Much like vegetables, we should all be concerned with the Kind of Data and the Quality of Data. Choices of data (and veggies) abound and we need to pick the appropriate set-combo. Quality of Data is a more complex issue that troubles mostly risk managers and regulators but should also be of great concern to the broader investment ecosystem. These issues – Picking Type and Checking Quality of Data – affect increasingly our risk-adjusted returns and their properties over time.

In this post, we will focus on the conventional investment subsector and explore what types of Data robo-advisors of all sorts use or will invest in. In future posts, we will look at the same issue – Data gathering and Quality control – in other WealthTech sectors, like marketplace lending and private markets.

The conventional investment subsector operates mainly with publicly traded securities and has been focusing mostly on individual stocks and wrappers like ETFs, mutual funds, and only lately PTFs that provide exposure to other asset classes (e.g. fixed income, real estate, digital currencies, or private equity).

Primary Sources of Basic Data

Stock exchanges have traditionally been feeding the market with basic, fundamental data for stocks and ETFs. In our coverage of the innovations out of stock exchanges, we looked at the areas of digitization and realized that focus is either in clearing and settlements or in private markets. In the area of primary data, we only note the adaptation of XBRL for financial reporting.

The Fintech Sandbox in Boston, is a non-profit that has been facilitating access to such basic quality data for startups to test their MVP reliably without incurring the usual costs. Xignite is the major disruptor in the space of primary financial data.

All robo-advisors use this type of data.

Secondary Sources of Data

Conventional Financial Data

Data related to mutual funds and indices, which at their core are nothing more than a portfolio of individual holdings, are the next kind of Data that the industry consumes insatiably. Lipper’s, Thomson Reuters, Bloomberg and Xignite, are the big providers in this space.

The rich variety of this kind of Data, starts with Fund Flows, insider activity, fund classification (e.g. value, growth, small cap etc), expense ratios, performance measures (e.g. Beta, volatility measures, return-to-vol etc), historical benchmarking, etc.

In this category, we include financial data that is used as an input for equity valuations, and projections on whether a stock is overvalued or undervalued. Data therefore, with estimates of earnings, revenues-sales, and growth, which traditionally have been provided by financial analysts on the Street. Estimize is the leading Fintech in this space, offering crowd sourced estimates for stocks.

Unstructured non-financial Data 

This is any data beyond the usual company filings. It can be website traffic and mobile access for online businesses, parking lot traffic for physical locations, human resources data (turnover), twitter trending key words, drug pipeline for pharmaceuticals etc.

1010Data, has been offering such data to hedge funds for both equities and fixed income. Thinknum is a Fintech that focuses also in this space.

Sentiment Data

Using sentiment data (i.e. optimism, fear, trust, capitulation, etc), to drive investment decisions, is an alternative way that will eventually be combined with the other kinds of data.

In our two part coverage of the Sentiment space, we identified a couple of early partnership that are combining primary and secondary data types (e.g. Thomson Reuters with Amareos, E-trade with TipRanks, Ameritrade with Likefolio). Since this is a nascent space, we have opened a conversation on the Fintech Genome that tracks the developments. Edit the wiki or simply post in this live conversation that monitors the adaptation of sentiment data and analysis into the mainstream.

There aren’t any robo-advisors using yet this type of data. Currently robo-advisors don’t allow for a scenario “Go cash, don’t invest now” or combining momentum with fundamental and therefore, sentiment data doesn’t add value.

Model generated Data – Alpha-generated, AI & ML

Using model-generated data to rebalance a robo-advisory portfolio is the next trend that we anticipate. We foresee, that the robo-advisory space will be moving from a primitive MPT portfolio-rebalancing method, to one that is dynamic and uses data generated by models to enhance the rebalancing and also allows for “I am out of the market for now”.

 In summary, robo-advisors are mainly using the primary resources of data. Any robo-advisors using any of the secondary sources of data?

Alpima is producing and using model generated data. They are actually offering a rich variety of model generated fundamental data. Similarly, Elm Partners is using Value-momentum models; combining fundamental analysis with momentum-technical signals. Logical Invest, is a Fintech focused on signal providing for financial advisors and therefore, producing model generated data.

Quandl is producing unstructured non-financial data, sentiment data, and alpha generating data. Their offering is rich and is coined, Alternative Data; and for now is refered to as “unorthodox data”. Is Quandl, the alternative for Yahoo Finance?

Daily Fintech Advisers provides strategic consulting to organizations with business and investment interests in Fintech & operates the Fintech Genome P2P Knowledge Network.  Efi Pylarinou is a Digital Wealth Management thought leader.