M Science, is a pioneer and leader in providing alternative data solutions to supplement core data in investment research. The company provides creative innovative approaches to buy-side clients through data partnerships, proprietary methodology and unique data sets to give intel to stay ahead of the competition and improve performance.
M Science started using alternative data 15 years ago. The first unique dataset used was counting items for sale on eBay (EBAY). The mission has been clear since: To forge unique relationships with non-traditional data providers and use proprietary technology to parse billions of data points every day to generate fresh insights and superior analysis.
They don’t issue buy-sell-hold recommendations, the mission is to provide market insight, so clients can stay ahead of the street, make smarter decisions and improve performance.
Growth has increased rapidly after a 2016 management-led buyout. In the last year, employee count has grown from 40 to about 100 employees and unique data partnerships have grown from 22 to 52. The typical client is a long/short or long only manager with a fundamental research focus. Strategic corporate clients have started to embrace research techniques for competitive purposes as well.
M Science has analyst and data science teams providing insights on 300+ companies and industries. Clients can also get a raw data feed to analyze data using in-house proprietary methods.
Michael Marrale, CEO of M Science, discussed the state of alternative data.
MODERN TRADER: With the expansion of new forms of datasets being so ubiquitous, is alternative data still an appropriate term?
Michael Marrale: The term alternative data, still holds. Alternative data refers to all data sources that are non-financial (no company filings or price and volume data). For example, the types of datasets that we use, be it transaction data, e-mail data or cellular activation data, are sources from real-world activities; consumers of the products that companies make. By the above definition, it would seem not alternative but given the context of the financial industry and their traditional usage of data, alternative here is synonymous with non-traditional, just a little catchier.
MT: Is alternative data a fad? Will it last?
Marrale: We strongly believe alternative data is not a fad. We take a strong view that alternative data will only be increasing in the next five to seven years. We have not even really started scratching the surface of ways to use data. We are only in the top of the second inning. There are many examples of new unique data for different industry groups. A great recent example is data from cloud providers like Amazon Web Services (AWS) or Google (GOOG). This was impossible data to source a year ago. Now we are using data from cloud providers to predict different key performance indicators.
MT: What are the challenges for hiring analysts and data scientists?
Marrale: This new alternative data is taking Wall Street by storm. Some current equity analysis methodologies are outdated and don’t work anymore. The millennial data-centric analysts want data and want to invent new data science techniques to predict outcomes. Wall Street employers must change to satisfy this talent pool of the future.
MT: Will earnings announcements and government reports be needed in the future?
Marrale: The way of the future is transparent real-time indexes. There are new datasets and innovative ways to predict earnings or inflation. Why do we need outdated government surveys or accounting reports on a quarterly basis to update our investment thesis? In the future, we will use these unique ways to predict earnings or inflation before announcement dates, and probably be updated hourly.
MT: If you had to invest $10 million in an alternative data trading strategy in the next six months, what would you do?
Marrale: There are two ways to approach this: The first would be to build an in-house data team of engineers and data scientists capable of processing alternative data sources at scale. The pro of this being that you can have complete control over all the processes from the ground up. The con being that it takes time, expertise and money to build this team and gain the requisite experience in dealing with alternative datasets. The second is to partner with a firm or vendor that has established themselves in this field. With the impending explosion of data pervasiveness, this helps a trading organization leverage the right skills; with the data partner providing expertise on the data and the trading firm focusing on extracting the right risk-adjusted value from it. Such a model also vastly reduces the time to market for such an alternative trading strategy.