Technical Trading

How does pairs trading of relatively diversified exchange-traded funds (ETF) compare with pairs trading of individual stocks? In his July 2016 paper entitled “Does Pairs Trading with ETFs Work?”, Philipp Doering tests a common pairs trading process on the universe of ETFs. Specifically, he:

1. Each month, scales all ETF prices to $1 and computes the sum of squared price deviations (SSD) for all possible pairs over the next 12 months, excluding those with missing prices.
2. Rescales prices of pairs with the lowest SSDs to $1 and, during the next six months, buys (sells) the relative loser (winner) of each of these low-SSD pairs whenever their prices diverge by more than two selection-interval standard deviations.
3. Closes pair trades when prices converge, one of the ETFs discontinues trading or the end of the 6-month trading interval, whichever occurs first. Within the trading interval, selected pairs may trade more than once.

He scales returns by the number of pairs selected for trading and averages across the six overlapping pairs portfolios to compute return on committed capital. He performs this process also for individual stocks for comparison. Using daily trading data for broad samples of ETFs and stocks during January 2001 through June 2016, he finds that: Keep Reading

How can investors integrate global asset class diversification, pre-eminent factor premiums and crash protection? In his July 2016 paper entitled “The Trinity Portfolio: A Long-Term Investing Framework Engineered for Simplicity, Safety, and Outperformance”, Mebane Faber summarizes a portfolio combining these three principles, as follows:

1. Global diversification: Include U.S. stocks, non-U.S. developed markets stocks, emerging markets stocks, corporate bonds, 30-year U.S. Treasury bonds, 10-year foreign government bonds, U.S. Treasury Inflation-Protected Securities (TIPS), commodities, gold and Real Estate Investment Trusts (REIT) .
2. Value/momentum screens: For U.S. stocks, each month first rank stocks by value and momentum metrics and then pick those with the highest average ranks. For non-U.S. stocks, each month pick the cheapest overall markets. For bonds, each month pick those with the highest yields.
3. Trend following for crash avoidance: For each asset each month, hold the asset (cash) if its price is above (below) its 10-month SMA at the end of the prior month.

The featured “Trinity” portfolio allocates 50% to a sub-portfolio based on principles 1 and 2 and 50% to a sub-portfolio based on principles 1, 2 and 3. Using monthly returns for the specified asset classes during 1973 through 2015, he finds that: Keep Reading

Do stock returns during extended market hours (4:00PM-8:00PM and 4:00AM-9:30AM) reliably predict subsequent returns during normal market hours? In their July 2016 paper entitled “Are Extended Hours Prices Predictive of Subsequent Stock Returns?”, Shai Levi, Joshua Livnat, Li Zhang and Xiao-Jun Zhang investigate whether extended hours stock returns predict returns the next day and over subsequent longer drift intervals. They focus on stocks with extended hours news (earnings releases, analyst rating changes or SEC form filings). They hypothesize that relatively informed institutional investors dominate extended hours trading and that their trading immediately on news reflects information rather than a need for liquidity. Using normal and extended hours stock returns and volumes for sessions around relevant news releases and for firms without news releases from News Quantified during 2006 through 2014, returns over subsequent drift intervals and earnings surprise data, they find that: Keep Reading

Extended stock market index winning and losing streaks elicit speculation about pending reversals. Does evidence support the what-goes-up-must-come-down view that likelihood of reversal grows with streak duration and magnitude? To check, we examine the “modern” (since 1990) behaviors of the S&P 500 Index and NASDAQ Composite Index during the one and two trading days after winning and losing streaks of at least three days. Using daily closes, highs and lows for the S&P 500 Index and the NASDAQ Composite Index during January 1990 through June 2016 (6,681 trading days), we find that: Keep Reading

Do equity trade recommendations from technical analysis experts beat the market? In his February 2016 paper entitled “Are Chartists Artists? The Determinants and Profitability of Recommendations Based on Technical Analysis”, Dirk Gerritsen evaluates technically based buy and sell recommendations for individual Dutch stocks and the AEX index. Specifically, he measures abnormal performance from 10 trading days before (including the publication date) through 20 trading days after recommendations. For individual stocks, “abnormal” means in excess of the return estimated by the four-factor (market, size, book-to-market, momentum) model. For the AEX index, abnormal means in excess of average index return over the year preceding the 30-day measurement interval. For recommendations that include stop-loss instructions, he measures also abnormal asset performance after any stop-loss actions. Finally, he examines whether recommendations agree with the consensus of eight kinds of simple technical trading rules. Using daily stock and and AEX index prices, total returns and trading volumes associated with 5,017 recommendations (3,967 with 500 stop-losses for individual stocks and 1,050 with 242 stop-losses for the index) from 101 experts on the Dutch stock market during 2004 through 2010, he finds that: Keep Reading

Should investors focus on past Sharpe ratio when picking individual stocks? In their June 2016 paper entitled “Don’t Stand So Close to Sharpe”, Angel Leon, Lluis Navarro and Belen Nieto compare 32 past performance metrics for effectiveness in selecting large capitalization U.S. stocks. They categorize these metrics into four groups:

1. Eight related to Sharpe ratio.
2. Six partial moment formulas (based on downside, or both downside and upside return variability, including Sortino and Omega ratios as special cases) for different levels of gain seeking/loss avoidance investment styles.
3. 14 tail risk measurements (such as value at risk) for different levels of gain seeking/loss avoidance investment styles.
4. Four measures of average return per unit of risk that do not fit within the other three groups.

