Sur paperjam.lu, nous utilisons des cookies pour mémoriser vos préférences, gérer la publicité, vous proposer des contenus toujours plus pertinents selon vos centres d’intérêt et améliorer sans cesse votre expérience utilisateur. En poursuivant votre navigation, vous acceptez l’utilisation de ces cookies.Fermer
Machine learning, big data and the study of human biases are leading to a deeper understanding of the emotional forces that influence investor behaviour. Combining these elements into the investment process creates a robust framework for guarding against common investor pitfalls.
During a visit to San Francisco I had the opportunity to meet with Dr Richard Peterson, the founder of MarketPsych and one of the leaders in the field of behavioural finance. The use of sentiment analysis was already incorporated in our investment process at that time so this was an opportunity to see how the field of big data and artificial intelligence was progressing. The innovative use of algorithms that can read news articles alongside impressive computing power enables market sentiment to be analysed, and quantified, at a much deeper level than had previously been available. Indices are constructed from thousands of web sources, many within milliseconds of publication, across both professional and social media domains.
The self-learning algorithms developed by Dr Peterson and his team can classify this vast amount of data into sentiment indices such as optimism, fear, joy or conflict. This provides real time insight into the sentiment, both negative and positive, that plays a role in driving markets. The events of Black Thursday in 1929, Black Monday in 1987 and the market crash of 2008 show the dramatic effect these sentiment shifts can have.
The philosophy of the Multi-Asset team at NN Investment Partners is a simple one, to combine the best of man and machine. We use robust statistical methods to assess whether the data inputs we are receiving has informational content that can be used in the construction of the toolkit that supports our investment decisions. The structure of this toolkit allows for a great variety of input signals, both fundamentally and behavioural based, and we believe that the behavioural elements can be just as accurately assessed for their predictive value as any other.
This toolkit makes use of digital news and social media feeds, converted into indices via text analytic techniques, to capture real time sentiment within different segments of the asset markets. These behavioural elements sit alongside the team’s fundamental analysis of markets and provide a unique perspective on how “the market” is feeling. The desire to measure the sentiment, or mood, of the market is not new. However, the emergence of social media, self-teaching algorithms and the ability to process large amounts of data, practically in real time, create an entirely new way to achieve this. We leverage on these techniques in our investment process.
The construction of this proprietary toolkit provides a framework for assessing markets, essentially the machine inputs that are then used by our strategists and portfolio managers to meld, with their own research and insights, into a coherent view on financial markets.
The other aspect of the decision-making process where we can benefit from machine discipline is to improve the ability of strategists and portfolio managers as forecasters.
What this means is that we can build a framework whereby the use of big data and self-learning algorithms can be used to build predictive models that evolve with the changing social and financial landscape.
This is not the end of the story, of course, big data and narrow artificial intelligence are currently the topic du jour, however, it is quite possible that our next round of innovation will be driven by something completely different. Learning, improving and adapting, in a never-ending cycle of technology and innovation, will continue to drive our investment process forward.