down
News

How can artificial intelligence assist investors in decision making?

Artificial intelligence (AI) is defined as the simulation of human intelligence processes by computing systems. This involves learning (the acquisition of information and rules of use), reasoning (the use of rules to arrive at conclusions) and auto-correction (improvements thanks to the retroactive analysis of conclusions). Since its beginnings in the 1950 article “Computing Machinery and Intelligence” by Alan Turing, AI has experienced highs and lows linked to technological process. In recent years, it has prospered thanks to a favorable environment combining a continual reduction in the price of computing power and a sharp rise in the volume of data from social networks and mobile devices. The increasing adoption of virtual assistants (Siri and Alexa) demonstrates recent progress in the AI sector, clearly one of the hottest trends in 2017, as explained in this article.

 

AI is being applied to chatbots, regulatory compliance, fraud detection, client behavioral analysis, and more. In market trend forecasting, for example, AI is being used to improve human decision-making, offering particularly interesting applications for buy-side stakeholders. Here, AI technologies are justified by the large volume of heterogeneous structured/non-structured data involved, near real-time analyses and recognition of complex configurations or patterns.

 

Given the historically low interest rates, directionless stock markets and sustained pressure on costs due to the rise of passive management, an urgent need has emerged for alpha-generating ideas (active management). A number of leading hedge funds are already exploring ways to exploit AI to boost fund performance. Automated or machine learning is being driven by massive volumes of financial data (prices, volumes, research notes, national macroeconomic data, corporate financial/accounting data, etc.), which it uses to refine its analyses, reveal patterns and infer investment strategies.

 

Companies such as Aidyia, Sentient Technologies, Rebellion Research and Two Sigma are developing approaches based on a combination of IA techniques, such as deep learning and evolutionary computation, to identify and implement investment strategies in complete autonomy. Another example of AI’s impact on investment processes is Kensho, a start-up specialized in financial data analysis. This company, in which Goldman Sachs acquired a multi-million dollar stake in 2014, aims to reduce human decision-making. Like Apple’s Siri, Kensho’s software is capable of replying, in a simple text box, to several million questions asked in natural language. This technology may well threaten the jobs of highly remunerated experts (quants, analysts, and so on) in the not too distant future.

 

While AI can help industrialize market trend forecasting based on fundamental analysis, it is particularly suited to sentiment analysis (SA), the evaluation of society/market opinions expressed in the media (news releases, forums, social media, chats, blogs, etc.). With their statistical approach to natural language processing (NLP), PsychSignal and MarketPsych enable buy-side entities to quantify familiar emotions (fear, happiness, uncertainty, confidence and urgency) to predict risk taking and integrate this information into their investment processes.

 

As AI technologies become more widespread, individual investors are also starting to witness the benefits. For example, online wealth management platforms such as Responsive and Quantenstein offer AI-based services that adjust portfolio content in near real-time.

 

Advances in AI are starting to provide investors with tools to make the most of the overwhelming flow of data facing them day to day. In particular, AI assists human decision making, helping investors make choices through an enhanced market trend forecasting. However, a number of obstacles continue to undermine the adoption of these technologies and certain key challenges must still be overcome, such as the shortage of AI skills, the existence of data silos and the need to identify relevant applications.