When I buy a steak, the motivation is relatively simple: I’d like to eat it. The same goes for a burger. Or indeed, some cheesecake. I suppose I could derive some value from snapping a photo of it and putting it on Instagram, without eating it, but that frankly seems like an utter waste. Ultimately, you buy food to eat. You go on a holiday, to go to somewhere sunny and fun, not in order to be in a metal tube in the sky, which flies in the sky (although that is inevitably part of the experience).
When it comes to financial firms using alternative data, it can be tempting to rush into the whole thing. This can involve hastily hiring massive teams of data scientists to do “something” with these funky datasets. Investors in a fund will be pleased that “something” is being done (at first..), and it might well be reflected in increases assets under management and so on. However, if firms are not careful, this can result in our Instagram snap without eating scenario, which I mentioned earlier. Ultimately, for data scientists to add value to the buy side (or indeed sell side firm), their insights need to be utilised somehow. Indeed, this is a subject discussed extensively in The Book of Alternative Data by Alexander Denev and I, which recently got published on Wiley.
Alt data needs buy in (as does any new initiative). This isn’t simply just buy in from management to provide the budget for data scientists, datasets and so on, which is a prerequisite. It also needs buy in from the rest of the firm. Data scientists need to understand the types of questions that are useful for the business. This involves communication with the rest of the business, in particular with analysts and portfolio managers on the buy side. The communications needs to be two sided, so that data scientists can talk about their findings and what’s important and actionable, without getting bogged down in the technical details. Their observations are monetisable only if they contribute to the trade generation process (in hopefully a profitable way!)
It is all a team effort when it comes to dealing with data. Whilst at first it might be the case this team might be very small, as it becomes more utilised more people might be hired. From data engineers to help store and maintain datasets, tech teams to help maintain tools, to data strategists who scour for new datasets and data scientists analysing data. Quants may build systematic trading signals and models with the data. Discretionary portfolio managers can use observations gleaned by data scientists and quants into their own trade views.
In particular during these turbulent times, the need for alt datasets has increased, as portfolio managers need to understand what’s happening to the economy on a high frequency basis, and not just waiting for official economic releases once a quarter or once a month. The key though is to make sure that the process is set up properly, so it provides value for the business and is an integral part of the decision making process (as opposed to an expensive exercise in hiring, without a monetisable outcome). Otherwise it can end up like an expensive exercise without a long term benefit to the business.