|A persuasive argument can be made that a trading strategy that uses no indicators also has the benefit of not having to choose the parameters for the indicators. You can change parameters to fit the situation, but the problem is how do you know if the situation has changed such that you now need to change the parameters? Those who trade using chart price patterns and volume are few but sometimes well known, such as Gary B. Smith at TheStreet.com.|
NR4 is an event. The event is that today's range, high minus close, is the narrowest range when compared to each of the past three days. NR4 is therefore a pattern. Some traders have reported that the markets will often make large moves after this event.
NR4 and an inside day are used in the article "Historical Volatility and Pattern Recognition," by Laurence A. Connors and Linda Bradford Raschke (Technical Analysis of Stocks and Commodities, September 1996) along with historic volatility. An inside day occurs when the high for today is less than yesterday's and the low for today is greater than yesterday's. I implemented the trading system proposed by Connors and Raschke for QQQ, aware that the examples they used were for coffee and heating oil, and had disappointing results. Their volatility measure does appear to be interesting, even if my QQQ experience was not, and is the following:
The six day standard deviation of historical price changes divided by the 100 day standard deviation of historical price changes,
where price changes are calculated by taking the natural logarithm of today's closing price divided by yesterday's closing price, and then yesterday's divided by the day before, repeating until the natural logarithm of six changes are calculated, or repeating for 100 days, and then taking the standard deviation of the last six natural logarithms of closing ratios or the standard deviation of the last 100 natural logarithms. Historical is arrived at by multiplying each by the square root of 260. Then for scaling, multiply by 100.
Taking the ratio of successive days is appealing because it is a ROC (rate of change) approach. Using the natural logarithm of the ratios has the benefit of seeing change in ROC, not with changes in powers of base 10, but of base 2.7183. . . Finally, using a standard deviation is sound way to find an unusual event, especially when comparing it to the standard deviation of longer-term data.
|However with QQQ, NR4 did a reasonable job of identifying points of price change. As I looked at the equity curve I noticed the biggest drawdowns occurred with downtrends. The same problem was evident in many places. While NR4 worked reasonably well for entry for an uptrend I couldn't get out fast enough (or switch to a short) in a downtrend. As much as I tried to finesse NR4 and inside days without resorting to an indicator with parameters to achieve reasonable equity performance I couldn't find a way. So in utter desperation I resorted to using my new highs versus new lows bull/bear indicator to keep me from getting in a downtrending bear market. This is the unimaginative use of the McClellan oscillator except I substitute new highs minus new lows for advances minus declines (Figure 1: second chart form top). For this McClellan like oscillator the system optimization ended up with 14 days for the first moving average and 28 days for the second because I made the second moving average two times the period of the first.|
|Figure 1: QQQ daily price and volume history (bottom charts), true/false condition of two values used in system testing (third chart from top), McClellan like oscillator using new highs new lows (second chart from top) and equity performance (top chart).|
|Graphic provided by: MetaStock.|
|Graphic provided by: Data vendor: eSignal<.|
|The Metastock code for this McClellan-like oscillator is:|
where security is a Metastock function that allows me to retrieve, in the case of nh, the Nasdaq number of new highs from a file called test on my C drive. I used Reuters data for the new highs and lows and DBC for the rest.
I next plotted the two conditions:
today's close > yesterday's close true and yesterday's NR4 true,
yesterday's NR4 true.
The third chart from the top of Figure 1 shows both conditions because I used the artifice of asking Metastock to show me -NR4, so that when NR4 is true (a fourth day range is smaller than the previous three) it charts as minus one and when NR4 is false, it's still zero. What you see is that NR4 events are prevalent from all the times NR4 reached minus one. The top half (Figure1: third chart from the top) charts the condition that NR4 is true yesterday, and today is an up day (today's close >yesterday's close).
When you see an event as prevalent as NR4 you know that you are very likely going to have to use it in conjunction with another event or indicator. The choice that Connors and Raschke made was to use the ratio of standard deviations of closing price ratios. Bollinger squeezes use a similar technique, with the rule being that the 20-day standard deviation of price be 1.5 times less than the six-month standard deviation of price (see Bollinger Bands (Part IV of IV), 3/21/2001).
It's not hard to imagine that if this trading strategy were used for shorts, as well as longs, there would be a considerable number of trades. You can see from the green buy arrows (Figure 1: annotation of price chart at bottom) that NR4 is quick to see any start of a potential uptrend. The problem as stated before is the exit condition or equivalently going short. With my eye I could see the opportunities but encoding NR4 into a trading system is a bit difficult.
When you think about what NR4 is in real life it is likely to be the end of either a bullish or bearish wedge or a pennant. Using the fourth day avoids the three-day reversal noise. Flags are fairly common and unfortunately NR4 is not going to be much help there. Since wedges are continuation patterns but are counter swings to the main trend, it is not surprising that NR4 events are points of decision. But the prevalence of NR4 events tells you it could be none of the above - so there is a significant chance of false signals. I added the condition today's close>yesterday's close the day after an NR4 to eliminate some of the false signals.
To only use NR4 in a bullish environment I added the condition that the momentum of the McClellan-like oscillator of new highs and lows be gaining and be above a threshold using the Metastock formulas
nhnlmom>-opt4 AND Mov(nhnlmom,opt2,E)>Mov(nhnlmom,opt3,E).
Opt4 optimized to 40, while opt2 and opt3 optimized to 4 and 14 respectively. In other words the McClellan-like oscillator should be above -40 and the four-day exponential moving average greater than the 14-day exponential moving average.
Trading QQQ allows the use of Nasdaq new highs and lows. Can you think of what the analogous device would be for say Microsoft?
The result is trading system performance that shows NR4 days can be of benefit:
Annual percentage gain: 21.65%,
Total winning trades = 15,
Total losing trades = 13,
Average win = $58.26,
Average loss = $27.49,
Largest win = $221.85,
Largest loss = $86.68.
|Company:||Technical Analysis, Inc.|
|Address:||4757 California Ave SW|
|Seattle, WA 98116-4499|
|Phone # for sales:||206 938 0570|
|Fax:||206 938 1307|
Traders' Resource Links
|Charting the Stock Market: The Wyckoff Method -- Books|
|Working-Money.com -- Online Trading Services|
|Traders.com Advantage -- Online Trading Services|
|Technical Analysis of Stocks & Commodities -- Publications and Newsletters|
|Working Money, at Working-Money.com -- Publications and Newsletters|
|Traders.com Advantage -- Publications and Newsletters|
|Professional Traders Starter Kit -- Software|
Click here for more information about our publications!