Momentum Investing

Does the January (turn-of-the-year) stock return anomaly affect value and momentum strategies applied at the country stock market level? In his June 2015 paper entitled “The January Seasonality and the Performance of Country-Level Value and Momentum Strategies”, Adam Zaremba investigates this question using four value and two momentum firm/stock metrics. The four value metrics, each measured over four prior quarters with a one-quarter lag and weighted by company according to the methodology of the associated stock index, are:

1. Earnings-to-price ratio (EP).
2. Earnings before interest, taxes, depreciation and amortization (EBITDA)-to-enterprise value (EV) ratio (EBEV).
3. EBITDA-to-price ratio (EBP).
4. Sales-to-EV ratio (SEV).

The two momentum metrics are:

1. Stock index return from 12 months ago to one month ago (LtMom).
2. Stock index return from 12 months ago to six months ago (IntMom).

He assesses strategy performance via returns in U.S. dollars in excess of one-month U.S. Treasury bill yield from hedge portfolios that are each month long (short) the equally weighted fifth of country stock indexes with the highest (lowest) expected returns based on each metric. He first reviews performances for all months and then focuses on turn-of-the-year (December and January) performances. Using monthly data for 78 existing and discontinued country stock market indexes during June 1995 through May 2015, he finds that: Keep Reading

How good can factor investing get? In his May 2016 paper entitled “Quantitative Style Investing”, Mike Dickson examines strategies that:

1. Aggregate return forecasting power of four or six theoretically-motivated stock factors (or characteristics) via monthly multivariate regressions.
2. Use inception-to-date simple averages of regression coefficients, starting after the first 60 months and updating annually, to suppress estimation and sampling error.
3. Create equally weighted portfolios that are long (short) the 50%, 20%, 10%, 4%, 2% or 1% of stocks with the highest (lowest) expected returns.

The six stock characteristics are: (1) market capitalization; (2), book-to-market ratio; (3) gross profit-to-asset ratio; (4) investment (annual total asset growth); (5) last-month return; and, (6) momentum (return from 12 months ago to two months ago). He considers strategies employing all six characteristics (Model 1) or just the first four, slow-moving ones (Model 2). He considers samples with or without microcaps (capitalizations less than the 20% percentile for NYSE stocks). He estimates trading frictions as 1% of the value traded each month in rebalancing to equal weight. Using monthly data for a broad sample of U.S. common stocks during July 1963 through December 2013 (with evaluated returns commencing July 1968), he finds that: Keep Reading

Subscribers have proposed that asset class momentum effects should accelerate (shorter optimal ranking interval) when markets are in turmoil (bear market/high volatility). “Asset Class Momentum Faster During Bear Markets?” addresses this hypothesis in a multi-class, relative momentum environment. Another approach is to evaluate the relationship between time series (intrinsic or absolute) momentum and volatility. Applied to the S&P 500 Index and the S&P 500 Implied Volatility Index (VIX), this alternative offers a longer sample period less dominated by the 2008-2009 equity market crash. Specifically, we examine monthly correlations between S&P 500 Index return over the past 1 to 12 months with next-month return to measure strength of time series momentum (positive correlations) or reversal (negative correlations). We compare correlations by ranked fifth (quintile) of VIX at the end of the past return measurement interval to determine (in-sample) optimal time series momentum measurement intervals for different ranges of VIX. We also test whether: (1) monthly change in VIX affects time series momentum for the S&P 500 Index; and, (2) VIX level affects time series momentum for another asset class (spot gold). Using monthly S&P 500 Index levels and spot gold prices since January 1989 and monthly VIX levels since inception in January 1990, all through April 2016, we find that: Keep Reading

Is there an objective way to benchmark the performance of trend-following Managed Futures hedge funds? In their March 2016 paper entitled “Adaptive Time Series Momentum – Benchmark for Trend-Following Funds”, Peter Erdos and Gert Elaut test a futures timing system that increases (decreases) allocations when trends are emerging (fading) per 251 equally weighted, volatility-scaled, daily rebalanced time series momentum (TSMOM) strategies. Strategy lookback intervals range from 10 to 260 trading days. Volatility scaling involves dividing momentum returns by an exponentially weighted daily moving average estimator of volatility over a 60-day rolling window. They account for trading frictions (bid-ask spread plus broker/market fees by asset class, estimated separately for old and new subperiods), exchange rates, one-day signal-to-trade execution delay and estimated management/performance fees. They apply the TSMOM system as a mechanical benchmark for trend-following Managed Futures hedge funds. They examine also a momentum “speed factor” that buys longer-term and sells shorter-term TSMOM strategies. Using daily prices for 98 futures contract series and monthly net-of-fee returns for 379 live and dead trend-following Managed Futures hedge funds during January 1994 through September 2015, they find that: Keep Reading

