Using AI to predict market reactions

The integration of artificial intelligence into investment research and portfolio management is becoming a competitive challenge for asset managers. For its part, Amundi is developing a tool for aggregating and interpreting information published about companies it tracks.

Analysing stock markets on the basis of data such as prices, fundamentals and trends only provides an incomplete picture. Based on this observation, the American startup Causality Link, located on the outskirts of Salt Lake City, applies artificial intelligence and natural language processing to better understand the financial markets. The company convinced Horizon SA, a French family office, to invest $5m in its technology in April 2020.

That was just beginning. Causality Link quickly established a partnership with the asset manager Amundi, as well as the Toulouse School of Economics, “interested in its innovative textual analysis method”, said Marie Brière, head of investor intelligence & academic partnerships at Amundi Institute. On 13 January, she co-authored an article on the capacity of artificial intelligence to anticipate stock market price trends. It was based on the analysis of 8,000 sources, namely press articles, broker analyses and company communications. The Toulouse School of Economics is also active in the partnership and contributes to the research project. Its president, Jean Tirole, who was awarded the Nobel Prize in Economics in 2014, sits on the Causality Link advisory board.

Corporate health in real time

Technology is beginning to be integrated into Amundi’s processes. “Analysts and portfolio managers have tools that allow them to monitor news and have access to signals aggregating in real time the degree of positivity of the news that hits a company,” said Brière. However, she qualified that: “We are testing strategies and the added value that this tool can have for management, but today there are no portfolios that are systematically managed on this type of signal.” And she also stressed the importance of the analyst’s role in verifying the information.

Analysts and managers, on the other hand, benefit from an efficiency gain. “The ability to aggregate a large amount of information into a single signal is extremely useful for the manager who does not necessarily have the time to follow all the sources.” The algorithm used constructs an aggregated signal on the basis of all the information, both positive and negative, that concerns a company, in real time.

We have the ability to monitor companies in real time on different dimensions, such as their financial results, their ESG performance, their communication, or the expectations of brokers and journalists about them.”Marie Brière, head of investor intelligence & academic partnerships, Amundi Institute

While investors used to focus on regularly scheduled company disclosures, the tool developed by Amundi and Causality Link provides instant information. “We have the ability to track companies in real time on different dimensions, such as their financial results, ESG performance, communication, or the expectations of brokers and journalists about them. By aggregating more than 1,700 indicators for each of the 4,000 companies we track, we can build a daily indicator reflecting the health of each company, as perceived in the press or in brokers’ analyses,” said Brière.

Identifying investors’ opinions

The signals generated by the tool are of some interest, according to Brière: “It reflects journalists’ view of company communications. This is very important, because this is often how investors assess the health of companies.” Using a series of indicators, the tool perceives whether the information reported by journalists is either positive or negative.

We have shown that markets react more to news about the future.”Marie Brière, head of investor intelligence & academic partnerships, Amundi Institute

In addition, the tool can identify the temporality of the news. “The software is able to understand whether the news reported about a company concerns the past, the present or anticipations about its future situation.” This is an important feature. “We have shown that markets react more to news about the future.”

Research to date has also found that financial information has a greater impact on investors’ decisions than ESG-related information. “This may come as a surprise,” said Brière. The reasons may be multiple. “It could be related to the fact that ESG is less followed by investors or that the impact of ESG news on companies’ profitability is more ambiguous.” In terms of impact on profitability, “if companies do not implement an ambitious climate policy, at some point they will be sanctioned by regulation or by virtuous investors, but on the other hand, these are also short-term costs.”

ESG is still a work in progress, as is the research project conducted by Amundi Institute. In the near future, other ongoing research projects will make it possible to exploit the causal links established by journalists. “This analysis will give us information on what journalists perceive as the real influencing factors, for example, on the profitability of companies,” said Brière. Even if analysts take different factors into account, those reported in the media shed light on the causal links perceived as important by investors in their asset allocation decisions.

This article was originally published in Delano. Read the article here.


Recent Posts

Follow Us