Pure Versus Buffered SMA Crossing Signals
Posted in Technical Trading
May 8, 2009
A reader observed: “One of the ‘problems’ with simple moving average (SMA) crossings is the churning from random price movements across the average. Lars Kestner proposes that an SMA crossing strategy can be improved by modifying the rules so that a BUY signal is generated when: (1) the close crosses over an SMA of the highs (rather than the closes); and, (2) the SMA of the closes is greater today than yesterday. A SELL signal occurs when the close crosses below an SMA of the lows (rather than the closes). These rules create a ‘self-adaptive band’ around the SMA to identify true trends, rather then random noise.” Do these “buffered” SMA crossing signals outperform “pure” crossing signals? To check, we compare the terminal values of contemporaneous pure and buffered signal strategies that buy on 200-day SMA crossovers and sell on 200-day SMA crossunders for both the long-run Dow Jones Industrial Average (DJIA) and its short-run exchange traded fund (ETF) proxy, the DIAMONDS Trust Series 1 (DIA). Using daily highs, lows and closes for DJIA and DIA over the entire periods available, we find that:
The following chart compares the terminal values of $1 initial investments in DJIA using both pure and buffered 200-day SMA crossings over the entire available dataset. We start each series at the first buy signal after the first sell signal, in August 1932 for the pure signal strategy and in March 1933 for the buffered signal strategy. (The mismatch in dates favors the buffered signal strategy in calculating subsequent cumulative returns.) The pure (buffered) signal strategy generates 252 (91) round-trip trades, with the last action being a sell on 5/20/08 (1/2/08). We assume trades are at the close on signal dates and test sensitivity to trading fees.
The chart shows that the pure signal strategy outperforms for round-trip trading fees less than 0.13%, while the buffered signal strategy outperforms for fees above this threshold. A buy-and-hold strategy started in August 1932 with the first pure strategy signal and ending 5/7/09 outperforms both SMA crossing strategies, but cumulative return comparisons are sensitive to start and stop dates. The SMA crossing strategies are likely much less volatile than buy-and-hold.
This analysis is artificial in that DJIA is not tradable, trading fees have changed considerably over the sample period and the impact of dividends is suppressed. Also, as noted, there is a mismatch in initial buy dates for compared strategies. For a more realistic (but shorter) comparison, we examine DIA scenarios.
The next chart compares the terminal values of $1 initial investments in DIA using both pure and buffered 200-day SMA crossings over the available dataset. We use nominal closing values for signal generation, but adjusted closing values for returns to account for dividends. We start both series at the first buy signal after the first sell signal for the buffered strategy, on 10/27/99 for the pure signal strategy and on 10/28/99 for the buffered signal strategy. The pure (buffered) signal strategy generates 56 (30) round-trip trades, with the last action being a sell on 5/20/08 (12/31/07). We assume trades are at the adjusted close on signal dates and test sensitivity to trading fees.
This chart shows that the pure signal strategy outperforms for round-trip trading fees less than 0.30% ($30 on a $10,000 trade), while the buffered signal strategy outperforms for fees above this threshold. A buy-and-hold strategy started on 10/27/99 and ending 5/7/09 outperforms both SMA crossing strategies, but cumulative return comparisons are sensitive to start and stop dates, and starting buy-and-hold on a signaled buy date favors buy-and-hold. The SMA crossing strategies are likely much less volatile than buy-and-hold.
It appears that traders with reasonably low fees and reasonably large trades achieve no long-term benefit from buffering SMA crossing signals as specified. Gains from the “noisy” trades generated by pure signals offset the benefit of reduced transaction fees based on buffered signals.
Both of the above analyses assume zero return on cash while out of the market. Including a risk-free return on cash would very slightly favor the buffered signal strategy. For example, from 10/27/99 through 5/20/08, the pure (buffered) signal strategy for DIA is out of the market 40% (44%) of the time.
In summary, limited evidence does not support a belief that buffered simple moving average crossing signals outperform unbuffered signals.
You May Also Enjoy...
- 10-month Versus 40-week Versus 200-day SMA
- The Washington Crossing Advisors Annual Stock Market Forecasts
It costs less than a single trading commission. Learn more here.
