INVESTMENT STRATEGY: Keeping watch on the short sellers

Backgammon boardLong-only investment managers could use hedge fund short-selling data as a sign post for stock prices, finds Lynn Strongin Dodds. But there are questions about whether data provides enough information about market direction and whether it is economical.

Short selling data has not caught the imagination of traditional institutions but the information garnered may help shine light into certain alpha generating trends.

The validity of short interest and broader short selling data to discern investment themes has been a long running topic in academic circles. 

“We have long held the belief that lending data provides valuable and timely market intelligence that should be a factor in both long-only and long/short portfolio managers’ investment decision processes,”  says Chris Holzwarth, head of global sales and relationship management for securities finance, State Street Global Markets. “To date, the use of such data by long-only managers seems to be less frequent than one would anticipate.”

One of the problems, according to David Lewis, a senior vice president in SunGard’s capital markets business, Astec Analytics, is that mainstream fund managers tend to look at only one side of the equation. “I think long-only fund managers are throwing away half of the data available and ignoring the short side of the market. They need to look at where the short side is placing bets so they don’t miss a trick.”

The bulk of the research into this area over the past 25 years has emanated from academia. The collective focus has been on the value that can be extracted from short interest, which is an indicator of what short sellers think about a particular stock. If it increases, then shorts are betting that the price will be heading down, but if it moves in the opposite direction, then the wager is on the stock bottoming out with the potential for an upward swing.

While conclusions vary, the general consensus is that short interest does have some predictive power, although the usual caveats about transaction costs and implementation issues apply.  The oldest study, which is still often referred to is by Douglas Diamond and Robert Verrecchia in 1987. The academics developed the rational expectation model which suggests that “increases in short interest is bad news and only informed investors conduct short selling”.

More recent studies have shown short that sellers do have the ability to forecast future negative abnormal returns, partly because they can adroitly process publicly available information.

However, there have also been reports that found short interest to be a neutral indicator, meaning that short sellers are no more informed than any other investors and that the data should not be used as a reliable indicator of subsequent returns.

One of the main issues with all of these findings is that the calculations relied on short interest data obtained directly from the exchanges, which are available with a significant delay.

The newer studies from firms such as S&P Capital IQ and Deutsche Bank relied on the dataset from Markit, a data firm, which covers more than $13 trillion (€9 trillion) of global securities in the lending programmes of over 20,000 institutional funds. It includes ten years plus of history with over three million intraday transactions.

Astec Analytic also offers a comprehensive picture with loan data for over 31,000 securities, including historical data on inactive securities, and over 31,000 lending portfolios, amounting to 1.5 million transactions daily, or $2 trillion on loan each day.

Although the studies from S&P Capital IQ and Deutsche Bank differed, they both found that this timelier, daily data provided additional signal strength over the traditional lagged data sourced from the exchanges. Both groups used DataExplorer’s database (which Markit acquired in 2012), to construct models using the same key and broad indicators of demand, supply and utilisation in their models to identify trends and whether to go long or short.

Each of these indicators produced their own signals. For example, according to the rational expectations model, stocks with high demand ratios should underperform those with low ratios. If supply is limited, then it could be a sign that a security is difficult to borrow, which leads to a tighter constraint on short selling. Utilisation, which measures the relationship between demand and supply, can offer an insight into how the interplay of demand and supply affects stock price movements while cost is important because if it is too high it can be a sign of limited supply due to low institutional ownership or strong demand, according to the S&P Capital IQ study.

The same study also found that overall daily securities lending information proved to be a valuable resource to derive non-traditional sources of alpha. By using the key indicators it generated an annualised long-short return spread and information ratio of 41% and 1.91% respectively, based on the Russell 3000 index from July 2006 to October 2011. 

James Clunie, fund manager at Jupiter Asset Management, also believes the information allows a greater probing into the behaviour of short sellers. “The data enables us to see the risk management behaviour of short sellers and what positions they may be building over time. It can help you avoid mistakes but it is also important to note that some short sellers are better than others.”

Clunie recently teamed up with academics Panagiotis Andrikopoulos and Antonios Siganos to investigate the extent to which UK short sellers are informed investors, in accordance with Diamond and Verrecchia’s  hypothesis. Using short selling data from Data Explorers from 2004 to 2012, their results suggested that heavily shorted stocks fail to consistently underperform their lightly shorted counterparts. Their ability to predict performance is limited to companies that struggle for survival, such as those about to enter bankruptcy or financial firms during the financial crisis. 

Will Duff Gordon, research director of Markit Securities Finance, says: “In the past, before aggregated data was available, fund managers would glean information about market trends by conducting research in-house with some internalising their securities lending activity. Now, institutional investors can gain insight into short selling activity.” 

Arguably, this is a useful signal for fund managers that can be used as one of many useful tools for asset allocation decisions, along with fundamental research such as earnings and cash flow. Some firms are calling this “quantamental”.

The theory behind combining these two distinct analytical styles is that quantitative and fundamental analysts basically employ similar factors including price-to-earnings multiple, past performance, market cap and sector classification. Putting a quantitative overlay on top of an established fundamental framework can strengthen and optimise outcomes.

Lewis also stresses the importance of looking at the whole picture. Just judging a stock on one fundamental piece of data is only one part of the story. He points to construction company Balfour Beatty. In January to February of 2013, he says, the longs were buying the stock, pushing the share price up. But on the other side, there was heavy short selling, and short sellers made a good return when the company came out with its shock profit warning due to weak performance at its UK construction arm. “If long-only managers had been looking at their activity, they may not have been blindsided.”

Holzwarth adds: “The mechanism by which it can be used to generate long-term alpha could be acting as a potential sell signal for long holders.

“If a portfolio manager saw rising utilisation rates for a particular security, it could prompt a review of the decision to hold that security. Through further analysis, the investment decision could be confirmed or the security sold, hopefully before further deterioration in valuation.”

Although many market participants see the value of using securities lending data as one tool, some question whether it is economical.

“The cost should be weighed against the possible benefits,” says Alec Nelson, specialist in securities finance at consultancy Rules Financial. “There are different trading strategies and a large proportion of securities lending is driven by programme and algo trading. This has an impact on the market but doesn’t necessarily tell you what the absolute direction of the market will be.”

Stuart Jarvis, sales trader in Citi’s equity finance securities lending team, also believes fund managers should look at who is generating the shorts. “It is not as simple as saying there is a lot of hedge fund short selling activity in the market and that may provide a challenge to long-only themes. There is too much noise in the market to get a clean signal which is one of the reasons I think it would be difficult to build a model that accurately identifies when to buy and sell securities off lending data alone.

“This is not to say information from securities lending can’t be useful as a qualitative overlay to fundamental research but there are more factors that need to be considered to produce concrete triggers which could indicate market direction.”

©2013 funds europe