Alternative Sector ETF Momentum Metrics
Posted in Momentum Investing
July 8, 2010
Readers have suggested two alternative metrics for the simple momentum strategy tested in the “Simple Sector ETF Momentum Strategy Performance”: (1) Sharpe Ratio over the past six months, and (2) slope of price over the past six months. Do these metrics outperform past six-month return in a momentum strategy applied to the following nine sector exchange-traded funds (ETF) defined by the Select Sector Standard & Poor’s Depository Receipts (SPDR), all of which have trading data back to December 1998:
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)
The two alternative strategies are: (1) at the end of each month, allocate all funds to the sector ETF with the highest monthly Sharpe Ratio over the past six months (SR6-1); and, (2) at the end of each month, allocate all funds to the sector ETF with the highest monthly price slope over the past six months (Slope6-1). For comparison, we include the strategy of monthly allocation to the sector ETF with the highest total return over the past six months (6-1). Using monthly dividend-adjusted closing prices for the nine sector ETFs and S&P Depository Receipts (SPY) over the period 12/98-6/10 (139 months), we find that:
For the purpose of these tests, we assume the risk-free rate is negligible (or unimportant) so that the SR6-1 signals derive from the mean monthly return divided by the standard deviation of monthly returns over the past six months.
The following chart compares the frequencies of sector ETF winners for the 6-1, SR6-1 and Slope6-1 strategies over the entire sample period. The 6-1 strategy generates the most switches. The SR6-1 strategy tends to spread winners more evenly across ETFs.
How do cumulative returns of the three strategies compare?
The next chart compares the cumulative values of $10,000 initial investments in the three sector ETF momentum strategies and in SPY over the sample period. Calculations derive from the following assumptions:
- Reallocate at the close on the last trading day of each month (assume that we can calculate momentum metrics for the ETFs just before the close).
- Trading (switching) friction is 0.25% of the balance whenever there is a change in holdings.
At the assumed level of switching friction, the original 6-1 strategy beats the two alternative metrics, but the SR6-1 strategy is often close. All three momentum strategies beat buying and holding SPY, but because of U.S. stock market conditions over the past decade, so does the typical random monthly timing strategy over this sample period (see “The 2000s: A Market Timer’s Decade?”).
How do average monthly returns, as alternative measures of strategy performance, compare?
The next chart depicts the average monthly net returns (with 0.25% switching frictions) and the standard deviations of monthly returns for the three sector ETF momentum strategies and for buying and holding SPY over the entire sample period. The three momentum strategies exhibit similar volatilities, with the 6-1 strategy having the highest average monthly return.
Note that iterating multiple tests on a given sample introduces data snooping bias (finds lucky indicators/parameter settings). There may also be incremental “second hand” data snooping bias via prior isolation of an indicator (for example, past six-month return) using a similar sample. In other words, those who first identified and publicized the 6-1 indicator may have tested many variations using similar data, and the bias so impounded carries forward to future tests that use a substantial part of the original development sample (or a highly correlated sample). In other words, the 6-1 strategy may be very lucky in the sample considered (see “Simple Sector ETF Momentum Strategy Robustness/Sensitivity Tests”). Data snooping bias is especially pernicious for small samples.
As a robustness test, we therefore check the persistence of outperformance for the best (6-1) strategy.
The final chart shows the trend in 6-1 strategy abnormal monthly returns (monthly 6-1 net returns minus same-month SPY returns) over the entire sample period. A best-fit trend line slopes downward, suggesting dissipation. In other words, most of the outperformance indicated above for the 6-1 momentum strategy comes from the early part of the sample. The sample period is short (only about 23 independent six-month intervals) for this kind of test.
In summary, evidence from simple tests does not support a belief that alternative measurements of momentum beat past six-month return for a sector ETF momentum strategy.
