of raw data are some of the common ways alternative data can be accessed. To this end and to remain competitive, organizations need to integrate AI driven by alternative data into the core fabric of their practices.

Implementing a successful AI program requires a great amount of strategic forethought. To settle on the right program for your firm, it’s crucial to engage in guided design thinking that’s aimed at identifying key areas where the firm can reap the most benefits from AI. It is essential to focus on the data-driven aspects of your organization and incorporate algorithms into your practice management systems to create concrete business cases that can benefit from AI. The most common risk is making the mistake of implementing AI that isn’t targeted at addressing legitimate needs. This only results in the firm incurring unnecessary costs without seeing any tangible benefits or solving problems.

For firms in the investment and asset management industry, there are several areas where these technologies can be put in place to achieve efficiency for business processes. One example would be an automated ingestion and analysis of 10K and 8K forms and earning calls transcripts, where natural language processing and machine learning techniques can analyze customized sentiment and extraction of numerical data from these documents. Firms may also consider using sentiment analysis of private and public portfolio holdings by performing analyses on companies and then correlating those findings to current and historical stock market prices to predict market movements based on current sentiment.

Another critical area where AI can be used is in the search, classification, and summarization of documents. AI tools will search through millions of research reports for analysts and select specific keywords and phrases, automatically categorize them, and then summarize individual documents to a pre-specified length. The process helps analysts to sift through huge volumes of public and private documents more comprehensively, efficiently, and accurately. Furthermore, investment and asset managers may use AI for predictive modeling for portfolio construction and management. This AI program can continuously monitor and access open source and paid news services, such as Reuters’ Lipper system, via an application programming interface (API). This allows for the generation of up-to-date sentiment on desired sectors, subsectors, and companies.

An additional example would be the combination of subjective knowledge, textual data, and structured data. Here, the goal is to create a system that predicts volatility of stock price movement not only based on an investor sentiment index alone, but also enhanced with analysts’ subjective knowledge and patterns extracted from structured financial data.

Finally, firms can use AI to make big data easily accessible via natural language search (demise of the structured data warehouse). AI tools will search distributed, structured, and unstructured “big data” sources by creating on the fly subjective natural language processing to mine internal firm data versus relying on canned data or rigid data warehouse structures. Technical competency or knowledge of the data’s location, content, or format is no longer necessary.

Predicting outcomes and streamlining decision-making is one of AI’s biggest draws, and firms can benefit immensely from insights into the development of markets and potential future trends. Once you’ve identified your key areas for AI integration, the final step is testing to make sure that it’s working before you do a full rollout. The best way to do that is with a proof of concept, which demonstrates how the new solutions will function, using a subset of the data to determine whether the program works or not. If the proof of concept meets expectations, it should become the basis for a prototype that your internal team can test to provide feedback. Skipping the prototyping step means missing out on crucial user feedback that’s necessary to maintaining and optimizing your technology to achieve its greatest utility. Doing a full-scale rollout of untested AI is a sure recipe for failure.

Once you’ve gone through development and are doing a full rollout of your chosen solutions, it’s key to train your staff on how to use them. Most importantly, your training needs to be tailored for each group of users. Investment managers may not get the same levels of training as support staff, depending on how the system will be used. Your technology department might get entirely different training to learn the back end of how things work and how to interact with the consultants who implemented the program.

While a fully trained rollout is unquestionably a huge accomplishment, the work doesn’t end there. You need to constantly maintain your program behind the scenes and update it to accommodate your changing business needs. Successful AI isn’t simply a plug-in. A tremendous amount of customization is required to make it successful, with consistent improvements to ensure that it remains useful and cost-effective.

Vigilant communication is critical to the success of any AI program. It’s easy to get caught up in the fanfare when you first decide to implement AI, but it’s just as easy for that fanfare to die down by the time you actually launch your solutions. If you constantly communicate the benefits of your AI program to your employees and keep them engaged in improving the technology by encouraging them to communicate feedback upward, you’re far more likely to see the long-term benefits of AI.

From the initial stages of building a program to its ultimate full-scale rollout, having an objective, experienced consultant to guide you through the process can mean the difference between wasting money on useless technology and implementing cost-effective solutions that will improve your business. At the end of the day, strategic planning and constant communication are what make an AI program successful. With the right approach, you can ensure that your technology is always working for you, not against you.


Arup Das is the Founder and CEO of Alphaserve Technologies. He is an expert in institutional level technology governance and operational risk management standards that are prevalent in hedge funds, private equity funds, venture capital funds and global law firms.