Friday, May 23, 2008

Strategy Lab: Moving average crossover

In this week's strategy lab I will take a look at a simple moving average crossover system; the strategy involves buying on a bullish cross of a 20-day MA by a 10-day MA and selling on the counter bearish cross. I have included data on two sets of stocks; the Active Trader list of U.S. stocks, and a set of European ADRs listed on U.S. markets.


Stocks:
US stocks: AAPL BA C CAT CSCO DIS GM HPQ IBM INTC IP JPM KO MSFT SBUX T WMT
European ADRs: ALU BHP BP SAP DT ASML STM BCS UN TOT ELN AZN DEO RYAAY LUX

Number of shares: 100

Commission: $9.95 (included in the loss calculation)

Trades: Round-trip only; partial trades were excluded.

The strategy was trade heavy with respect to the total number of trades made, making it unsuitable for a regular commission account. The no-stop crossover system generated 893 trades. Adding a stop did not increase substantially the number of trades the system made:

Profit: -$11,164
Winners: 338
Losers: 555
Win percentage: 38%
Profit factor: 0.88

There was a polar opposite return when this strategy (with zero commission) was applied to U.S. and European stocks. Only the 4% stop loss strategy suggested any relationship between the two groups of stocks:


Without further testing of European stocks it is hard to conclude if there is a meaningful difference in trading quality between U.S. and European stocks - but the discrepancy as shown here is not small. One factor contributing to these findings was the lower liquidity of European ADRs compared to the U.S. test stocks. Another is the frequent gaps commonly found in ADRs traded on U.S. markets. Both factors likely contributed to some of the differences seen here. What works for one set of conditions or stock groups may not work for another, and this must always be in the back of your mind when it comes to the practical application of your strategy.

What of a three, randomly selected, 1-year test periods?

The three periods were November 2006/07, February 2004/05, and November 2007/current day. Only U.S. stocks were tested to keep outputs consistent with earlier strategy lab tests.


Because of the lag characteristics of moving averages, entered trades came a few days after reversal points in the market, so using a broader stop prevented whipsaw from the frequent backtests common after a reversal. Using a stop was more profitable than using the strategy as a 'pure system'; i.e, letting crossovers dicate entry and exit signals only.

The win percentage was respectable, around the 42-43% area.

Moving average trading systems perform very well in strong trending markets. The 2006/07 period was very profitable (with the exception of a 3% stop). Non-trending markets tend to hit the system more in commission costs rather than absolute losses. Using longer moving averages would have resulted in fewer trades and therefore fewer transaction costs and would be more suitable for commission based accounts.

Dr. Declan Fallon, Senior Market Technician, Zignals.com the free stock alerts, market alerts, and stock charts website

0 comments: