Monday, May 26, 2008

Average True Range, Part2: Stops

This week we are going to be referring to two previous studies:
A Closer Look at Profit Targets
Average True Range, Part1

In the first Average True Range Study, we concluded that ATR multiple breakouts entries did provide an edge over percentage breakouts with regard to expectancy and efficiency. The logic behind that is based on the adaptive nature of ATR breakouts to conform to various markets, time periods, and conditions. This week, I set out to determine if that same logic would apply to stops, thus offering an advantage over the percentage stops tested last week. The entry criteria will be exactly the same as last week’s study and we will also be testing protective stops and profit targets using ATR multiples instead of percentages.

The logic for these stops are as follows:

Protective Stop
ATRStopPrice=EntryPrice-(Average(TrueRange,10,0)*ProtStopX)
ExitLong(ATRStopPrice,Stop,Day)

Pofit Target
ATRProfitTargetPrice=EntryPrice+(Average(TrueRange,10,0)*ProfTgtX)
ExitLong(ATRProfitTargetPrice,Limit,Day)

What is interesting about these is that the stop and target prices are calculated daily based on the previous 10-days ATR and therefore change from day-to-day to adapt themselves to the security’s recent volatility. They expand as volatility increases and contract as volatility dries up.
In order to make these results comparable to last week's, I began by determining what ATR multiples for stops and profit targets gave me similar average losses and average wins. What these preliminary test concluded was that an ATR multiple of 1.0 closely corresponded to 4% of price movement over many, many trades (with equities). Therefore we will step through ATR multiples of .5 (from .5-2.5) for protective stops and 1.0 (from 1-10) for profit targets to closely resemble last week’s study. You will note when making comparisons that the average wins and losses for these are similar. Comparison Study here

As usual, we will run the test on three separate portfolios to increase our confidence that the results are systemic and all tests will be run from 01/01/1998-12/31/2007.

Let’s being by looking at EDR=Expectancy per dollar risked. It is calculated by Average Trade/Average Loss. The results are in dollars with .28 representing 28 cents per dollar risk.





It is never as much fun to write about inconclusive tests but unfortunately that seems to be the case here. Of the three portfolios, ATR stops only outperformed on the Nasdaq portfolio with a maximum EDR of 1.34 vs 1.01 for the percentage based stops. The average EDR for the Nasdaq portfolio was .61 and and .5 respectively. That is a substantial improvement but this points out why multiple portfolios is critical if one is serious about finding robust methods. On the IBD portfolio, the ATR stops had a maximum EDR of .73 and an average of .36 while the percentage stops resulted in a maximum of .81 with an average of .4. Not as large a difference but clearly no edge for the ATR stops with our IBD portfolio. The S&P results were about equal.

Now lets look at a more conventional performance barometer, PF=Profit Factor. This ratio is derived simply by dividing gross profit by gross loss.





The results are very similar here as well with only the Nasdaq portfolio experiencing an advantage of the ATR stops over the percentage based ones.

Finally, for EER=Efficient Expectancy Ratio. This is our EDR/Avg. Days in Winning trades. This number is meant to reveal the most efficient use if capital as it calculates expectancy per dollar risk per day in trade. The results are in dollars with .01 representing 1 cent per dollar risk per day in trade.





Ah ha, our good old EER ratio again presents a unique picture. With regard to trade efficiency, the ATR stops do reveal an edge across all three portfolios with varying degrees of success.
IBD ATR - max:.030 avg:.012
IBD % - max:.024 avg:.010
NAS ATR – max:.043 avg:.021
NAS % - max:.029 avg:.013
S&P ATR - max:.020 avg:.011
S&P % - max:.012 avg:.005

Although we are dealing with small numbers here, the advantage in percentage terms is not inconsequential.

So how do we make sense of these mixed results? I want to make it clear where facts end and speculation begins … here.

It is my belief that ATR stops are effective when using extremely tight stops which is where we see the EER test really excel. The looser the stops, the less a role daily noise plays as more significance is offloaded to vehicle selection, overall trend, and structural relationships. The evidence does seem to suggest that ATR is effective at allowing stops to adapt to and remain clear of daily volatility when deployed properly under appropriate circumstances. Personally, I would strongly consider and test ATR stops further if my system objectives required tight stops or all of my orders were executed mechanically with my brokerage. The success with the Nasdaq portfolio was substantial enough to warrant further testing. Below, I will post the raw data for futher study. Hope everyone here in the U.S. enjoyed their Memorial Day weekend.




2 comments:

Damian said...

Another top notch study. Tradersstudio should be paying you!

I've studied ATR extensively, and I've never found a system, using stocks, that worked really well. They generally don't beat buy and hold - and work, essentially, as a risk reducer ala a moving average. Also like moving averages, churning kills you. Your tests essentially prove what I learned by backtesting.

The same cannot be said of commodities where trending is more normal - ala the Turtles.

Most of the stuff I've looked at is based on the Chandelier buy/sell - so a moving stop hung off of the high.

bhh said...

Thanks, I only wish. Maybe they would be into advertising on our blogs, who knows?

I personally use an ATR breakout for entry into trends with pretty good success under the right market conditions but will be sticking with percentage based stops for now. ROC stops have some potential based on some exploratory studies I've done. I will do more on those in the future.