Tips for investors: buying alternative data

Posted by: Charlotte Werger on

Alternative data and fintech are booming. An increasing amount of alternative data is hitting the financial industry. As such, it is starting to become increasingly difficult to decide what’s the right type of data for you and your organisation. So let’s discuss which things to consider when exploring alternative data for your investment strategy or research.

Alternative data providers usually take some form of crowd sourced or unstructured data (think images and text) and bring structure to it, such that it can be easily consumed by others. By doing so, they create tremendous value, as previously hidden insights are now uncovered. Working with large amounts of unstructured data usually requires highly technical skills and experience. Therefore, acquirers of alternative data get access to those skills, without having to build everything in-house. Investors are always on the lookout for alpha. Alternative data can be a way to provide alpha and trade on new factors. But adding alternative data to your portfolio isn’t always straightforward, and usually you need to do your research to trade it successfully.

So here are a few crucial things you need to consider when it comes to adding alternative data:

  • Uniqueness of the data
  • Coverage
  • Frequency of updating
  • History of data
  • How to combine
  • Service level of the vendor
  1. Uniqueness of the data
    When we’re talking about “uniqueness” you need to ask yourself the following: What are the unique insights this data is uncovering, and how is this different from what’s already out there?
    But you got to think broadly about this. Test how the data is correlated to other things you already have access to, but also test whether it leads or lags those other insights. It’s likely that the data is measuring something earlier than standard indicators which are slow to update. It can also be that the data is measuring some sort of sentiment that can function as a contradicting view to consensus, etc. Another thing you need to ask yourself is: “who else has access to this data?” Is the data widely distributed or available to a select few. If it turns out to be hugely profitable, is there a way to gain exclusive access?
  2. Coverage of companies
    Do you have hundreds of companies in your investment portfolio, or just a handful? If it’s the former then you want to think about how well the data covers your investable universe. And more specifically, does it cover large cap or most micro cap. In that sense, make sure you know how robust your investment strategy is to growing AUM. Does your Sharpe ratio deteriorate quickly when you try trading with a larger portfolio? 
For very concentrated portfolio, alternative data will actually be hugely valuable as well. It can allow you to go much further in depth and discover things that are hard to measure by standard data alone. When using alternative data for concentrated portfolios, you need to think about how many good trades can you make with this data in a year.
    Coverage can also be interesting from a regional perspective. Alternative data can be an excellent way to reach those regions that have unreliable official data. An example is using satellite images to measure economic development in regions that have no reliable measurement of economic activity.
  3. Frequency of updating
    You might think: more frequent is always better. But this is not necessarily always the case. A data vendor might be incentivised to sell you a high frequently updating data stream, as this usually will cost more. However, the underlying patterns/relations you’re measuring might not actually change that fast over time, so lower frequency is enough. On the other hand, you don’t want to miss out structural breaks in the data, or events where patterns suddenly change. So think about these things, and consider how they would fit in the time horizon of your investment strategy.
  4. History of data
    A drawback of most alternative data vendors is that they can’t provide a long history of their data. This has usually to do with the fact that the websites/hardware/raw data they use simply didn’t exist a few years ago. In this day and age, we have got to move away from the paradigm that everything you put into a portfolio needs to have a track record of at least 5 years. With new techniques and new data, that approach simply does not work, otherwise you will be missing out on the new stuff. So what to do then? If you can build a strong investment thesis as to why a certain approach works, or why your alternative data can predict XYZ, then you can start “incubating” these things in your strategy, by assigning a small asset allocation to it. You don’t need large amounts of history, what you need is some test data and a solid investment thesis to back it up.
  5. How to combine
    Some alternative data is best when combined with other data and other alternative data functions very well standalone in a strategy. You definitely need to consider how you can combine existing insights with your new alternative data, and how many different investment strategies you can build with your alternative data stream. There are usually many different ways to trade the insights, and sometimes it requires getting creative. Think about the case again where alternative data can cover regions that are hard to reach with standard data. Those insights can for example be used to augment your existing strategy and make it one with “global” coverage. Another example is combining different sources of alternative data. In this paper, they combine crowd-sourced earnings estimates with another alternative data provider on Twitter sentiment to create value. They confidently claim: “We estimate that gross of costs, the “alpha” from tweet sentiments post-earnings announcement may be as high as ~10-20% per year”.
  6. Service level of the vendor
    When buying data from a vendor, you have to consider whether that vendor can reliably deliver that data on a real-time basis. You can ask them questions to check how often their API/FTP server goes down, and how quickly they can react to that. What do they do when you find out in the morning that your data is missing and you can’t trade. Are they in the same time zone?
    As you might be dealing with many different data vendors, you don’t want this to become a headache. From personal experience, I know large established data vendors can have these issues too. So working with smaller companies that are lean and nimble might actually be a positive thing, rather than a drawback.

Now that you have a sense of what to think about when buying alternative data, go and have some fun with it. It will give you an excellent opportunity to connect with your clients and explain how you’re using new high-tech insights in your investment strategy. And if you have questions about how to use satellite images in your portfolio, don’t hesitate to reach out to