|There are a number of common themes that run through trading. Trading ranges -- high minus low -- are sometimes used with trading strategies built around inside and outside days. An inside day occurs when today's high is less than yesterday's high and today's low is greater than yesterday's low. Similarly, an outside day occurs when today's high is greater than yesterday's high and today's low is less than yesterday's low. |
The conventional wisdom is that inside days can be a precursor to breakouts, which means you would buy if a breakout occurs above the previous day's high and sell if a breakout occurs below the previous day's low. An inside day represents a day with less volatility. It's another characterization of Bollinger band squeezes, which are often precursors to market trend changes. An inside day is like a candlestick harami pattern, which is a reversal pattern.
For outside days, the conventional wisdom is that on the following day you buy in a downswing and sell if it's a rally. An outside day is a characterization of an engulfing pattern, which like the harami, is a candlestick reversal pattern. You need to be aware that the reliability of both harami and engulfing patterns are tied to how well the trend is known. Both are two-day interday patterns.
Most traders who use inside and outside days are by necessity forced to apply trend indicators and momentum indicators. Some use both long and short term.
Next you need to consider if what you see is a constricted range day, for which NR4 (see posting on 8/02/01) tends to be favorable, as traders avoid using three days because of third day reversals. Constricted range days are treated the same as inside days, that is, setups for either upside or downside breakouts.
From the previous posting you know NR4 days are prevalent events. The issue I faced to get a facsimile of reasonable equity performance ($1,500 gain from $1,000 invested) for NR4 was to employ market breadth for QQQ on the downside. I introduced momentum in as innocuous a way as I could by using market momentum. I made the upside work by requiring today's close to be greater than yesterday's close on the trading day following the NR4 event. Some might call that a rally event, although it's a bit of a stretch. Connors and Raschke recognized that NR4s could be the end of a wedge or pennant, which is a continuation pattern, and used a device similar to a Bollinger squeeze to see if a particular NR4 was in fact a precursor to a reversal.
A rally day occurs when today's high is greater than the previous day's high and today's low is greater than or equal to the previous day's low. Similarly, a reaction day occurs when today's high is less than or equal to the previous day's high and today's low is less than the previous day's low.
|In the December 1990 issue of Technical Analysis of Stocks and Commodities, Alan Friedman wrote that Roger Cole devised a method to use rally and reaction days. Cole disregarded inside and outside days to find three rally days in a row or three reactions in a row. This is like three soldiers or three crows in candlestick pattern analysis, except Cole allowed for inside or outside days to intervene. Three soldiers and three crows, while rare, are extremely reliable. So Cole gave us some relief from a strict definition, but traded off increased frequency at the expense of reliability. Cole also used moving average crossovers to augment the two-day (rally, reaction, inside, outside day) results. Again, two days didn't work reliably on their own. |
Cole, in fact, tied volume into his definition of rally and reaction. If used with QQQ it becomes overly restrictive. Notice that inside, outside, narrow range, constricted range, rally and reaction days are all defined without regard to close. They work because the market is always testing the previous day's high and low. They work because pivots work. They work because the market is continually expanding and contracting in volatility. They work because Bollinger squeezes work. NR4 events are like constricted range events and to resolve the ambiguity between what might be a continuation event and a reversal a statistically significant reduction in volatility needs to occur. A Bollinger squeeze is a potential reason for inside days or NR4 days. It is also why inside days will work, but they are ambiguous, as are NR4s, as are harami patterns, as are --- as Gann would tell you.
Jack L. Weinberg ("The Range Indicator," Technical Analysis of Stocks & Commodities, June 1996) devised an indicator that measures the changes in the average day's intraday range (high to low) as compared to the average day's interday range (close to close) as a precursor to change in trend. What would happen if you combined Cole's rally/reaction work with Weinberg's range indicator?
|Figure 1: QQQ Daily Price and Volume (bottom two charts), and Equity Performance (top chart). Long and short entries are indicated by green (long) and red (short) arrows. By requiring all three conditions to happen the number of trading opportunities was purposely held down.|
|Graphic provided by: MetaStock.|
|Graphic provided by: Data vendor: eSignal<.|
|Instead of making $500 trading QQQ with an NR4-based strategy, you make $4,400, but I also added short positions to reach $4,400 because there were very few upside moves missed. |
I reconstructed Weinberg's range indicator for verification. In a previous posting I have covered stochastic RSI, a Chande creation, and came to the conclusion that stochRSI is superior to RSI. In a similar fashion Weinberg calculates a stochastic range. But range is modified to be the ratio of today's true range divided by today's close minus yesterday's close. The power of ATR is that it attempts to give a single measure of today's happenings relative to yesterday's. The Metastock formula for Weinberg's true range indicator is below. I annotated the formula with comments that are enclosed in "[ ]".
q:=Input("q stochastic period",3,20,10);
m:=Input("m exponential smoothing period",1,10,3);
[q is used as the look back period for finding the highest high and lowest low values of the ratio of ATR/(today's close-yesterday's close); m is used to do an exponential smoothing of the stochastic range - Weinberg recommends a default of 10 for q and 3 for m]
[value1 is set to today's true range (ATR) divided by (today's close - yesterday's close) if today's close is greater than yesterday's close, otherwise set value1 equal to today's ATR]
[find the lowest and highest values of value1 over the last q periods]
[determine as a percentage the where value1 fits in the highest high and lowest low value1s]
[smooth stochrange using an exponential moving average of m periods]
The Metastock expert system formulas are below. I wanted a form of Cole's three day rally when Weinberg's range indicator indicated I had a possible trend change event, and then figure out which way momentum is going now compared to what the range indicator was indicating in the recent past. What I saw was that the range indicator could be leading, which is not unusual when using measures of volatility. Another thing I did was logically "AND" all of the conditions that require simultaneous events, with the intention of changing from long to short OCCASIONALLY to avoid being whipsawed. This goes against the grain of having a large number of trades to validate a trading strategy.
For both long and short entries I used a similar approach. I used Cole's notion of three rally days in a row, except I didn't let inside or outside days intervene. I didn't look at closes in calculating the variable "value." I wanted the range indicator to be greater than opt3 for both longs and shorts - you could do better if it had separate values for each. I also wanted to use the same range indicator for long and short entries, so I used opt1 and opt2 in each. What I did want to vary was the look back, opt4 and opt7, for long and short since I was concerned with faster reversals in a bear market. Opt 4 optimized to 10 and opt7 to 12.
I was mostly concerned about my simple-minded momentum indicator of close momentum and let it vary differently between long and short entry. I decided to cheat a little and substituted an optimized momentum indicator for the use of combinations of long term and short term momentum and trend indicators, using a narrow range of periods. I was more interested in seeing the results of the Cole/Weinberg combination, rather than worry about trend and momentum indicators.
bull:=Alert(RangeIndicator(opt1,opt2)>opt3,opt4) AND value>1 AND Mov(C,opt5,E)>Mov(C,opt6,E);
bear:=Alert(RangeIndicator(opt1,opt2) > opt3,opt7) AND value1 > 1 AND Mov(C,opt8,E) < Mov(C,opt9,E);
The values of opt1 through opt8 are 10, 4, 66, 10, 3, 8, 12, 7, and 10. Weinberg recommends 10 and 3 for opt1 and opt2. I let Opt3 vary a little and then just set it to 66 because it was what I would choose reading the chart with the range indicator. I fixed Opt5 at 3.
total winning trades: 11
total losing trades: 5
average win: $352.07
average loss: $121.45
largest win: $1,158.50
largest loss: $224.41
most consecutive wins: 6
most consecutive losses: 1
|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!