Artificial intelligence (AI) is often considered a threat to skilled jobs, especially in the financial sector. However, AI is above all a tool to help make sound investment decisions.
It’s been a long time that we saw traders clinging to their phones to get information and place orders. The computerization of capital markets started more than 30 years ago and has progressed rapidly in recent years. Thanks to artificial intelligence (AI), and particularly Machine Learning, the asset management industry is now experiencing a new transformation with regard to data processing and analysis.
The markets have entered the age of data
When making investment decisions, today’s asset managers have access to more data than ever before. Balance sheets, profit and loss statements, performance ratios, credit ratings, news articles: there are thousands of data points providing a remarkable amount of information which – in theory – should help to build better-performing investment portfolios.
But data has become so numerous that it is easy to drown in it. Or should we say that a human being can easily be overwhelmed, because this seems to be the case: despite the amount of data at their disposal, asset managers are still struggling to beat their benchmarks.
And with good reason: humans not only struggle to process all the available data, they also have "cognitive biases", i.e. a tendency to perceive information through a filter of personal experience and preferences. Whilst this is a useful mechanism to help the brain cope with the input it receives any second, it also makes decisions less objective. Therefore, good-quality data alone is not enough – you need to know how to properly process it in order to extract its value.
Eliminating human bias, a main benefit of AI
Cognitive biases are by no means specific to the financial industry, we all have them.
The best-known is probably the confirmation bias, which subconsciously makes us reuse the same strategies because they have worked in the past. But in the financial world, this approach does not always pay off. Similar to sports, the systematic use of the same strategy over and over again usually stops working at some point.
There are other classic human biases which are prevalent in financial management, such as overconfidence, risk aversion or selective memory, i.e. only very good or very bad experiences are remembered. All these biases have been analyzed thoroughly by behavioral finance studies.
The solution: combining human intelligence with AI
Artificial intelligence has become a useful tool to help avoid human bias. Created to make rational decisions based on objective criteria rather than emotions, AI can help investment managers to eliminate many of their personal biases.
This doesn’t mean that humans are stripped of their responsibilities or decision-making powers. Quite the contrary: AI can be an invaluable support for investment managers, allowing them to make more informed decisions. Moreover, human oversight is essential to compensate for the possible biases of AI.
In fact, machine learning means that algorithms learn by themselves from data samples selected by humans. Therefore, selection biases can exist and an AI, however sophisticated it may be, will only reproduce these biases. And there are other reasons for AI bias, such as a low representativeness of data samples or incorrect framing of a problem which may lead to unintended outcomes. This is why it is crucial that a human manager is in charge who is an expert in his field so as to identify any potential issues early on and maximize the benefits of using an AI system.
Some major financial institutions are already exploring the co-existence of human intelligence and AI. However, evangelization is still in its infancy in the asset management industry as a whole. But faced with stiff competition from ETFs, active asset managers are undoubtedly looking for effective solutions to regain their ability to generate alpha. For them, AI can offer real benefits to help them outperform the market.