Technical Trading

A subscriber asked for confirmation that the strategy described at “Meb Faber’s 12-Month High Switch” (the strategy) is attractive. This strategy at each monthly close:

• Calculates the 12-month high for: SPDR S&P 500 ETF Trust (SPY), iShares MSCI EAFE ETF (EFA), Vanguard Real Estate Index Fund ETF Shares (VNQ), SPDR Gold Shares (GLD) and Invesco Optimum Yield Diversified Commodity Strategy No K-1 ETF (PDBC) based on dividend-adjusted prices.
• Allocates 20% of the portfolio to each asset currently within 5% of its 12-month high.
• Allocates any remaining funds to iShares 7-10 Year Treasury Bond ETF (IEF) if IEF is within 5% of its 12-month high, or otherwise to cash with return estimated by the yield on 3-month U.S. Treasury bills (T-bill).
• Rebalances to equal weights each month and applies trading frictions of 0.1% per 1-way trade for any switches in holdings.

To investigate, we replicate the strategy, substituting Invesco DB Commodity Index Tracking Fund (DBC) for PDBC to get a longer sample period that includes the 2008-2009 financial crisis. We use an equal-weighted (EW), monthly rebalanced combination of SPY, EFA, VNQ, GLD and DBC as a benchmark. We apply 0.1% frictions to ETF switches but not to simple rebalances, which are generally small. Using monthly dividend-adjusted prices for SPY, EFA, VNQ, GLD and DBC during February 2006 (DBC inception) through June 2024, we find that: Keep Reading

Is industry trend-following an attractive strategy over the long run? In their June 2024 paper entitled “A Century of Profitable Industry Trends”, Carlo Zarattini and Gary Antonacci evaluate the long-term performance of a long-only industry trend-following (Timing Industry) strategy, modeled as follows:

• Entry – buy an industry when its daily closing price crosses above the upper band of either its 20-day Keltner Channel (with a multiplier of 2 for the high-low price range component) or its 20-day Donchian Channel.
• Sizing – each day for each open position, calculate 14-day past return volatility as an estimate of its future volatility and resize all open positions so that they contribute equally to overall portfolio volatility, limiting overall portfolio leverage to 200%.
• Exit – each day for each open position, close the position if it crosses below a stop loss represented by the lower band of either its 40-day Keltner Channel (again with a multiplier of 2 for the high-low price range component) or its 40-day Donchian Channel. However, do not ever lower the stop loss. When a position closes, reinvest proceeds into 1-month U.S. Treasury bills.

For a long-term test, they apply these rules to nearly 98 years of daily returns for 48 hypothetical annually rebalanced, capitalization-weighted industry portfolios constructed by assigning a Standard Industrial Classification (SIC) Code to each stock traded on NYSE, AMEX and NASDAQ. For a recent and more realistic test, they apply these rules to 31 sector exchange-traded funds (ETF) offered by State Street Global Advisors. Utilizing daily returns for the 48 industry portfolios since July 1926 and for the 31 sector ETFs as available (inceptions January 2005 to June 2018), all through March 2024, they find that:

Keep Reading

Is the Buffett Indicator, the ratio of total U.S. stock market capitalization (proxied by Wilshire 5000 Total Market Index W5000) to U.S. Gross Domestic Product (GDP), a useful indicator of future U.S. stock market performance? W5000/GDP clearly has no stable average value over its available history (see the first chart below), so the level of the ratio is not a useful predictor. We therefore consider the following variables based on W5000/GDP as predictors of W5000 returns at horizons up to two years:

1. Quarterly change in W5000/GDP.
2. Average quarterly change in W5000/GDP over the past two years (eight quarters).
3. Average quarterly change in W5000/GDP over the past five years (20 quarters).
4. Slope of W5000/GDP over the past two years.
5. Slope of W5000/GDP over the past five years.

We consider two kinds of tests: (1) a linear test relating past changes in these variables to future W5000 returns up to two years; and, (2) a non-linear test calculating average next-quarter W5000 returns by ranked fifths (quintiles) of past changes in these variables. Using quarterly levels of W5000 (with extension), Shiller’s P/E10 lagged by one quarter and quarterly GDP lagged by one quarter during the first quarter of 1971 through the first quarter of 2024, we find that: Keep Reading

A subscriber asked: “Some pundits have noted that U.S. stocks have greatly outperformed foreign stocks in recent years. What does the performance of U.S. stocks vs. foreign stocks over the last N years say about future performance?” To investigate, we use the S&P 500 Index (SP500) as a proxy for the U.S. stock market and the ACWI ex USA Index as a proxy for the rest-of-world (ROW) equity market. We consider three ways to relate U.S. and ROW equity returns:

1. Lead-lag analysis between U.S. and ROW annual returns to see whether there is some cycle in the relationship.
2. Multi-year correlations between U.S. and next-period ROW returns, with periods ranging from one to five years.
3. Sequences of end-of-year high water marks for U.S. and ROW equity markets.

