The purpose of this appendix is to show how Projected Implied Volatility (PIV) can

Transcription

1 Volatility-Based Technical Analysis: Strategies for Trading the Invisible By Kirk Northington Copyright 2009 by Kirk Northington APPENDIX B The PIV Options Advantage Using Projected Implied Volatility to Trade the Butterfl y, Condor, Strangle, and Straddle The purpose of this appendix is to show how Projected Implied Volatility (PIV) can improve on traditional technical analysis methods to find high probability options spread trades. Support and resistance is the classic method for finding option spread trade candidates, and then selecting strike prices. This appendix does not propose to change that. I do believe, however, that the basis of determining support and resistance can be greatly improved when it incorporates volatility. We will use MetaSwing N Bands and S/R lines to define volatility - based support and resistance. A more intensive use of chart - based technical analysis could certainly benefit many private and professional options traders. We will explore here how volatility - based support and resistance should be a primary selection criterion for Butterfly and Condor trade selection. The method will be described in detail, and then back - tested to show its statistical performance. We present also statistical proof that PIV - based support and resistance can accurately forecast a substantial reduction in short - term volatility. Then, in the second part of this appendix, I focus on using N Bands and S/R lines in multiple time frames to best trade corporate earning releases with strangle and straddle option spreads. Also included will be MetaStock exploration code for finding historically high movement stocks within any specific month of the year. PIV FOR THE BUTTERFLY AND CONDOR When searching for good Butterfly and Condor trade candidates, the objective is to select an option - capable security that you believe will trade within a low volatility range for three to four weeks. It is desirable to find a stock that has been on the move, but going 427

2 428 APPENDIX B forward, it will experience a pause in its movement. This previous movement creates greater levels of implied volatility, and thus a better profit - making environment for the trade. In short, everyone else thinks it will continue its trading direction, but you know that it is less likely to do so. A traditional technical analysis method for butterfly and condor trade selections involves identifying a stock that has established a trading range. It would have preferably experienced a minimum of two clear highs and two clear lows, with each high and low pair at the same level. The highs are used to identify resistance and the lows are used to identify support. A trading range is thus formed. This sounds a lot like the formation of a rectangle pattern, doesn t it? You may recall that we explored trading rectangle patterns in Chapter 10. There we stated that the majority of the technical analysis community believes that the only rectangle pattern related price activity that has shown predictability are throwback and pullback retracement patterns, which occur outside of the rectangle; movement inside the rectangle is unpredictable (refer to Figure 10.3 ). This hardly seems desirable for a butterfly or condor. This contradiction of methodology shows how classic technical analysis has not always been helpful to option traders. It would be much more advantageous to identify the trading range prior to its formation. In reality, it is more necessary than advantageous. The technology and activity of programmed trading by large institutions has significantly nullified the predictability of visually recognizable price patterns. The M eta S wing Compression The MetaSwing method for Butterfly and Condor trade candidates is called a Compression. There are four different types of Compression setups. We explore one of them in this appendix. The goal of the Compression method is to identify a narrow zone of extreme resistance and support in a weekly periodicity. The more narrow the zone of support and resistance the better, because the ultimate profitability of the trade will be greater. This zone should increase the probability of price behavior that subsequently results in a temporary trading range for four weeks. Furthermore, this method should produce a trade candidate that is exhibiting sufficient implied volatility levels to make a Butterfly or Condor trade profitable. We see in Figure B.1 a weekly MetaSwing chart of Thomas & Betts Corporation (TNB). Only N Bands and S/R lines are plotted on the chart. Note how price rallies through the first half of 2007 and begins to encounter resistance around S/R 4. This larger view of the price activity shows plenty of movement, and that can be a dangerous environment for a low - volatility option spread trade. Now let s take a closer look at the anatomy of the Compression. Figure B.2 shows another look at the Thomas & Betts weekly chart, which zooms in on the trade setup. The week in which the trigger occurs is denoted by the vertical dashed line. The

3 The PIV Options Advantage 429 FIGURE B.1 This weekly chart of Thomas & Betts (TNB) shows an absence of well - defined trading ranges. Chart created in MetaStock. Chart created in MetaStock. Chart uses the MetaSwing Add - on, by Northington Trading, LLC. All rights reserved. Compression signal appears as a right - facing arrow with the letter C above it. On further inspection, the daily chart (not shown) indicates that price closed above the S/R 3 line on Tuesday, July 3, The criteria that define a Compression setup are listed here. Figure B.2 is notated to identify these criteria. The Compression setup is found on a weekly chart. Price rises up through the S/R 4 line and closes above the S/R 3 line. Price has not closed above the S/R 3 line in the past three weeks. The price high has not crossed above the S/R 1 line in the past three weeks. The distance between S/R 2 and S/R 3 is less than one unit of ATR(40). This means that the distance between S/R 1 and S/R 4 is less than three units of ATR(40).

