Sandalwood Research Blog
Five Things to Remember When Using Alternative Data
Alternative data is a hot topic in investment management, with over 90% of hedge funds we speak to saying they plan on incorporating unique data into their research process. However traditional CIOs and portfolio managers often don’t know where to begin.
We often hear new users of data say things like, “I’m not going to pay for data unless I can trade on it”. There is the expectation that the data purchased is the end-all-be-all of whatever the data is supposed to provide a perspective on. Or they say, “The data isn’t perfect, I’ll pass”. These are understandable perspectives. Clients want to spend limited research budgets in the most effective way possible. But this type of thinking is unrealistic, and will often lead analysts to ignore data sources that may otherwise be highly useful. It is wishful thinking that one can purchase a single data source and expect to base a large position upon it. Should a perfect data source exist, it would easily be digested by machines, supplanting the role of the investment analyst. Alpha generated from the dataset would eventually be gone. On the other hand, a savvy investment analyst can still build a meaningful edge using more complex and nuanced data sources.
Here are our top 5 things to keep in mind when utilizing alternative data:
- Data is not a magic wand. Would you place a trade on a single expert network call? Would you base your entire thesis on a single source? The answer is a resounding “NO”. Yet over 90% of the hedge funds evaluate a data source on their ability to trade on that single alternative data source. If the data source isn’t able to provide a tradable insight in isolation, then it’s not worth anything right? This unrealistic expectation makes analysts dismiss data that otherwise would be highly useful and actionable when combined with other information sources. Purchasing a dataset doesn’t mean you are suddenly armed with all you need to make an investment. Can data be powerful? Yes. Can data help generate alpha? Absolutely. But keep in mind the following:
- Do not expect precision from one dataset. Analysts often expect to make precise revenue estimates using a single alternative data source. Is precision possible? Maybe, if you have access to multiple data sources and understand the nuances of each data set. You can have precision if you have 100% of the puzzle. However this is often not the case, and nor should it be expected.
- Do not expect perfection, ever. A corollary to the above is the unrealistic expectation that the data be perfect. There is no dataset that is perfect in its predictive value, where it is 100% accurate, 100% all of the time. You should not expect this of any dataset you purchase. Equity research recommendations are hardly 100% accurate – yet they still provide value.
- DO expect inflection points. The easiest way to generate alpha from alternative data is use it to track directional changes and inflection points. Alternative data is very good at providing inflection points ahead of the street. Whether it be sentiment data, transaction data, or shipping data – give it long enough and you will see trends shift the other direction. If you can identify an inflection point earlier than everyone else, then precision isn’t necessary. Position yourself first, and watch other investors pile in after you. A dataset that can accurately call directional changes in a company’s revenues is not precise, but it is still useful.
- Data should tell you something useful and unique. Instead of evaluating a data source by whether you can place trades using that data, seek to evaluate whether the data tells you something new. If you are using the data to help model a company’s sales, expect the data to tell you something new about the direction, degree, or nature of the company’s sales. When evaluating a new data vendor, ask yourself the following questions:
- Does the data tell me something I don’t already know? Data should tell you something new or help verify a theory you may have about a company. If it brings you closer to conviction or challenges your thesis – then it’s worth evaluating further.
- Is the information I obtain using this dataset unique? Data is most valuable if similar insights could not have been gleaned from another source. If you could have gotten the exact same insight from a channel check or expert network call, then maybe the data isn’t worth buying. But if the data tells you something about a company’s sales that an expert would never be able to tell you – then perhaps you have a dataset worth purchasing.
- Each data source is just a piece of the puzzle. If you’re looking for the data to give you a view of a company’s sales, don’t expect it to give you 100% of the company’s sales. Not only would such data be extremely expensive, it’d likely have compliance issues. Data should provide you with a useful piece of the puzzle, not the entire picture. It is up to you to both collect as many pieces as possible and to make better use of the pieces you have generate alpha.
- This point cannot be underscored enough – because data can provide a piece of the puzzle, it doesn’t mean you can extrapolate that piece to inform the entire picture. The data may be accurate, but only for its portion of the puzzle. Investors will rarely be given the entire picture outside of a company’s official disclosures. But each increment puzzle piece is valuable, and even more valuable when combined with other pieces.
- Data is synergistic – therefore use multiple (data) sources. If a data source just one piece to the puzzle, then you want as many puzzle pieces as possible. Information can be synergistic and data is one of the rare situations where 1+1=3. Combining two or more data sources often results in a unique view that could not be achieved from a single source. Purchase data that tells you something you don’t already know, combine it with other information, and build enough of that puzzle until you have enough of the picture to have conviction. Does the data source you are evaluating make your existing information more actionable or useful? Analysts are already used to piecing together information about a company using various information sources – alternative data is just an additional source of information.
- Unique data requires unique investment. The Pareto Principle states that 80% of the output comes from 20% of the invested input. In building your information puzzle, it’s likely that 80% of the required information for an investment comes from easily obtainable sources such as company filings, equity research reports, and earnings transcripts. The final 20% of the puzzle will be the most time consuming and expensive to obtain. But it is also the most valuable, since few others will have it. The final 20% will require 80% of your effort, and likely your budget. But the fruit of the final 20% is the opportunity to generate alpha. Analysts often expect data to be both cheap and easy to use. If that were true, then then everyone would already have access to the data and the insights, and any edge obtained from the data would be gone. Therefore if the data is truly unique and useful, then you should expect it to be expensive and / or require effort to use. The barrier to entry will either be cost, effort, or both. Keep the following in mind:
- Unique data will be expensive. If you can get the same data from 4 different vendors, then don’t expect to pay a premium for it. But if there is only one source for the data – expect to pay up.
- Invest time in analyzing the data. Analysts are extremely busy and it is very rare to see an analyst spend significant time with the data or ask unique questions of the dataset. Analysts often dismiss the data as useless or draw the wrong conclusions from the data after only a cursory examination. Simply spending more time on the data will offer you an edge over 90% of the hedge funds out there in finding unique insights. If you aren’t willing to put in the work, don’t expect results – data is not a shortcut to instant alpha.
Sandalwood Advisors is Asia’s first alternative data platform. We offer investors actionable data sources focused on Chinese consumer spending. You can also follow us on Twitter @sandalwooddata or on LinkedIn.
Sportswear in China – Who’s winning the race between Anta, Adidas, Nike, Li Ning, Xtep? Sportswear brands in China are benefitting […]
What has the impact of THAAD been on duty-free operations of Lotte & Hotel Shilla? Due to the ongoing THAAD controversy, […]
China’s resurgent appetite for branded luxury goods has driven luxury stock prices through the roof. Kering, Moncler, Burberry, LVMH are […]