Momentum Investing

Is stock price all-time high a consistently effective trigger for trend following? In their January 2025 paper entitled “Does Trend-Following Still Work on Stocks?”, Carlo Zarattini, Alberto Pagani and Cole Wilcox revisit and extend the results of prior 2005 research on long-only trend following based on stock price all-time high that used 1980-2004 data. They assemble survivorship bias-free data for all liquid U.S. stocks backward through 1950 and forward through November 2024. They define liquid as unadjusted closing price above $10 and 42-trading day average dollar volume over $1 million (adjusted downward for past years based on inflation) at the time of portfolio reformation. Their trading/rebalancing rules are:

• For each qualifying stock not in the portfolio, if the daily adjusted close (considering splits and dividends) reaches or exceeds its all-time high, buy the stock at the next open.
• For each stock in the portfolio, if the daily close is below the daily stop-loss level, sell the stock at the next open. The daily stop-loss level is either the previous stop-loss or, if higher, a new stop-loss computed from the All-Time High price (ATH) and 42-day Average True Range (ATR) at the most recent close, as follows:

For portfolio testing, they focus on Russell 3000 stocks from 1991 through the end of the sample period. They use 42-trading day actual volatilities to set position sizes for selected stocks to achieve approximately equal expected contributions to a 30% annualized portfolio target volatility. They allow up to 200% leverage to achieve these positions sizes, with adjustment to position size when higher leverage is indicated. They recompute stock weights at each close to reflect new portfolio entries and exits and changes in expected stock volatilities. They assume frictions/costs that cover broker commissions, slippage (impact of trading) and interest/borrowing costs. Using daily interest rates and daily prices, dividends and other price adjustments for a broad sample of U.S. stocks during January 1950 through October 2024, they find that:

Keep Reading

A subscriber wondered whether choosing the fourth place asset class exchange-traded fund (ETF) rather than the third place class ETF for monthly reformation of the Simple Asset Class ETF Momentum Strategy (SACEMS) would matter if the difference in respective past returns over the ranking interval is less than 0.5%. To investigate, we take a broad, systematic approach and test the following two scenarios:

1. Impose a buffer of -0.5% when reforming the SACEMS portfolio. Specifically, each month subtract 0.5% from the past returns of the first, second and third places of last month before reranking. This test captures the subscriber question as a subset, but tends to increase trading due to small ranking return differences.
2. Impose a buffer of 0.5% when reforming the SACEMS portfolio. Specifically, each month add 0.5% to the past returns of ETFs for the first, second and third places of last month before reranking. This test tends to suppress trading due to small ranking return differences. 

For the second scenario, we also look at effects of buffers larger than 0.5% for the Equal-Weighted (EW) Top 2 SACEMS portfolio. Using monthly SACEMS outputs during June 2006 through November 2024, we find that: Keep Reading

Does degree of co-movement among commodity exchange-traded funds (ETF) predict whether momentum or reversal is imminent? In their September 2024 paper entitled “How to Improve Commodity Momentum Using Intra-Market Correlation”, Radovan Vojtko and Margaréta Pauchlyová investigate whether the relationship between short-term and long-term average pairwise return correlations indicates when to pursue momentum and when to pursue reversal among commodity ETFs. Based on prior research, they consider four ETFs: DBA (agriculture), DBB (base metals), DBE (energy) and DBP (precious metals). Their strategies consists of each month:

1. Ranking the four ETFs by 12-month past return.
2. Calculating average pairwise 20-day and 250-day daily return correlations for the four ETFs.
3. If the average short-term correlation is higher (lower) than the average long-term correlation, executing an equal-weighted momentum (reversal) strategy by buying (selling) the two top-ranked ETFs and selling (buying) the two bottom-ranked ETFs.