4 430 APPENDIX B FIGURE B.2 A closer look at Thomas & Betts gives us a detailed view of the Compression setup and characteristics. Chart created in MetaStock. Chart uses the MetaSwing Add - on, by Northington Trading, LLC. All rights reserved. Figure B.2 clearly shows how the distance between S/R 2 and S/R 3 is less than 1 ATR (40). The weeks that follow the Compression signal are labeled one through four, just above S/R 1. It is obvious to see how price is rising aggressively until it hits resistance at the S/R 1 level. The Upper N Band is on a continuous rise in the weeks preceding the Compression signal, and also after the signal. This helps profit potential of the Butterfly trade because implied volatility readings are rising. The effectiveness of the Compression signal is derived from the unusually tight configuration of S/R lines. Once price rises to cross S/R 3, it is well into the center of a compressed area of support and resistance. The result is price movement that pauses at S/R 1 resistance, and often oscillates between that level and S/R 4. At the right of Figure B.2 are the three option strike prices that would be logical choices for a narrow Butterfly trade. During weeks three and four, price goes on to close at the sweet spot for this example, which is the $60 option strike price. This is a teaching example that works out very well, but, of course, not all do. The question any trader

5 The PIV Options Advantage 431 asks at the outset of the trade is, What is my probability of success? That s the topic of the next section. A Study of Compression Performance When we attempt to conduct a study of option trading strategies, a very real problem presents itself. Historical option premium prices are not as readily available as stock prices. Capturing and storing the multiple matrixes of prices, which represent an option chain, for even a single stock, is a daunting task. Historical option premium price data are available, but it is prohibitively expensive for all but very large institutions. Because of this dilemma back - testing an option strategy with results based on actual option premium prices is simply not an option to most of us. This forces us to be more resourceful. Instead of using the option premium price, we will use the underlying stock price movement as a correlation to option profitability. To further ensure a positive correlation, we will overachieve by setting the trading range goal to be typically narrower than the actual option spread strike prices. Compression Testing Method For this backtest, the primary measure of trade success is determined by analyzing how well price remains between the S/R 1 and S/R 4 lines during the period of a typical Butterfly or Condor trade. Through observation, the zone existing between these to price levels is significantly narrower than commonly available option strike prices. Figure B.3 shows a diagram that outlines how we will measure movement, which is, in effect, performance. Since the goal of the trade is for price to remain within a narrow trading zone, the optimum underlying stock price is identified as the Center Line. The Center Line is exactly half way between S/R 1 and S/R 4. Therefore, from the Center Line to S/R 1 is equal to +1 Unit of Movement. Consequently, when the underlying stock price changes by that amount, we state that it has moved a +1 Unit of Movement. If the underlying price decreases from the Center Line to S/R 4, then it has changed 1 Unit of Movement. The actual changes in prices are reported as positive or negative Units of Movement. It is very important to understand that the numbers shown in the S/R 1 S/R 2 S/R 3 S/R 4 Distance = +1 Unit of Movement Center Line Distance = -1 Unit of Movement FIGURE B.3 This is a diagram of Compression movement measurement.

