• Rasa Sosnovskytė, Head of SEO and Growth Marketing at Oxylabs.io


• 29.06.2022 11:45 am
#data

The momentum effect is the tendency for stocks to show persistence in performance. In simple terms, a stock that’s performing well is likely to continue to do so, even if it’s overvalued. The same goes for underperforming stocks.

While the effect itself is relatively established, the source of these tendencies remains subject to much debate. The momentum effect is so unusual, in fact, that it’s considered to be a market anomaly.

Momentum flies in the face of our expectation that the markets are completely rational. Investors, however, aren’t often too interested in the philosophical underpinnings of how markets act, whether they are rational or irrational. What we will be looking at here is how, regardless of rationality, alternative data can help us generate alpha through anomalies.

Underreaction and overreaction

Explanations of the momentum effect are often relegated to the realms of behavioural economics rather than finance theory. An explanation that is often given is that investors are not entirely rational. Cognitive biases cause them to overreact or underreact to certain news or the performance of stocks.

Since these same investors play no minor role in the valuation of stocks, even extremely small biases can start compounding into a snowball-like phenomenon until their effects become reflected in prices.

Numerous cognitive biases might coalesce to have the aforementioned impact. In cases of underreaction, sunk cost fallacy, anchoring bias, and many others might be involved. Overreaction might be due to some measure of hype, fear of missing out, or something of the like.

Research seems to indicate that underreaction is the safer option as overreaction causes a feedback loop that keeps pushing pricing away from fundamental value. Price corrections are bound to happen, and they might be more violent than normal.

Underreaction, however, isn’t subject to the same pressure. Instead of moving away from fundamental value, it will trend towards reduced mispricing, eventually evening out. As such, there’s little to no risk when price correction happens.

Numerous bubbles that have happened over the years are a perfect example of investor overreaction. Prices seem to rise without reason until the market buckles under the pressure and a sudden correction happens.

Finding momentum

Since the momentum effect is a market anomaly, regular finance theory strategies might be less useful than in most other cases. As we’ve established, the likely source is human biases or heuristics that may cause mispricing.

Either way, if you’re involved in investing, you’re likely less interested in the theoretical underpinnings of the momentum effect and more likely interested in how one could estimate such happenings. There have been attempts to derive benefits through generic momentum strategies, but these are subject to a price reversal.

Robeco, an asset management company, however, seems to have uncovered fairly reliable ways to evaluate the momentum effect in certain equities. While they have primarily focused on underreaction, a more risk-prone investor might also be interested in overreaction.

As I’ve outlined previously, there are several factors at play here:

1. Markets are not completely rational if we take into account the momentum effect.

2. The momentum effect is caused by overreaction or underreaction to certain events.

3. Overreaction or underreaction happens due to cognitive biases, emotions, and heuristics.

I’d like to add a fourth point here. Both underreaction and overreaction are, in some way, expressed through overall sentiment. We have seen such effects realized through the collection of news headlines and other alternative data sources previously.

Various strategies have been developed that involve ways to measure overall sentiment. Unfortunately, if the information is easily accessible, it likely has been already adjusted to by the market. Such sources might be anything from traditional datasets.

What we’re left with is, mostly, alternative data in its various shapes and forms. As a source, it is usually defined in contrast to traditional ones. So, it’s everything, but fiscal reports, government documents, etc.

Usually, you’ll have things like satellite imagery, news headlines, forum comments, and many others headed under alternative data. The issue, primarily, is that the signals derived from such datasets can be weak (in comparison to traditional sources), but the collection efforts, if manual, are enormous.

Capturing momentum

Web scraping is a method that’s commonly utilized in alternative data-gathering processes. While web scraping can only acquire publicly accessible data from the Internet, slightly limiting its reach, it’s sufficient for sentiment analysis.

For a quick rundown, web scraping is a process in which an automated application goes through some amount of URLs and downloads all the data found within. That data is then parsed and the meaningful bits are extracted, usually output into some data-analysis-ready format.

There have been numerous studies done that show that sentiment analysis has a place in price predictions, producing varying degrees of accuracy and return. As we’ve established that momentum is inextricably linked with the sentiment, there’s reason to believe that we can use web scraping for our benefit.

In fact, in my previously cited source, sentiment analysis and momentum effect have been studied through news headlines and produced some interesting results. Primarily, what has been the issue in using alternative data for momentum effect-driven investment has been the high entry barrier. Alternative data used to be prohibitively expensive, and its quality varied wildly.

Now, however, the tides are changing. Web scraping affords even relatively small groups of researchers or investors to acquire significant datasets from any publicly available source. As such, it can be used to generate alpha through market anomalies, namely the momentum effect.