{"id":19934,"date":"2024-04-26T10:47:40","date_gmt":"2024-04-26T09:47:40","guid":{"rendered":"https:\/\/affidaty.io\/blog\/?p=19934"},"modified":"2024-04-26T11:59:07","modified_gmt":"2024-04-26T10:59:07","slug":"finance-artificial-intelligence","status":"publish","type":"post","link":"https:\/\/affidaty.io\/blog\/en\/2024\/04\/finance-artificial-intelligence\/","title":{"rendered":"Finance and Artificial Intelligence: the Fox and the Grape"},"content":{"rendered":"\n
\n

The article reflects solely the opinions of the author and does not intend, therefore, to represent official positions of the Institute to which he belongs.<\/mark><\/p>\n<\/blockquote>\n\n\n\n

<\/div>\n\n\n\n

Introduction<\/h2>\n\n\n\n

The use of advanced algorithms for predictive analysis in investment processes is certainly not a novelty in the field of finance. However, the current development and spread of Artificial Intelligence (AI)<\/strong> systems, facilitated by increasingly powerful computational capabilities and the availability of previously unthinkable storage<\/em> devices, are revolutionizing not only the organizational and operational approaches of banks and other financial intermediaries but also those of the supervisory authorities themselves.<\/p>\n\n\n\n

In recent years, operators in the financial sector have progressively redefined the production and delivery processes of services through the use of technologies such as cloud<\/em> computing, Machine Learning (ML)<\/strong>, predictive analysis tools, blockchain<\/em>, and Natural Language Processing (NLP)<\/em>. The risks associated with their use are the subject of increasing attention from supervisory authorities, which have initiated specific assessments to verify the degree of technological dependence on outsourcers<\/em>, the awareness of corporate bodies, the level of regulatory compliance with traditionally technology-neutral supervision provisions, and inspections at technology providers themselves (increasingly concentrated).<\/p>\n\n\n\n

The discussion about the possible applications, emerging risks, and prospects associated with AI systems in the financial sector is now more heated than ever. The recent approval of the AI Act by the European Parliament strongly raises the eternal dilemma about the ability of regulation to “embrace” technological evolution in support of efficiency and economic progress without reducing its innovative scope (at best) or unduly hindering its path (at worst).<\/p>\n\n\n\n

The relevance of the phenomenon has also been confirmed by the International Monetary Fund, which dedicated the December 2023 issue of the F&D – Finance & Development magazine to Artificial Intelligence, addressing the theme “What AI means for economics<\/a>” from various perspectives.<\/p>\n\n\n\n

In my professional experience, I came across algorithmic trading and the use of machine learning systems in finance during my studies in Financial Engineering and Asset Management at the Swiss Finance Institute in the early 2000s. In particular, with the late Prof. Salih Neftci, we discussed what then appeared statistically as an anomaly in the mutual fund sector: the Medallion Fund managed by Renaissance Technologies.<\/p>\n\n\n\n

The fund was established in the late 1980s by James Simons, one of the most brilliant American mathematicians of the last century, who graduated at the age of twenty in mathematics and literature from the Massachusetts Institute of Technology, collaborated with the U.S. Defense Department in cryptography and decoding of military messages, and became a department chair in mathematics at thirty. After receiving important scientific recognition and having an outstanding academic career, Simons decided to dedicate himself to the world of finance by founding Renaissance Technologies with other mathematicians. After initial attempts, in 1988 he established the Medallion Fund, which since then has achieved profits of over $105 billion, making Simons one of the richest men in America (with an estimated current wealth of over $31 billion) and generating an average annual return of 40% for its subscribers (net of high fees). No other financial instrument has been able to generate such high profits for such a long time.<\/p>\n\n\n\n

The specific algorithmic trading<\/mark> techniques used by the fund remain a mystery to this day, but it is still possible to reflect on the methodological approach. First, Simons decided to surround himself not with finance experts, but with mathematicians, engineers, statisticians, physicists, and data scientists, who still constitute over 90% of the personnel of the company he founded. Among the first to be involved were also IBM engineers working on artificial intelligence projects.<\/p>\n\n\n\n

Secondly, we know that the strategies used involve the generation of millions of buy and sell orders per year, executed within fractions of a second. During our Swiss studies, we discussed how Renaissance Technologies’ algorithmic approach partly replicated that of speech recognition (a technology that IBM itself developed starting in the 1960s).<\/p>\n\n\n\n

Just as in voice dictation, the computer must concatenate almost in real-time the uttered phonemes, anticipating the word about to be completed and inferring, based on the most probable statistical correlations, the next word, in algorithmic trading, computational capacity aims to interpret all market signals (the “financial phonemes”) to estimate, as quickly as possible, the probability that the individual security may experience an appreciable movement (upward or downward) in the next few seconds (or fractions of a second) upon which to generate the buy or sell order. In other words, predictive modeling algorithms are used to identify patterns<\/em> in financial markets through machine learning<\/em>.<\/p>\n\n\n\n

With the increasing complexity of financial markets, coupled with the emergence of new forms of investment with derivative and structured components, Simons’ company’s hardware has also strengthened: currently, their computational systems rely on 50,000 computer cores<\/em> using a proprietary research database that enriches itself by 40 terabytes per day, leveraging connectivity of over 150 gigabits per second (source: https:\/\/www.rentec.com<\/a>).<\/p>\n\n\n\n

<\/div>\n\n\n\n

AI Systems in Finance: Main Uses<\/h2>\n\n\n\n

Considering the various uses of AI in finance, it is evident that competitive pressures in a highly regulated sector such as banking and finance have favored a rapid adoption of artificial intelligence\/ML, allowing for efficiency gains, increases in individual productivity, and cost savings, reshaping customer interfaces, improving forecast accuracy, risk management, and compliance<\/em> adjustments.<\/p>\n\n\n\n

The competitive wedge imposes, in fact, continuous investments in technology on banks and financial intermediaries not only to maintain and consolidate their market shares but also to support the feverish pursuit of high performance<\/em> capable of attracting new investors and strengthening capital levels, in accordance with the capital constraints imposed by supervisory authorities.<\/p>\n\n\n\n

Among the main areas of application, the following can be mentioned:<\/p>\n\n\n\n