Xiaowei Sun
The use of alternative data to aid investment strategies is being largely considered in quant finance. However, the actual application of the vast amount of data in investment strategies is not as wide-spread yet. There are kinks to be worked out, but the potential is there. In this article, Xiaowei Sun, Risk Methodology Analyst at Deutsche Bank, explores these possibilities.
Big data is quite a hot topic at present and making marvellous influences on various industries. Appropriately applied, it is absolutely an effective driver for quantitative finance as well. Quant finance is such a grand study field. The aim of this essay is to briefly present my perspective how big data will play a part in strengthening investment strategies.
We are currently living in a world with data explosion. Billions of Google search requests and social media posting are just circulating around. In the uninhabited suburbs, there might be hidden information storage centres that are operating day and night to process data in all shapes. In the CBD, we observed that more and more FinTech start-up companies were established and they are making financial innovations by means of data science.
Big data is an extremely large collection of datasets that may be analysed to reveal patterns, tendencies, and associations, especially relating to human behaviours and interactions. It is exactly the sort of information that investors care about. Big data will bring unprecedented opportunities to the financial area through expanding the data variety, structured or unstructured, to feed various financial models.
Looking back to the traditional mode of stock investment, a human equity research analyst is only capable of taking care of limited number of companies in a particular industry and performing the fundamental analysis on accounting numbers in financial statements. A human trader might be overwhelmed by emotion or mislead by a piece of media news.
The high-frequency trading can cover much more abundant factors than human and avoid emotion interference to grasp possibilities in the market. Nevertheless, it is gradually entering a bottleneck stage. As market participants increase, outperforming others is getting more demanding. Many algorithms and models in use are basically open or semi-open. HFT is racing against time. Eventually it is the hardware and even the broadband speed that may determine your profit.
Big data is bringing about an additional and broad input source as well as the powerful processing techniques, to provide a much wider variety of data in all shapes and sizes to feed existing models or train newly data-driven models. In particular, we clearly see the potential and values contained in the unstructured data such as social media, search engine requests and geographic information. They do imply relevant information but play trivial roles in investment strategies for the moment.
For instance, the text in the media can be digitalised, the key terms are captured based on the dictionary defined beforehand and the positive/negative wordings are classified to score a business. The recently published news reflects market sentiment in a quantitative way and helps make forecasting on stock price movements.
Another example is the number of search requests submitted in engines, for instance googling an enterprise’s name is positively related to the trading volume of this company’s stock. The browsing footprints help predict future market activities.
Matt Napoli, 1010data, on alternative data's past, present, and future
To summarise, the next big thing to impact quant finance, in my viewpoint, is big data, especially the alternative data it brings as extra information source and powerful data science techniques – hardware, programming system and algorithm – to solve problems.