Are (Stock) Bubbles Rising?

• Global equity markets rebounded strongly in the second half of 2020 and into January 2021, pushing valuations to levels not seen since the tech bubble. This led some to claim markets, or certain segments, may be in a bubble.
• We propose a model for quantifying the “bubbliness” or crowdedness of individual stocks and their peers. This measure can be aggregated to the portfolio level.
• We examined data behind the two “bubbliest” stocks, as of Jan. 27, 2021. One of them, GameStop, was among the most crowded stocks since September 2020, according to our crowding model.

In 2020, the MSCI USA Index gained 21.4%, pushing its aggregate trailing price/earnings (P/E) ratio above 30 for the first time since 2002, as we see in the exhibit below. The returns in the MSCI USA information-technology and consumer-discretionary sector indexes were more than twice as large, prompting some to suspect bubbles may be forming.

MSCI USA Index Had Highest P/E Ratio in Nearly 20 Years in 2020

But how does one identify or quantify a bubble? We propose a framework that allows one to assess the “bubbliness” of stocks and portfolios. Intuitively, we believe that bubbles are largely driven by the same forces, and can be identified by the same characteristics as crowded trades, a topic we have researched extensively at the factor level. Specifically, we believe stocks in a bubble state are characterized by exceptionally high values in the five metrics in the exhibit below.

Metrics Used to Quantify Bubbles and Crowding in Stocks

Specific bubbles may have additional characteristics, such as sentiment as measured by unusual patterns in options-trading activity, news mentions, retail sentiment or discussions on social media. But we believe the five listed above are common to all, and are elevated in bubbly stocks in both the cross section (relative to other stocks) and time series (relative to a stock’s own history).

With the exception of trading volume, the five metrics above are the same used in the MSCI Factor Crowding Model. We have taken these five metrics, evaluated them in the cross section and time series and applied them to individual securities to create a security-level crowding model, which we can use to create an overall crowding score for individual securities. While the model does not explicitly distinguish between crowding on the long or short side, the score of the short-interest component provides an indication of the potential for short squeezes, although our research found that highly crowded stocks generally underperformed over the 1996 to 2020 period.

The Security-Level Crowding Model in Action

The exhibit below lists the 10 most crowded stocks, as of Jan. 27, 2021, according to the security-level crowding model and shows how the scores of the top two, MicroStrategy Inc. and GameStop Corp., evolved since Jan. 1, 2020. Both stocks had extremely high scores on nearly all five crowding metrics on Jan. 27. We note that while GameStop became a very popular bubble story at the end of January 2021, our model placed it among the most crowded stocks since September 2020.

Most Crowded Stocks as of Jan. 27, 2021

Crowding scores are in units of standardized z-scores. Roughly half of stocks will have scores above zero and half will have scores below zero. Typically, only ~2% of stocks will have scores above 2, which we would consider very elevated.

What Other Data Might Help Understand Bubbly Stocks’ Behavior?

We have previously written about unusual options-trading activity in GameStop, and observed high levels of shorting in GameStop from data on hedge-fund holdings.¹ Here we use MSCI FactorLab² data to examine several measures of options and news sentiment or activity for GameStop and MicroStrategy: put/call trading-volume ratio, options-trading volume and news volume. We see that both stocks had enormous increases in options trading, which was initially dominated by calls, but that put volume overtook calls in GameStop after the stock passed USD 100 per share at the end of January.

Options Volume and Put/Call Ratio for GameStop and MicroStrategy

Source: OptionMetrics, MSCI

Average put volume over the last month divided by average call volume over the last month. Source: OptionMetrics, MSCI

Looking at news sentiment, specifically news volume, we see dramatic differences between the two stocks. GameStop gained large media attention, which caused its news volume to increase greatly, while the news volume for MicroStrategy remained much lower and more stable until Feb. 8 when Tesla Inc. announced they purchased USD 1.5 billion in bitcoin and would accept the cryptocurrency to purchase their vehicles.

News Volume for GameStop and MicroStrategy

Average daily news volume in Dow Jones Newswires over the trailing one month. Source: Ravenpack, MSCI

Fear of Crowds?

Many investors may choose to simply avoid crowded or bubbly stocks altogether. But knowing which companies are most similar to the crowded ones may be valuable to compare the behavior and characteristics of a crowded stock to its peers before deciding whether to replace or liquidate it. We used MSCI Peer Similarity Scores³ to find the companies most similar to MicroStrategy and GameStop. Those companies and their crowding scores, as of Jan. 27, are listed below. Note that only some of the peers of both stocks had what we would call moderately high crowding scores — i.e., approaching two.

Crowding Scores of Companies Most-Similar to GameStop and MicroStrategy

We believe understanding crowding can provide insight into the behavior and potential risks of certain “bubbly” stocks and may be a valuable addition to understanding portfolio risk and performance drivers.

¹MSCI Hedge Fund Intel data aggregates the holdings of hedge funds’ long and short positions on a monthly basis.

² MSCI FactorLab is a large library of MSCI-curated factor and descriptor data that incorporates MSCI’s latest factor innovations and can be quickly integrated into investment strategies and research.

³MSCI Peer Similarity Scores analyze the text of company filings, company fundamentals, news mentions, return correlations and ESG key-issue weights to quantify similarity between all pairs of companies.

Further Reading

Is There an Options Sentiment Factor?

MSCI Integrated Factor Crowding Model

MSCI FactorLab

Peering into Peer Selection

Looking for a Better Hedge

Anatomy of Hedge Fund Portfolios