The "Reversal" Play: A Study of Company Size, Industry, and Sector

In technical analysis, the “reversal play” is a well-known strategy used by traders to take advantage of imbalances in momentum. The strategy roots itself in the idea of mean-reversion, large decreases in the share price of a stock are typically followed by a period of increases as the value roughly regresses to a mean. From a fundamental point of view, the investor assumes that a sudden spike in share price (which this data looks at) is not accompanied by an immediate fundamental change, therefore, the company has become undervalued.

This phenomenon has been heavily researched before showing that this strategy is capable of producing excess returns. Marc Bremer and Richard Sweeney in “Reversal of Large Stock-Price Decreases” (1991) found that “extremely large negative 10-day rates of return are followed on average by larger-than-expected positive rates of return over following days.” Bremer and Sweeney found that the recoveries they studied were relatively slow and “inconsistent with the notion that market prices quickly reflect relevant information.” This conclusion seems to question the legitimacy of the efficient market hypothesis, a rule that is often rebutted by the existence of reversal plays.

MIT researcher Wesley Chan published “Stock Price Reaction to News and No-News: Drift and Reversal After Headlines” (2001) documenting reversals accompanied by news. His results lead to two conclusions, “first, investors are slow to respond to valid information,” and “second, investors overreact to price shocks, causing ‘excess’ trading volume and volatility.” Once again, academic research suggests that the efficient market hypothesis can’t explain these plays which show a pattern of excess return.

It is in this spirit of a loose interpretation of the efficient market hypothesis that this paper is written. Information asymmetry and a wide variety of valuations help fuel movements in a stock’s price that doesn’t seem fueled by fundamental shifts. This research will look at these movements across different sectors, industries, and company sizes to bring to light any trends of the reversals within those subcategories.

Method

The data was collected from Alpha Vantage which provides stock price data going back to 2000. In particular, this data looks at 1,824 different companies with market capitalizations above 2 billion dollars. This constraint on company size has been enforced because of the volatility of smaller companies with low share prices that move in erratic manners not representative of the technical environment in which reversals occur. These companies are organized in 11 sectors and 70 industries based on their main line of business. In total, 110,404 data points of share price drops in the range of 3 percent and 50 percent were used. Table 1 shows the distribution of these points by sector.

Sector	Data Points	Tickers
Basic Materials	11,255	127
Communications Services	9,461	53
Consumer Cyclical	14,740	243
Consumer Defensive	15,297	93
Energy	15,069	148
Financial Services	18,569	284
Healthcare	4,787	166
Industrials	5,283	226
Real Estate	3,332	128
Technology	9,620	274
Utilities	2,995	81
Total	110,404	1,824

Table 1. Size of data set and tickers used for each sector.

Data Distribution

For each of the drops observed, measures of price movement 5-days, 10-days and 20-days were recorded to determine the development of the reversal. These values could take on negative or positive values of varying magnitudes. In all sets of the rebound data for most sector, large right tails developed and formed skewed distributions. The average skewness of each sector’s rebound set (5-day, 10-day, and 20-day rebound) is shown in Table 2.

Basic Materials	10.21
Communications Services	9.45
Consumer Cyclical	5.70
Consumer Defensive	2.26
Energy	31.26
Financial Services	16.94
Healthcare	11.94
Industrials	39.26
Real Estate	28.76
Technology	51.9
Utilities	1.29

Table 2. Average skewness of rebound data sets of 5-days, 10-days, and 20-days. Positive values represent skewness to the right.

The skewness of the datasets shows which sectors are prone to above average returns when a reversal develops. Technology has the largest right skew reflecting its accelerated movement in the current stock market. As illustrated by the boom and bust in the early 2000’s, these stocks are prone to speculation and, therefore, volatile movement. However as technology grew fundamentally strong, the reversals became stronger. Energy and Industrials were the next highest skewed with their dependence on fluctuating commodity prices. Movements in commodity prices can often cause unbased drops in a stock which is otherwise fundamentally strong. The Utilities and Consumer Defensive sectors came in as the lowest as these stocks typically move slower than the market and are shielded from speculation. Reversals are typically tepid on the positive side in these sectors.

It is interesting to note that the distributions of the rebound sets are not just skewed but lognormal. The rebound data sets can be transformed into a roughly normal data set, through the following calculation. After this transformation, the skewness of each distribution is update in Table 3.

:

Basic Materials	-0.46
Communications Services	0.14
Consumer Cyclical	-0.09
Consumer Defensive	-1.26
Energy	1.39
Financial Services	-0.51
Healthcare	0.03
Industrials	0.15
Real Estate	1.50
Technology	0.84
Utilities	-1.00

Table 3. Average skewness of rebound data sets of 5-days, 10-days, and 20-days after transformation

After the logarithmic transformation, the distributions’ skewness falls dramatically and even become negative. This shape only occurs in instances of a heavy positive tail suggesting that reversal developments have smaller negative extremes relative to the positive extremes. While negative outcomes are still likely, the magnitudes of these observations are significantly lower. This is an important observation to be made by an investor looking to reduce risk of large drawdowns and increase the chances of an extremely positive result.

Analysis

This paper seeks to identify various sector and industry trends within the reversal data. To do this, one can observe the recorded average drops and average 5-day, 10-day, and 20-day rebounds. For the following analysis, visuals can be found on the author’s Tableau workbook titled, “Average Share Price Drop and Rebound By Sector and Industry.” Chart 1 is an extract from the workbook showing the average drop and 5-day rebound of all the industries.

Chart 1. Drop and 5-day rebound by industry (colored by sector).

There is a low positive correlation between average drop and average 5-day rebound when organized by industry. A few outliers stick out. Biotechnology within Healthcare had drops that averages 7.52 percent in magnitude and an average 5-day rebound of 3.28 percent. Volatility in this industry is significantly higher than its peers within Healthcare due to pharmaceutical trials that, more or less, determine a company’s future. Although, this is mostly true for smaller-sized biotechnology companies. Communication Equipment in Technology is another industry that showed drops and similar swings of high magnitude.

Of the most resilient industries, Specialty Insurance within Financial Services and Homebuilding and Construction within Consumer Cyclical were the most. Within 5-days, Insurance had recovered more than half lost in an average drop while Homebuilding and Construction had almost recovered two-thirds of the average drop. Interestingly enough, both industries might be the ones that are affected by natural disasters the most. Natural disasters are the ideal event for a reversal as, typically, a company’s fundamentals do not change and the negative effects are temporary and often exaggerated.

Industries that had the strongest drift from 5-days out to 20-days out (20-day-5-day spread) included Insurance, Biotechnology, Communication Equipment, Coal, Airlines, and Metals and Mining. Coal is an irrelevant case because of its unwinding over the past decade. Three of the industries with the largest spread are those already identified as the most vulnerable based on the 5-day rebound number, therefore it makes sense that these are seeing strong drift. The interesting observations are Metals and Mining and Airlines. Metals and Mining and Airlines grew 1.89 percent and 2.29 percent from 5-days after to 20-days after. Both industries tend to be capital intensive and have large barriers of entry.

Chart 2. Average 5-day, 10-day, and 20-day rebound by market capitalization.

The chart above shows each average rebound measure by market capitalization. Three size groups are represented in the data, mid-capitalization ($2 billion - $10 billion), large-capitalization ($10 billion - $50 billion), and mega-capitalization ($50 billion and over). Following the general observation that smaller stocks tend to be more volatile, the data shows this trend to be true for reversals as well. The 5-day rebound averages are relatively equal, but the drift to 20-days is strongest in mid-caps. This is also explained by a larger amount of extreme positive rebounds that pull the mid-cap data higher.