Objective research to aid investing decisions

Value Investing Strategy (Strategy Overview)

Allocations for February 2020 (Final)

Momentum Investing Strategy (Strategy Overview)

Allocations for February 2020 (Final)
1st ETF 2nd ETF 3rd ETF

Doing Momentum with Style (ETFs)

| | Posted in: Momentum Investing, Size Effect, Value Premium

“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 set momentum ranking (lookback) interval. We focus on the baseline ranking interval from the “Simple Asset Class ETF Momentum Strategy (SACEMS)”, but test sensitivity of findings to ranking intervals ranging from one to 12 months. As benchmarks, we consider an equally weighted and monthly rebalanced combination of all six style ETFs (EW All), and buying and holding S&P Depository Receipts (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 consider 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 December 2019, we find that:

To accommodate the longest lookback interval in the sensitivity test (12 months), we commence rankings in August 2002 and strategy return calculations in September 2002. We start with a 4-month lookback ranking, as used in SACEMS. 

The following chart shows the distribution of Top 1 style ETFs over the available sample period for a 4-month ranking interval. This specification selects small-growth (the most volatile of the six ETFs) most often and large-value (close to the least volatile) least often.

How does this distribution translate into cumulative Top 1 performance?

The next chart compares the gross cumulative values of $100,000 initial investments in Top 1, EW All, SPY, SPY:SMA10 and Top 1:SMA10 over the available sample period. Calculations assume:

  • Reallocate/rebalance at the close on the last trading day of each month (assume that accurate estimates of ranking interval returns for the ETFs are available just before the close).
  • Ignore ETF switching frictions, EW All rebalancing frictions and dividend-reinvestment frictions.
  • Ignore any tax implications of trading.

Top 1 mostly beats SPY, but does not outperform EW All or SPY:SMA10 benchmarks. In other words, results do not support a belief that the momentum strategy adds value to simple style diversification. Compound annual growth rates (CAGR) are 10.5%, 10.4%, 9.7%, 9.6% and 11.4% for Top 1, EW All, SPY, SPY:SMA10 and Top 1:SMA10, respectively. Maximum drawdowns (MaxDD) based on monthly measurements are -54%, -52%, -51%, -12% and -19%, strongly favoring the two strategies with SMA10 market timing.

Regression of Top 1:SMA10 versus SPY:SMA10 monthly returns suggests that outperformance of the former comes partly from alpha (0.08%) and partly from beta (1.10). In other words, it comes partly from the inherently higher return and partly from the higher volatility of the former compared to the latter.

For another perspective, we look at monthly return statistics by rank.

The next chart summarizes average gross monthly returns with one standard deviation variability ranges returns for style ETF ranks 1 through 6 (for a 4-month ranking interval) and the three benchmarks over the available sample period. Top 1 is not exceptional compared to other ranks. Ratios of average monthly return to standard deviation range from 0.17 for rank 6 to 0.21 for ranks 2 and 4 (compared to 0.35 for SPY:SMA10 as the highest).

Is Top 1 performance sensitive to length of ranking interval?

The final chart summarizes Top 1 CAGRs and MaxDDs for ranking intervals of one to 12 months, and for the three benchmarks. Based on available data, shorter ranking intervals work better than longer ones. A 3-month ranking interval is optimal for the sample period, but appears anomalous/likely to be lucky.

The 4-month ranking interval used above is about the middle of the strongest range.

For reference, the following tables report the performance metrics used in “Momentum Strategy (SACEMS)” for style momentum Top 1 (4-month ranking interval).

The first table covers mostly monthly statistics. Rough Sharpe Ratio means average monthly return divided by standard deviation of monthly returns.

The second table covers annual statistics. Annualized means CAGRs. The Sharpe ratio calculation uses average monthly T-bill yield during a year as the risk-free rate for that year.

In summary, evidence from the available sample does not support belief that a simple size-value style ETF momentum strategy enhances an equally weighted benchmark.

See also “Simple Sector ETF Momentum Strategy Update/Extension”.

Cautions regarding findings include:

  • Sample size is modest (about 18 independent measurements for the longest momentum ranking interval and only 22 independent SMA10 intervals).
  • Testing multiple ranking intervals on the same data introduces data snooping bias, such that the best-performing interval (as noted) overstates expectations.
  • The iShares style definitions may not be concentrated enough to exploit style peculiarities identified in some research.
Daily Email Updates
Filter Research
  • Research Categories (select one or more)