Open:FactSet Forum

How Firms Value Alternative Datasets


(Timothy Gavin) #1

On the back of the recent survey performed by Coleman Parkes in partnership with FactSet, the head of FactSet’s Content & Technology Solutions Rich Newman published an article with Global Banking & Finance Review.

The results provide insights into the types of data quantitative analysts value and use today, and what types they are likely to value more in the future.

(Daniel Mathon) #2

Hi @Timothy.Gavin

Did the survey contain any questions about the overall perceived value of traditional vs alternative data, current and future?

And any thoughts on what’s driving the big differences in perceived value of alternative data sets currently and in the future? Is this primarily a function of the supply side (availability/quality of data) or the demand side?

(Richard Newman) #3

Hi Daniel,

Thank you for the questions.

The survey resulted in some interesting feedback particularly in the area of current value vs perceived future value. Although alternative data has been a receiving a lot of attention, the survey and our own empirical work has shown that within the traditional institutional investment community, it is just beginning to be incorporated into investment processes. However, there is tremendous interest in testing various types of alternative data and, clearly, the value firms see is how it can help differentiate strategies.

The greatest challenges firms are facing is how to vet the vast quantities and types of alternative data as well as integrating it with their traditional core content (e.g., fundamentals, estimates, pricing) and then quickly determine if the data can lead to additional alpha or minimize risk. With so many alternative data options from both small fintech startups and large firms, the challenges are vast.

You’ll notice that Sentiment and ESG data are both seen as providing high value in the future. They both complement core content and provide actionable information. As we move forward on the alternative data journey, data that best complements core data and can be more easily integrated will continue to provide the most value.

Please feel free to ask additional questions and provide feedback. The forum provides an excellent way for partners, clients, academics and others to share information and ideas. We are excited to have launched it and with the activity.

(Daniel Mathon) #4

Hi @Richard.Newman,

Thanks for getting back to me. Some follow-ups:

You mention qualitatively that it’s still early days of adoption, but did the survey include any questions about perceived value of traditional vs alternative? (The output in the article only breaks down within traditional and alternative, with no mention of the level above that).

You mention Sentiment and ESG. Sentiment however is very low in the list for current value and only half way up the list for future value. Any reason for highlighting that specific data set? Or were you referring to News as part of Sentiment as well?

I do find it interesting that the current and future value outcomes are so different on the alternative data side and am keen to understand what’s driving that. Taking some specific examples:

Shipping moves up significantly. I can think of a few different reasons for this:

  • Data isn’t currently available (as much), but is expected to become (more) available
  • Data quality isn’t good enough currently for it to be useful, but is expected to get better
  • Data is available and of high quality but other, additional information necessary for it to be valuable (e.g. linkages to ticker level information) isn’t yet available, but is expected to become available

App Installs conversely moves down significantly. Again, there could be multiple reasons, incl:

  • Data is currently useful for an information edge and hence alpha generation, but is expected to become commoditized and arbitraged away.
  • Relevance of app installs expected to decrease
  • Goodhart’s Law kicking in, with app install numbers getting games
  • Or perhaps simply that other data sets (especially ESG and News) increase in value, thereby lowering the relative value of other data sets


(Richard Newman) #5

Great points, Daniel.

First, yes, sentiment includes news. For example, analyzing twitter feeds as well as unstructured deep history for StreetAccount.

Although the study provided some interesting feedback from the market, my response also included my observations since we launched the Open:FactSet Marketplace (OFM) and direct user feedback. The key feedback is that it’s typically not a specific data set that a user or strategy is interested in, but rather it’s the integration and overlap among different content types. For example, sentiment by itself may or may not be interesting. But sentiment combined with ESG, GeoRev, pricing, and fundamentals may be. The excitement and potential I see for alternative data are the infinite opportunities to combine the data with traditional content. Using any set in isolation and siloed does not provide as much value.

We have moved from a world of traditional quant attributes related to core data such as earnings surprise, price momentum, etc. to a world of unlimited content and attribute definition. Ultimately, what is considered alternative now will become core. But with the increase in computing power and data sources, there will always be new types of information entering the market. As I wrote previously, a big challenge is how to vet, integrate and test the disparate data before deciding which to introduce into production strategies and how best to combine data from disparate sources.