Opinions expressed by Entrepreneur contributors are their own.
Alternative data is no longer a secret, especially in the finance industry. Novel and unique datasets were always in the sight of investment companies, hedge funds, venture capital firms, etc. A firm example of the rising trend is ESG data, which seems to (so far) provide tremendously powerful signals for investment decisions.
While there are many shapes and forms of alternative data, one most frequently used and researched is sentiment analysis. Web scraping is relatively easy to access nowadays, making the acquisition of these datasets cheap and efficient.
Sentiment analysis, for example, has previously been shown to potentially generate valuable insights. With the hype surrounding data, however, it’s easy to be led astray by the supposed potential it holds.
Sentiment and other types of alternative data
Few articles have deeply discussed the concept of sentiment. In most cases, it’s the automated processing of content (text-based, usually) to derive emotional states and judgments. Advanced tools like Google’s Natural Language AI can uncover the sentiment, assign entities (objects which are being talked about), etc.
Investor emotions and sentiment have a clear effect on the stock market. The momentum effect is a great example of mid-term market inefficiencies caused by the emotional backdrop of investors. Both underlying forces for the momentum effect, underreaction and overreaction, are essentially founded upon emotional states.
An important caveat I rarely see mentioned, however, is that emotional states aren’t long-lasting and can only provide predictions for extremely short time windows. In fact, as I intend to show further, most of alternative data, if not all of it, captures a short-term (less than 5 years) forecast.
Investors and businesses
From a bird’s eye view, stock prices are affected by investors and issuers. The former have a direct impact on the price, as they are the driving force behind demand. The latter have an indirect effect, because business performance has an impact on the perception of the investor and stock valuation.
There are also potential black swan events that might shatter the balance at play. I won’t be discussing these, however, as it would add many unpredictable variables to my hypothesis without providing much additional value.
Businesses have some indirect impact on both short-term and long-term pricing. Current cash flows, consumer sentiment, marketing strategies, etc., can have an impact on quarterly or yearly earnings. Business strategy, growth options and other factors influence long-term performance.
Xerox has always been an example of how a changing business strategy and competitive landscape can have a tremendous impact on an organization’s performance. It has both held dominant positions in the market and struggled through decades.
Since company performance and stock prices are intimately tied together, savvy investors need to take into account both short-term and long-term prospects. Alternative data, however, predominantly contains information about the former.
Related: How to Forecast Revenue and Growth
How far can alternative data go?
Various information sources can be called alternative data, since it is so loosely defined. Mostly, however, it will be web scraped information, satellite imagery and credit card data, as they are among the most used sources.
Each of these sources, whether by gathering sentiment from public social media posts or analyzing cash flows through credit card data, point to the current state of the business. In other words, it is a reflection of current performance. Additionally, these sources aren’t supposed to be long-lasting. They’re snapshots of a decidedly short period within the company’s history. Data has to be collected over some longer period for any type of forecasting to be done.
Yet, they cannot reveal anything about the abstract approach to business. No strategy can be unveiled. If a company or organization pivots and changes direction, none of that information can be gleaned from alternative data sources.
They can, however, unveil how effective these pivots or changes are. As such, alternative data can provide valuable insights into the short-term expected performance of a business, but not much else.
We have some evidence that financial services organizations have begun employing alternative data for, at least, similar means. A survey conducted by Censuswide and Oxylabs has shown that financial services organizations rank web scraping, an alternative data acquisition method, as one of the most impactful in terms of revenue generation.
Alternative data has a lot of buzz surrounding it. There’s good reason to be excited, especially if you’re involved in the finance industry. It’s a completely new source of data that can provide an edge against the competition.
I would, however, like to caution against running head-first into it. Alternative data is surprisingly good, due to its nature, at providing signals that show short-term performance. It is the same nature that prevents it from being as useful or useful at all at long-term performance predictions.
In the end, most of alternative data hides signals about short-term performance. As such, it should be used exactly for that reason alone. Trying to apply the signals to longer-term, such as five-year forecasts, might even be detrimental.
Finally, alternative data should also be weighed against opportunity cost. Getting the signals out of it requires more work than with traditional datasets. Working with alternative data means detracting resources from other potential strategies.