Their asset sample is all stocks continuously in the S&P 500 Index over the sample period. They rank these stocks daily over a 264-day rolling window of past returns for each of the 32 metrics and reform respective equally weighted portfolios of the top 20 stocks (about 5%). They compare these portfolios based on next-day return statistics and on overlap of stocks selected, with the Sharpe ratio portfolio as a benchmark. Using daily total returns for the 424 stocks that are continuously members of the S&P 500 Index during January 2005 through September 2014, they find that: Keep Reading

Is nearness to 52-week highs or lows informative about future stock returns? In their June 2016 paper entitled “Nearness to the 52-Week High and Low Prices, Past Returns, and Average Stock Returns”, Li-Wen Chen and Hsin-Yi Yu examine the power of extreme price levels (52-week highs and lows) to predict stock returns, and whether any such predictive power is distinct from the momentum effect. They focus on the left (right) tail of nearness to 52-week low (high), because these stocks may attract the most investor attention. They determine 52-week highs and lows with monthly data. Specifically, they each month form value-weighted portfolios that are:

1. Long the bottom 10% and short the top 90% of stocks sorted on nearness to 52-week low.
2. Long the top 10% and short the bottom 90% of stocks sorted on nearness to 52-week high.
3. For comparison, long the top 10% and short bottom 10% based on returns from 12 months ago to one month ago (momentum strategy).

Using monthly prices (ignoring dividends) for a broad sample of non-financial common U.S. stocks and monthly factor portfolio returns during July 1962 through December 2014, they find that: Keep Reading

Does availability of liquid tracking products change short-term trending/reversal tendencies of equity indexes? In their May 2016 paper entitled “Indexing and Stock Market Serial Dependence Around the World”, Guido Baltussen, Sjoerd van Bekkum and Zhi Da investigate how introduction of index futures, exchange-traded funds (ETF) and mutual funds affects measures of index serial dependence. They hypothesize that technical interplay in index products among investors, market makers and arbitrageurs stimulates short-term reversal. They measure serial dependence with daily lags and one-week lag in two ways: (1) simple autocorrelations; and, (2) returns to a “MAC(5)” trading strategy based on a weighted average of autocorrelations for lags 1 to 4, with positive (negative) returns indicating trending (reversal). Using daily data for 21 major global equity indexes and associated index futures and ETFs and for mutual funds tracking the S&P 500 Index as available through mid-May 2013, they find that: Keep Reading

When does a cointegration test, which looks for a connection between two apparently wandering price paths, work for pairs trading? In their May 2016 paper entitled “Cointegration and Relative Value Arbitrage”, Binh Do and Robert Faff investigate the conditions under which cointegration successfully identifies stocks for pairs trading. Their basic pairs trading strategy is to each month:

1. Identify cointegrated pairs based on daily total returns over the last 12 months.
2. Over the next six months, buy (sell) the relatively undervalued (overvalued) stock when cointegrated pair spread exceeds its selection interval mean by two standard deviations.
3. Close positions when the spread reverts to its historical mean or the trading period ends, whichever occurs first.
4. Closed trades may be reopened as signaled, if there is more than a month left in the trading interval.

They then refine the strategy by constraining selected pairs to those that are close economic substitutes, corresponding to a low cointegration coefficient. Pairs passing (failing) this constraint move together in the long run without any price scaling (only with scaling of prices for one member of the pair). While they focus on pairs of individual stocks, they also consider trading of pairs of small groups (baskets) of stocks. Their benchmark is a conventional pairs trading strategy that identifies pairs with the smallest sums of squared differences in normalized daily prices over the past 12 months, and then trades as specified above over the next six months. Using daily data for a broad sample of U.S. common stocks during July 1962 through December 2013, they find that: Keep Reading

A reader asked whether the gold-gold miner stocks arbitrage-like argument in Jay Kaeppel’s February 2010 article “Don’t Give Up On Gold Stocks Just Yet” (for which his September 2004 article “Gold Stock and Gold Bullion” is a more robust antecedent) supports frequent timing of these assets. For example, if SPDR Gold Shares (GLD) and Market Vectors Gold Miners GDX) diverge over some recent interval, do they then reliably converge quickly? To check, we examine the relative price behaviors of these funds. Using weekly dividend-adjusted closes for GLD and GDX during late May 2006 (inception for GDX) through mid-May 2016, we find that: Keep Reading