Using daily adjusted closes for the selected ETFs from the end of 2007 through early 2024, they find that: Keep Reading

Is momentum a rational firm earnings growth proxy rather than a manifestation of investor underreaction/overreaction to news? In their August 2024 paper entitled “A Unified Framework for Value and Momentum”, Jacob Boudoukh, Tobias Moskowitz, Matthew Richardson and Lei Xie present an asset pricing model that treats value and momentum as complementary inputs to a present value of earnings estimate. They view momentum, return from 12 months ago to one month ago, as a noisy proxy for earnings growth. They test this view by relating momentum retrospectively to actual earnings growth. They further construct an asset pricing model based on a single growth-adjusted value factor and compare its effectiveness to that of the widely used 4-factor (market, size, book-to-market, momentum) model. They calculate growth-adjusted value factor returns via monthly, 5-year smoothed bivariate value-growth regressions, with three alternatives for earnings growth adjustment: (1) momentum as a proxy for growth; (2) a combination of momentum and analyst earnings forecasts as a proxy for growth; and, (3) retrospective actual earnings. They focus on individual U.S. stocks, but also look at U.S. industries, stocks across 23 developed equity markets and Japanese stocks. Using monthly book-to-market ratios, stock returns, next-year earnings growth forecasts and actual annual earnings as available for Russell 3000 stocks since the end of March 1984, for stocks in 23 developed country markets since the end of January 1989 and for stocks in the MSCI Japan Index since the end of August 1988, all through December 2019, they find that:

Keep Reading

A subscriber requested evaluation of versions of the Simple Asset Class ETF Value Strategy (SACEVS), the Simple Asset Class ETF Momentum Strategy (SACEMS) and a combination of the two, as follows:

1. For SACEVS Best Value, which chooses the asset proxy for the term, credit or equity premium that is most undervalued (if any), use only every third signal rather than every monthly signal. We start these quarterly signals with the first available signal in July 2002.
2. For SACEMS Equal-Weighted (EW) Top 2, use only every third signal rather than every monthly signal, rebalancing to equal weights only with these quarterly signals. We align quarterly signals with those for SACEVS.
3. For the SACEVS-SACEMS combination, use 30%-70% weights (per “SACEVS-SACEMS for Value-Momentum Diversification”) and rebalance only quarterly. We align quarterly rebalancings with those for SACEVS.

The overall approach is to suppress trading by limiting portfolio reformations/rebalancings to quarterly, while retaining the informativeness of monthly inputs. We apply these changes and compare results to those for the tracked (baseline) versions of SACEVS Best Value, SACEMS EW Top 2 and their 50%-50% combination. Using monthly SACEMS and SACEVS data during July 2006 (limited by availability of a commodities proxy in SACEMS) through June 2024, we find that: Keep Reading

“Beat the Market with Hot-Anomaly Switching?” concludes that “a trader who periodically switches to the hottest known anomaly based on a rolling window of past performance may be able to beat the market. Anomalies appear to have their own kind of momentum.” Does momentum therefore work for style-based exchange-traded funds (ETF)? To investigate, we apply a simple momentum strategy to the following six ETFs that cut across market capitalization (large, medium and small) and value versus growth:

iShares Russell 1000 Value Index (IWD) – large capitalization value stocks.
iShares Russell 1000 Growth Index (IWF) – large capitalization growth stocks.
iShares Russell Midcap Value Index (IWS) – mid-capitalization value stocks.
iShares Russell Midcap Growth Index (IWP) – mid-capitalization growth stocks.
iShares Russell 2000 Value Index (IWN) – small capitalization value stocks.
iShares Russell 2000 Growth Index (IWO) – small capitalization growth stocks.

