Sunday, August 31, 2008

Making the Trend Your Friend: A Remarkably Consistent Trading System

I want to thank Delbert Dunmore of Aurora, IL for this very promising foundation for a trading system. Del, as you might recall, was the subject of behavioral finance research into the "Reverse Midas Effect" before he was hired by a major investment bank to head up their Contrary Opinion Desk.

The key to Del's system is constructing a price channel consisting of a 10-day moving average of daily high prices and an 8-day moving average of daily low prices. When the current day's closing price closes above this channel, buy at close. When the current day's closing price closes below the channel, sell short at close. The trades are closed when SPY closes within its price channel.

Going back three years in the S&P 500 Index (SPY), we find that this short-term trend following system has been remarkably consistent in its profitability. A total of 104 trades were generated, averaging 5 days per trade. The system generated 27 profitable trades and 77 losers, impressively losing about three-quarters of the time. But how about average win/loss per trade? Here is where the system shines: the average loss is 1.40 points (14 S&P futures points), but the average gain is 1.47 points (14.7 S&P futures points).

That's right: the system loses three times as often as it wins, but keeps winners about even in size with losers. Over the three years, the system has lost 67.9 SPY points or a whopping 679 S&P futures points. Hats off to Del and all other short-term trend followers for another trading system that is so bad that it's promising.

P.S. - The system went long on 8/29.


littlemog said...

lol... Dr Brett... this is the most interesting thing I've heard of. (in all week)

Firebird said...

Dr. Steenbarger,

Tsk, tsk, such a negative attitude!

Haven't you read the latest research on the power of positive attitude in trading? It's endorsed by 9 out of 10 trading co(ckro)aches:

Best trading,


John said...

Reversing the system and going back to Jan 2000 gives a return of 14.1% annualized with 14.8% volatility.

Zlatko said...

It seems to me that the usual problem with system development is that the results are normally distributed, not bad. As long as the results are skewed that much in one direction, one can just flip the rules.

A quick test on my software shows that flipping the rules and doing only the long trades ( buy whenever price closes below it's 10-day High SMA and sell when it closes above ) would from 31.08.2005 to 31.08.2008 have given a win percent above 80% and a 38% return (44% if proceeds were reused), not counting commissions.

I didn't bother calculating the short trades as I'm not too familiar with the systems testing part of my software package, so maybe the results are different there.

abel said...

A extremely competent and accomplished sailor I used to sail with, had among his skillful and talented list of attributes, an uncanny sense of irreverence, that to this day I consider among his most endearing qualities.

A dash of irreverence suits you well.

GS751 said...

another great post.

Marc said...

I guess the monkey throwing darts is in jeopardy of being outperformed ;)

Kevin said...

Zlatko: As long as the results are skewed that much in one direction, one can just flip the rules...

Perhaps Dr. Steenbarger already knows this and you've missed the irony of his post?