For the first two analyses, we relate the U.S. stock market to itself as a control (to assess whether ROW market behavior is distinct). Using monthly levels of the S&P 500 Index and the ACWI ex USA Index during December 1987 (limited by the latter) through April 2024, we find that: Keep Reading

Is momentum better measured by a granular fitted line or beginning-to-end return? In their March 2024 paper entitled “Trended Momentum”, Charlie Cai, Peng Li and Kevin Keasey investigate use of an analytically/visually clear linear stock price trend to enhance conventional momentum. They measure price trend clarity (TC) as R-squared for a regression of daily price versus date over the past 12 months. Specifically, they each month:

• Sort stocks into fifths (quintiles) based on conventional momentum, return from 12 months ago to one month ago.
• Further sort the top momentum quintile into finer quintiles based on TC.
• Form  equal-weighted or value-weighted portfolios of resulting sorts and compute their gross returns and 3-factor (market, size, book-to-market) alphas over the next six months.

To confirm use of TC to measure clarity of price trend, they separately conduct an experiment that relates analytical TC to trend clarity perceived by sample of 128 individuals each evaluating 10 pairs of stock charts. Their sample includes daily price data for U.S. common stocks from January 1927 through December 2020. Analyses requiring earnings start in 1964, while those involving investor sentiment span 1967 through 2018. They groom all variables to exclude outliers. In further analyses, they employ global stock price data. Using the specified methodology and data, they find that:

Keep Reading

Can day traders get rich with an Opening Range Breakout (ORB) strategy that buys (sells) unusually active stocks with positive (negative) opens that break out to new highs (lows) during the first five minutes of the trading day? In their February 2024 paper entitled “A Profitable Day Trading Strategy For The U.S. Equity Market”, Carlo Zarattini, Andrea Barbon and Andrew Aziz test a 5-minute ORB applied to stocks with unusually high daily trading volume (Stocks in Play). Rules for this strategy start with screening listed U.S. stocks for:

1. Opening price above $5.
2. Average daily trading volume at least 1,000,000 shares during the last 14 trading days.
3. Average True Range (ATR) over the last 14 days more than $0.50.
4. Opening range interval volume relative to the last 14 days (Relative Volume) at least 100% and among the 20 with the highest Relative Volumes.

Each day for each stock surviving this screen, they place a stop order to buy (sell) if the stock moves up (down) in the first five minutes and then again reaches the high (low) of this range after the first five minutes. For each executed trade, they set a stop-loss order at 10% ATR distance from the executed entry price. If the stop loss does not trigger intraday, they close the trade at the market close. They size each trade such that the loss on a triggered stop-loss would be 1% of capital deployed and impose a 4X leverage constraint. They assume $25,000 starting capital and impose $0.0035 per share commission (per Interactive Brokers Pro Tiered as of December 31, 2023). Using the specified data for all U.S.-listed stocks (over 7,000) during January 2016 through December 2023, they find that:

Keep Reading

“Turn-of-the-Month Effect Persistence and Robustness” indicates that average absolute returns during the turn-of-the-month (TOTM) are strong for both bull and bear markets. Does a strategy of capturing all bull market returns and TOTM returns only during bear markets perform well? To investigate, we apply four strategies to SPDR S&P 500 ETF Trust (SPY) as a tradable proxy for the stock market:

1. SPY – buy and hold SPY.
2. SMA200 – hold SPY (cash) when SPY closes above (below) its 200-day simple moving average (SMA200) the prior day.
3. TOTM – hold SPY from the close five trading days before through the close four trading days after the last trading day of each month and cash at all other times (TOTM).
4. SMA200 or TOTM – hold SPY when SPY closes above its 200-day SMA the prior day and otherwise use the TOTM strategy.

We explore sensitivities of these strategies to a range of one-way SPY-cash switching frictions, with baseline 0.1%. Using daily dividend-adjusted SPY from the end of January 1993 through early January 2024 and contemporaneous 3-month Treasury bill (T-bill) yields as the return on cash, we find that: Keep Reading

What are optimal intrinsic/absolute/time series momentum (IM) and simple moving average (SMA) lookback intervals for different asset class proxies? To investigate, we use data for the following eight asset class exchange-traded funds (ETF), plus Cash:

• Invesco DB Commodity Index Tracking (DBC)
• iShares JPMorgan Emerging Markets Bond Fund (EMB)
• iShares MSCI EAFE Index (EFA)
• SPDR Gold Shares (GLD)
• iShares Russell 2000 Index (IWM)
• SPDR S&P 500 (SPY)
• iShares Barclays 20+ Year Treasury Bond (TLT)
• Vanguard REIT ETF (VNQ)
• 3-month Treasury bills (Cash)

For IM tests, we invest in each ETF (Cash) when its return over the past one to 12 months is positive (negative). For SMA tests, we invest in each ETF (Cash) when its price is above (below) its average monthly price at the ends of the last two to 12 months. We focus on compound annual growth rate (CAGR) and maximum drawdown (MaxDD) as key metrics for comparing different IM and SMA lookback intervals since earliest ETF data availabilities based on the longest IM lookback interval. Using monthly dividend-adjusted closing prices for the asset class proxies and the yield for Cash over the period July 2002 (or inception if not available by then) through December 2023, we find that:

Keep Reading

“Distance Between Fast and Slow Price SMAs and Stock Returns” finds that extreme distance between a 21-trading day simple moving average (SMA) and 200-trading day SMA, as applied to individual U.S. stock price series, may be a useful return predictor. Does this finding apply to non-U.S. stock market indexes? In their December 2023 paper entitled “Market Timing with Moving Average Distance: International Evidence”, Menachem Abudy, Guy Kaplanski and Yevgeny Mugerman test ability of the distance between fast and slow SMAs to predict future returns across 92 international stock market indexes. Specifically, they each month:

• Measure moving average distance (MAD) for each index as the ratio of its 30-calendar day SMA to its 300-calendar day SMA in local currencies.
• Sort the indexes according to MAD into fifths (quintiles) or tenths (deciles).
• Reform an equal-weighted hedge portfolio that is long indexes in the top quintile or decile with MAD values above one and short indexes in the bottom quintile or decile with MAD values below one.
• Adjust portfolio returns to U.S. dollars via local currency exchange rates.

They consider the full sample of 92 indexes and three subsamples: (1) 46 countries with the highest United Nations development ratings; (2) the MSCI 25 developed markets; and, (3) the MSCI 30 emerging markets. Their benchmarks are buy-and-hold the MSCI World Index (large and mid-size firms in 23 developed countries) and the S&P Global 1200 (30 markets representing about 70% of global market capitalization). Using daily levels of 92 international stock market indexes as available since June 1980, associated U.S. dollar exchange rates and international stock factor model returns, all through November 2020, they find that: Keep Reading

Does degree of difference between fast and slow simple moving averages (SMA) for a stock price series predict future stock return? In the December 2023 revision of their paper entitled “Moving Average Distance as a Predictor of Equity Returns”, Doron Avramov, Guy Kaplanski and Avanidhar Subrahmanyam test distance between a 21-day SMA (SMA21) and 200-day SMA (SMA200) for the stock price series as a return predictor. Specifically, they each month:

• Calculate for each stock the SMA21-to-SMA200 ratio.
• Sort stocks into tenths (deciles) by this ratio.
• Calculate the standard deviation of the ratio across all stocks.
• Select stocks from the top decile with ratios greater than one plus one standard deviation for a long portfolio. Select stocks from the bottom decile with ratios less than one minus one standard deviation for a short portfolio. 
• Specify the Moving Average Distance (MAD) for a stock as 1 if it is in the long portfolio, -1 if it is in the short portfolio, and 0 otherwise. Stocks in the two portfolios are market capitalization-weighted.

They then assess the magnitude and reliability of MAD portfolio performance. They estimate breakeven trading frictions for MAD portfolios based on zero alpha relative to different factor models of stock returns. To assess uniqueness of MAD indications, they control for 18 firm characteristics and several technical indicators. Using daily prices adjusted for splits and dividends for publicly traded U.S. common stocks priced at least $5 during July 1977 through December 2018, they find that:

Keep Reading