Machine learning has become a core component of the finance industry. The landscape is rapidly changing with the increased adoption of artificial intelligence. 

In recent years, the growing field of data science and AI has already started grasping in finance. Commonly used in tasks such as risk management, predictive analysis, fraud detection, and consumer analysis.

That trend is showing no signs of stopping, with a 260% increase in data science teams in financial services since 2018. 

This comes from the results of a report from Refinitiv which was conducted surrounding the industry. The 2020 report shows how business strategies are continuously changing and how Covid-19 has accelerated these results.

Here we’ll take a look at the highlights and key takeaways from the survey. The results of the survey show what the industry’s current state is and can be used by new career candidates to understand the market they are entering.




The survey conducted by Refinitiv was designed to examine the artificial intelligence and machine learning landscape as of 2020. This includes topics such as the level of adoption in financial services, firms’ strategies for various processes, and use cases of ML. The survey also aimed to report the changing role of data scientists in the finance industry and the tools used.

The survey is based on 423 phone interviews conducted between June 29 and August 14 of this year. Respondents included a mix of data scientists, quants, and tech decision-makers across various geographies.

Numerical breakdown of the data surveying data science in the finance industry. [Source: Refinitiv]


The trend is quite clear. The number of teams using AI and ML is increasing significantly. Key findings in the survey suggest that financial firms are scaling AI/ML capability across multiple business areas. They are progressing from proofs of concept to developing mature ML models. 

Around 80% of respondents said that significant investments were being made in the technology. 72% reported it being a core component of their business strategy already.

The survey shows that data-driven businesses are using the technology across the board. There is now an abundance of business use cases for AI and ML with 64% of the data science teams now working in business units. This need will continue to grow as businesses wish to stay competitive and manage risks. 

We also see that financial services firms are utilizing multiple cloud services platforms to run their models. This helps increase their resilience and scale easier. Among the more popular platforms are Microsoft’s Azure and Amazon’s WS which have begun adopting AI-optimized hardware. This tech helps adjust computing power to meet surging demands as they arise.

Given that companies are diversifying their cloud provides, prospective professionals would be well off to do the same. Beginning with learning about the most popular, Microsoft’s Azure. Furthermore, understanding the business goals, error tolerance, and relevant data will be essential to building the AI systems.



The average number of data science roles in financial firms reach 83 in 2020, a 26% increase since 2018. What’s more, 35% of the firms believe their numbers will continue to grow over the next year.

The average number of data science teams that these roles are spread across is 7.1 in 2020. And more interestingly, the number of companies with more than 4 teams increased from 3% to 28% over the two years.


Respondents were also asked about how involved the data science teams were with the decision making process of procuring data. Results show that data scientists are key influencers in deciding on trialing data, with 83% involvement. However when it comes time to buy data sets the decision process is taken over by procurement and executive teams.

What does this mean for prospective data scientists entering the field? They need to have a solid understanding of the AI-data relationship. As some procurement responsibilities shift to data scientists, they need to intimately know what is required to build the systems for the desired business goals. Additionally, the survey shows that the industry demand for professionals is still growing and a good place to look when entering the field.


In the past, deep learning has been seen as mostly academic technology with some niche applications in the tech industry. Implementing such techniques was an expensive endeavor that wasn’t considered worth it. That outlook has shifted as well, with 75% of firms investing in deep learning in particular.

The deep learning framework that is utilized most often is TensorFlow, followed by PyTorch. 

“We use TensorFlow and PyTorch in Refinitiv Labs to build AI/ML models, so it is not surprising to see their growing popularity in the data science community. What is surprising is the percentage of firms using deep learning, which takes time and resources to get right, but can deliver exceptional results,” noted the head of Refinitiv Labs, Geoff Horrell.

Natural language processing (NLP) is a field of AI that analyzes unstructured data such as text and voice data. This technique opens up more possibilities and makes it easier to work with unprocessed data. Its use is becoming more widely adopted as firms are gearing up to get new insights for business decisions.


As the landscape changes so do the challenges faced by most data science teams across the field. The barriers to getting funding and finding the talent to fill the teams are decreasing. This comes as a result of firms investing more in AI systems and placing greater emphasis on the results.

In contrast, the technical hurdles of implementing the systems are still getting more attention. Namely, poor data quality and data availability. As firms continue to use unstructured data the barrier will also grow.

Furthermore, the change in focus is reflected in emerging use cases which enter a whole new level of interaction with the market. These rely heavily on high-quality data and third-party data. To find unique investment opportunities and remain at the forefront of the competition companies will need:

  • Unique data with a history to prove the strategy works
  • Diverse data sources to create new data combinations that your competitors will struggle to replicate
  • The ability to join and link disparate data sets that may not be used in the market today


You can’t discuss statistics and the finance industry in 2020 without touching on the effect that the Covid-19 pandemic has had.

The survey reports that 72% of firms’ existing models were negatively impacted by the pandemic. This is due to the unprecedented data and times. Previously, testing on alternative data sets was considerable risk and effort. Then during the pandemic, models were incapable of handling the economic shifts.

Many firms reported that their models need to be updated and need to be more dynamic. “Our models depend on data quality and a small mistake can lead to a major disaster in the system. Hence, we now need a large amount of quality data sets for the development of new models,” reported a data scientist at a commercial bank in Germany.


To conclude the report, Refinitiv provides a breakdown of how three global regions compare. The regions are the Americas, EMEA, and Asia-Pacific. For the most part, all three were in agreement. The Americas are slightly ahead in the adoption of AI and anticipate the greatest number of increases in data science roles.

The results of the survey given some valuable insights into the current status of the finance industry and an outlook for future trends. The growing importance and emphasis on data to build business strategies will mean a growing need for well trained data scientists.

For the full details check out the report on Refinitiv’s site.

Interested in learning more about the finance industry and data science? Whether you just want to pick up a few new skills or launch a new career, Lantern Institute provides programs in finance and data science to help you get started.

5 1 vote
Article Rating