Bill Rempel, a.k.a. NO DooDahs!

Late last year, I mentioned a desire to work on futures trading as one of my resolutions for 2007. I’m revisiting and updating some of my backtesting work on crude oil futures, and thought it might be worth sharing.

There are five important questions in any system. In this system, the first answer is “the Crude Oil CL contract.”

The second question is all about the setup. Here, it is a triple moving average system with a closing filter, designed for long-term trend following.

I’ll get to the third question later.

The fourth question is about entry. It is signaled by an alignment of the closing price with three exponential moving averages exists, in this case, the 25-, 70-, and 90-day exponential moving averages. If this alignment does not exist on day one, but does exist on day two, then entry is at the open of day three, and is either long or short, depending on the alignment of the close and the moving averages.

The fifth question is about exits, both stop loss and profittaking. Exit is also taken on the morning after a closing signal is given. There are both a hard stop and a trailing stop, both of them are 2 times the 45-day ATR of the contract, with the hard stop based on the entry price and the trailing stop reset nightly by the closing price and ATR of the contract.

This system is evaluated from March 8, 1995 to May 9, 2007, on back-adjusted data (not continuous contract). It has about a 50/50 win ratio, but wins more than twice what it loses, and generates a good expectancy.

The stats here are in terms of dollar-points per contract. The average win of 3.441 points per contract traded equates to $3,441 of actual gain per contract, and the average loss of 1.499 points is $1,499 of actual loss per contract traded. This is big money. It’s made bigger money by the maintenance margin required, if a minimum commitment is made. With a $3,000 per contract maintenance margin and a $60.80 contract price, the contract has 21.5 to 1 leverage. Based on a $4,050 initial maintenance margin, the contract has 16 to 1 leverage. Given these example margin levels, a fully-levered entry (1 contract backed by $4,050 in the account) would allow only a $1,050 loss (or $1.05 on the contract!) before a margin call was triggered and the trader was out of the game!

So now we come back to the third question, how much to buy. In this example, I simulate backing the initial purchase (selling) of 1 contract per $12,000 of capital backing, or approximately 7 to 1 leverage. If money falls below $12,000 per contract, the system continues to trade until it runs out of availability; if money exceeds $24,000 at the time a buy signal occurs, 2 contracts are bought (sold); for $36,000 there would be 3 contracts bought (sold), etc.

The account at the end of 12 and 1/6th years has grown from $12,000 to $623,380, including unrealized gains. This is a 38.4% return. It generates an impressive equity curve.

It does this at an impressive price in terms of volatility, however. Look at this drawdown!

On at least two points, this system has 70% drawdown of equity! As of the ending point, the system is 34% down from peak equity about a year earlier, where annualized returns had approached 48%! This is not a system for the faint of heart … and a key point in trading this may be the addition of uncorrelated strategies traded simultaneously on other contracts.

More on this later.