Big Data has become a household name in recent years with many Banks, Asset Managers and Hedge Funds initiating projects or purchasing software solutions. Here we ask whether this is something old or something new for Investment Management and what the opportunities are.
Some suggest Big Data is not new in the sense that many quant funds have applied unconventional data processing approaches since the 1980s. However Big Data is more than volume, it is often characterised with other differentiators to ‘normal data’ such as variety and velocity. On the velocity point it’s again clear that many quant funds have spent a great deal of time trimming microseconds from trading signals and execution processing times. However on the variety point many quant funds have historically pursued relatively similar investment strategies leading to intense competition within particular asset classes and quant skills. A hypothesis for this is that quant funds traditionally hate having to spend time manually or semi-manually sourcing, cleansing and structuring data. Manual work raises the costs of the trading strategy and is also often seen as janitorial work rather than the work of an experienced Quant or Data Scientist. This is why many quant managers trade the same liquid asset classes for which data is readily available from mainstream vendors. It’s also one of the reasons for overcrowding negatively impacting investment performance – a herd mentality that has led to game theory being increasingly incorporated into quantitative strategies.
Competing for the same trading signals isn’t however a ‘quant only’ problem and many other managers are seeking to avoid overcrowding by diversifying into more complex and less liquid asset classes such as real estate, infrastructure or ‘Real Assets’.
Big Data is mainly documents
Big Data growth is only partly fuelled however by increasing velocity of prices and other signals. In most organisations variety is the principle driver – i.e. most big data is in fact documents. This data has traditionally been the preserve of discretionary Investment Managers who have built up significant expertise in areas such as Fundamental Analysis to extract value from sources such as earnings calls, company reports, broker research, news stories and other sources. However few have successfully incorporated Big Data techniques to perform this kind of analysis at scale. A reason for this is because the skills and technologies required are not those traditionally used for Big Data problems in Investment Management – they require skillsets such text mining, linguistics and knowledge management. Again this is partly cited as a reason for overcrowding and the pursuit by many asset managers of ‘Alternate Data’ sources.
Neural Alpha – the Big & Alternate Data fuelled research platform
Searching for relevant information typically consumes up to 40% of an investment researcher’s time. Once found however there are also further steps required to extract and structure data from this content, standardise it and link it with other useful data sets to gain insights. This makes sector specific research at scale expensive and complex.
We significantly reduce the costs of acquiring and leveraging alternative and connected data sources to create highly connected risk, research and screening tools. If you are interested in finding out more or have an idea for a research project you would like to commission please do contact us.
June 01, 2016