We test a simple Top 1 strategy that allocates all funds each month to the one style ETF with the highest total return over a specified momentum ranking (lookback) interval. We focus on a 6-month ranking interval as often used in prior research, but test sensitivity of findings to ranking intervals ranging from one to 12 months. As benchmarks, we consider an equal-weighted and monthly rebalanced combination of all six style ETFs (EW All), and buying and holding SPDR S&P 500 (SPY). As an enhancement we consider holding the Top 1 style ETF (3-month U.S. Treasury bills, T-bills) when the S&P 500 Index is above (below) its 10-month simple moving average at the end of the prior month (Top 1:SMA10), with a benchmark substituting SPY for Top 1 (SPY:SMA10). We employ the performance metrics used for SACEMS. Using monthly dividend-adjusted closing prices for the six style ETFs and SPY, monthly levels of the S&P 500 Index and monthly yields for T-bills during August 2001 (limited by IWS and IWP) through June 2024, we find that:

Keep Reading

“Simple Sector ETF Momentum Strategy” investigates performances of simple momentum trading strategies for the following nine sector exchange-traded funds (ETF) executed with Standard & Poor’s Depository Receipts (SPDR):

Materials Select Sector SPDR (XLB)
Energy Select Sector SPDR (XLE)
Financial Select Sector SPDR (XLF)
Industrial Select Sector SPDR (XLI)
Technology Select Sector SPDR (XLK)
Consumer Staples Select Sector SPDR (XLP)
Utilities Select Sector SPDR (XLU)
Health Care Select Sector SPDR (XLV)
Consumer Discretionary Select SPDR (XLY)

Here, we update the principal strategy and extend it by adding equal-weighted combinations of the top two and top three sector ETFs, along with corresponding robustness tests and benchmarks. Using monthly dividend-adjusted closing prices for the sector ETFs and SPDR S&P 500 ETF Trust (SPY), 3-month U.S. Treasury bill (T-bill) yield and S&P 500 Index level during December 1998 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

Are there interactions between stock return momentum and days of the week? In their March 2024 paper entitled “Same-Weekday Momentum”, Zhi Da and Xiao Zhang investigate how momentum interacts with days of the week. They first perform regression tests to evaluate abilities of same-day and other-day past returns to predict day-of-the-week momentum. They then evaluate economic significance of findings by comparing three trading strategies:

1. Standard Momentum – each month, reform a value-weighted hedge portfolio that is long (short) stocks that are in the top (bottom) tenth, or decile, of stocks with the highest (lowest) average monthly returns from 12 months ago to one month ago.
2. Same-Weekday Momentum – each weekday during a month, reform a value-weighted hedge portfolio that is long (short) the decile of stocks with the highest (lowest) average daily returns on the same day of the week from 12 months ago to one month ago.
3. Other-Weekday Momentum – each weekday during a month, reform a value-weighted hedge portfolio that is long (short) the decile of stocks with highest (lowest) average daily returns on other weekdays from 12 months ago to one month ago.

Using daily data for publicly listed U.S. stocks, excluding those priced less than $5 and those in the bottom tenth of NYSE market capitalizations, during 1963 through 2021 and daily equity fund/institutional trading data as available, they find that: Keep Reading

A subscriber asked about the stability of the momentum measurement (lookback) interval used for strategies like the Simple Asset Class ETF Momentum Strategy (SACEMS). To investigate, we run two tests on each of top one (Top 1),  equal-weighted top two (EW Top 2) and equal-weighted top three (EW Top 3) versions of SACEMS:

1. Identify the SACEMS lookback interval with the highest gross compound annual growth rate (CAGR) for a sample starting February 2006 when Invesco DB Commodity Index Tracking Fund (DBC) becomes available and ending each of May 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023 and 2024. We consider lookback intervals of one to 12 months, meaning that earliest allocations are for February 2007 to accommodate the longest interval. The shortest sample period is therefore 5.3 years. This test takes the perspective of an investor who devises SACEMS in May 2012 and each year adds 12 months of data and checks whether the optimal lookback interval has changed.
2. Identify the SACEMS lookback interval with the highest gross CAGR for a sample ending May 2021 and starting each of February 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018 and 2019. The shortest sample period is again 5.3 years. This test takes perspectives of different investors who devise SACEMS at the end of February in different years.

Using monthly SACEMS inputs and the SACEMS model as currently specified for February 2006 through May 2024, we find that: Keep Reading