Do smart beta indexes efficiently exploit factor premiums? In his April 2016 paper entitled “Factor Investing with Smart Beta Indices”, David Blitz investigates how well smart beta indexes, which deviate from the capitalization-weighted market per mechanical rules, capture corresponding factor portfolios. He consider five factors: value, momentum, low-volatility, profitability and investment. He measures their practically exploitable premiums via returns on long-only value-weighted or equal-weighted portfolios of the 30% of large-capitalization U.S. stocks with the most attractive factor values. He tests six smart beta indexes:

1. Russell 1000 Value.
2. MSCI Value Weighted.
3. MSCI Momentum.
4. S&P Low Volatility.
5. MSCI Quality.
6. MSCI High Dividend.

Using monthly data for the five factor portfolios and the six smart beta indexes as available through December 2015, he finds that: Keep Reading

How should investors think about stock factor investing? In his April 2016 paper entitled “The Siren Song of Factor Timing”, Clifford Asness summarizes his current beliefs on exploiting stock factor premiums. He defines factors as ways to select individual stocks based on such firm/stock variables as market capitalization, value (in many flavors), momentum, carry (yield) and quality. He equates factor, smart beta and style investing. He describes factor timing as attempting to predict and exploit variations in factor premiums. Based on past research on U.S. stocks mostly for the past 50 years, he concludes that: Keep Reading

Is there a most practical way to make value and momentum work together across stocks? In the April 2016 version of their paper entitled “Combining Value and Momentum”, Gregg Fisher,  Ronnie Shah and Sheridan Titman examine long-only stock portfolios that seek exposure to both value and momentum while suppressing trading frictions. They define value as high book-to-market ratio based on book value lagged at least four months. They define momentum as return from 12 months ago to one month ago. They consider two strategies for integrating value and momentum:

1. Each month, choose stocks with the highest simple average value and momentum percentile ranks. They suppress turnover with buy-sell ranges, either 90-70 or 95-65. For example, the 90-70 range adds stocks with ranks higher than 90 not already in the portfolio and sells stocks in the portfolio with ranks less than 70. 
2. After initially forming a value portfolio, each month buy stocks only when both value and momentum are favorable, and sell stocks only when both are unfavorable. This strategy weights value more than momentum, because momentum signals change more quickly than value signals. For this strategy, they each month calculate value and momentum scores for each stock as percentages of aggregate market capitalizations of other stocks with lower or equal value and momentum. They suppress turnover with a 90-70 or 95-65 buy-sell range, but the range applies only to the value score. There is a separate 50 threshold for momentum score, meaning that stocks bought (sold) must have momentum score above (below) 50.

They consider large-capitalization stocks (top 1000) and small-capitalization stocks (the rest) separately, with all portfolios value-weighted. They calculate turnover as the total amount bought or sold each month relative to portfolio size. They consider two levels of round-trip trading frictions based on historical bid-ask spreads and broker fees: high levels (based on 1993-1999 data) are 2.94% for small stocks and 1.06% for large stocks; low levels (based on 2000-2013 data) are 0.82% for small stocks and 0.41% large stocks. They focus on net Sharpe ratio as a performance metric. Using monthly data for a broad sample of U.S. common stocks during January 1974 through December 2013, they find that: Keep Reading

Does adding crash protection based on global market breadth enhance the reliability of dual momentum? In their April 2016 paper entitled “Protective Asset Allocation (PAA): A Simple Momentum-Based Alternative for Term Deposits”, Wouter Keller and Jan Willem Keuning examine a multi-class, dual-momentum portfolio allocation strategy with crash protection based on multi-market breadth. Their principal goal is consistently positive returns, at least 95% (99%) of 1-year rolling returns not below 0% (-5%). Their investment universe is 13 exchange-traded funds (ETF), 12 risky (SPY, QQQ, IWM, VGK, EWJ, EEM, IYR, GSG, GLD, HYG, LQD, TLT) and one safe (IEF). Each month, they:

