A by-product of the internet, Big Data technology is now being applied within numerous economic sectors including the asset management industry. The uses to which Big Data can be put in this specific sector are potentially huge as they encompass activities ranging from "client-profiling" and operational risk management right through to the investment process itself.
Big Data comprises the collection and handling of a huge quantity of diverse, unstructured data. The technical response to this challenge is two-phased: a Hardware part (IT architecture based extensively on parallelism) and a Software part, with a dedicated operating system (Hadoop). A whole service industry based on these technologies has, consequently, considerably grown in recent years (Cloud Computing, Big Data data analysis tools, …).
The analysis approach and the tools used to extract value from Big Data are actually very similar to those traditionally used in finance. After a series of transformations (filters, normalisation, management of anomalous/missing data, …), the data are analysed by "forecasting/regression" or "classification/machine learning" quant-type models, the best-known of which include the artificial intelligence methods (neural networks) commonly used to identify patterns and even the so-called "clustering" methods used, for example, in client-profiling.
Asset Management, just like other industries, is also interested in Big Data. Indeed, investment companies seek to better anticipate the needs and reactions of their clients in order to offer them better-adapted products and actions. They also see in this technology a tool for improving their operational risk. As for developing a systematic investment process based exclusively on Big Data, we are, for the moment, only at the research stage, with very few fund launches implicated: the industry is as yet unable to objectively judge the performance of such strategies. Nonetheless, it does have a promising and relevant contribution to make to the generation of new sentiment indicators and near real-time important macro-economic indicators (inflation, unemployment, GDP, etc.).
Candriam has already longstanding experience in most of the data analysis/Machine Learning techniques traditionally used in Big Data Analysis, specifically applied in many of our management processes (CTA, Quant Global Macro, Index Arbitrage…). Additionally, the company has developed, over the past decade or so, proprietary expertise in the storing and handling of large quantities of data. Finally, there have been regular discussions at Candriam – to which all departments have contributed – designed to substantially enhance data use (via a Cloud solution). To this end, four main priorities have been identified: Client & Distribution, Innovation, Digital Enterprise and Data Value.