1. Measure the momentum of each risky ETF as ratio of current price to simple moving average (SMA) of monthly prices over the past 3, 6, 9 or 12 months, minus one.
2. Specify the safe ETF allocation as number of risky assets with non-positive momentum divided by 12 (low crash protection), 9 (medium crash protection) or 6 (high crash protection). For example, if 3 of 12 risky assets have zero or negative momentum, the IEF allocation for high crash protection is 3/6 = 50%.
3. Allocate the balance of the portfolio to the equally weighted 1, 2, 3, 4, 5 or 6 risky assets with the highest positive momentum (reducing the number of risky assets held if not enough have positive momentum).

The interactions of four SMA measurement intervals, three crash protection levels and six risky asset groupings yield 72 combinations. They first identify the optimal combination in-sample during 1971 through 1992 and then test this combination out-of-sample since 1992. Prior to ETF inception dates, they simulate ETF prices based on underlying indexes. They assume constant one-way trading frictions 0.1%, acknowledging that this level may be too low for early years and too high for recent years. They focus on a monthly rebalanced 60% allocation to SPY and 40% allocation to IEF (60/40) as a benchmark. Using monthly simulated/actual ETF total return series during December 1969 through December 2015, they find that: Keep Reading

Does intricate optimization of the relationship between past month-by-month returns and future month-by-month returns substantially outperform a simple stock return momentum strategy based on some fixed past return interval? In their March 2016 paper entitled “Tree-Based Conditional Portfolio Sorts: The Relation between Past and Future Stock Returns”, Benjamin Moritz and Tom Zimmermann apply the machine learning concept of tree-based conditional portfolio sorts to determine which past monthly stock returns provide independent information about future monthly returns. This methodology handles a large number of independent variables, exposes non-linear relationships and emphasizes systematic out-of-sample testing. Their solution (“intricate” momentum) is an average model that smooths potentially anomalous predictions of many specific models, each employing different subsets of predictive variables on different subsamples (to mitigate overfitting). They make intricate momentum adaptive by annually updating the average model based on the last five years of data to determine how each of the monthly returns during the last 24 months predict each of the monthly returns over the next 12 months, generating a total of 45 annual predictions commencing five years after the start of the sample. Their test portfolio takes equally weighted long (short) positions in the tenth of stocks with the highest (lowest) predicted returns during each of these 12 months. Using monthly returns and stock/firm characteristics for a broad sample of U.S. stocks during 1963 through 2013, they find that: Keep Reading

Subscribers have requested extension of the momentum ranking interval robustness test in “Simple Asset Class ETF Momentum Strategy Robustness/Sensitivity Tests” to portfolios other than the momentum winner (Top 1), which each month ranks the following eight asset class exchange-traded funds (ETF), plus cash, on past return and rotates to the strongest class:

PowerShares DB Commodity Index Tracking (DBC)
iShares MSCI Emerging Markets Index (EEM)
iShares MSCI EAFE Index (EFA)
SPDR Gold Shares (GLD)
iShares Russell 1000 Index (IWB)
iShares Russell 2000 Index (IWM)
SPDR Dow Jones REIT (RWR)
iShares Barclays 20+ Year Treasury Bond (TLT)
3-month Treasury bills (Cash)

We consider the following additional five portfolios: equally weighted top two (EW Top 2); equally weighted top three (EW Top 3); loser (Bottom 1); equally weighted bottom two (EW Bottom 2); and, equally weighted bottom three (EW Bottom 3). We consider momentum ranking intervals ranging from one month (1-1) to 12 months (12-1), all with one-month holding intervals (monthly portfolio reformation). The sample starts with the first month for which all ETFs are available (February 2006) and portfolio formation starts with the first month allowed by the longest momentum ranking interval (March 2007). We focus on gross compound annual growth rate (CAGR) and gross maximum drawdown (MaxDD) as key portfolio performance statistics, ignoring monthly reformation costs. Using monthly total returns for the specified assets during February 2006 through February 2016, we find that: Keep Reading