6 432 APPENDIX B TABLE B.1 M eta S wing Compression Performance Results 2.5 Weeks 3.5 Weeks 4.5 Weeks Distribution of Unit Movement Distance from Center Line: Close Between 1 and % 70.6% 64.5% Close > % 18.8% 22.4% Close < 1 6.9% 10.6% 13.0% Arithmetic Mean of Close Median Close Arithmetic Mean Distance from Center Line Geometric Mean Relative from Center Line Standard Deviation from Center Line Trades were conducted on the S & P 500 stock population from July 2000 to June Total Quantity of Trades = 851. (Represents one of four types of MetaSwing Compression methods.) study results are expressed in Units of Movement. Examine Figure B.3 closely before reading the performance results. The testing period is the same as the studies conducted in previous chapters, which is July 2000 through June All trades occurred in the S & P 500 population of stocks. One of the four MetaSwing Compression setup methods produced 851 trades during the test period. Results from all 851 trades were analyzed, thus no random sampling was needed. In the example shown in Figure B.2, the signal day can be any day of the week. If the signal week is considered week zero, then the average trade duration for that week can be assumed to be ½ week. Therefore the close of week two signifies an average trade duration of 2.5 weeks. Compression Test Performance Results The performance results of 851 Compression signal trades is listed in Table B.1. These results are given for three distinct hold periods: 2.5 weeks, 3.5 weeks, and 4.5 weeks. The most common trade duration for a Butterfly or Condor seems to be three to four weeks. Therefore, the middle column showing 3.5 week data is likely the sweet spot for performance. To reiterate, the results are presented mainly to demonstrate the likelihood of price to remain between the S/R 1 and S/R 4 price levels. Consequently, any result of less than one unit of movement achieves that. The top three rows of results in Table B.1 describe the distribution of performance for the three different periods. The top row lists the percentage of trades that close within the S/R 1 to S/R 4 boundary; that is, less than one unit of movement away from the Center Line. As we would expect, each successive week away from the Compression signal shows a higher percentage of closes outside the boundaries. At 3.5 weeks, 70.6 percent represents favorable odds. Figures B.4, B.5, and B.6 show the unit of movement distribution histograms for weeks 2.5, 3.5, and 4.5 respectfully.

7 The PIV Options Advantage 433 Frequency Weeks 80% close between -1 and More Units of Movement from Center Line FIGURE B.4 This shows Compression week performance distribution. Frequency Weeks 71% Close between -1 and More Units of Movement from Center Line FIGURE B.5 This is an illustration of Compression week performance distribution. Frequency Weeks 65% close between -1 and Units of Movement from Center Line FIGURE B.6 This shows Compression week performance distribution. Row four of Table B.1 shows the arithmetic mean of all the movement data. At week 3.5, the average price closes.15 Unit of Movement above the Center Line. The positive progression of data for higher prices at each successive week, for the arithmetic mean and median, suggests that upward price momentum carries it higher after the period of reduced volatility. The progressing right translation of the histograms support the same implication.

8 434 APPENDIX B TABLE B.2 M eta S wing Compression Cumulative Performance Results 2.5 Weeks 3.5 Weeks 4.5 Weeks Cumulative Measure of Unit Movement: Below S/R 1 and Above S/R Above S/R 1 and Below S/R % Unit Distance Below S/R 1 and Above S/R 4 85% 69% 58% Above S/R Below S/R Trades were conducted on the S & P 500 stock population from July 2000 to June Total Quantity of Trades = 851. (Represents one of four types of MetaSwing Compression methods.) The middle section of Table B.1 shows the key descriptive statistics. These values were derived from the absolute value of the movement data. All of these measurements imply that for the majority of trades, price remains within the narrow trading zone. Suppose we measure the raw data in a different way. We know how many were inside, above and below the S/R 1 - to - S/R 4 zone. Another important piece of information would be to know how far above 1 or below 1 the trade failures reached. Knowing this would help us to quantify the severity of the outliers. Likewise, it would be good to know how far inward from the support or resistance level the winners reached. Table B.2 shows the Compression signal s cumulative performance. This serves as another correlation to profitability. In other words, this table shows movement away from the S/R 1 line for values closing above the center line and movement away from the S/R 4 line for values closing below the center line. Unlike Table B.1, it is using the S/R 1 and S/R 4 lines as the points where measurement begins. For example, if a trade closes at +.25, which is above the center line, Table B.2 reports it as a. 75 Unit of Movement away from the S/R 1 line, in the first row. If a trade closes at +1.45, Table B.2 reports it as. 45 away from the S/R 1 line in row 2. What Table B.2 tells us is that the sweet spot is indeed week 3.5. In row 3, we see that 69 percent of the movement away from the theoretical breakeven points is profitable. I say that this is a theoretical breakeven because the S/R boundaries are known. The actual option strike prices should form the real breakeven price points for the trade, but those aren t known in this study. After my own observation of many Compression chart signals, I can state that the functional option strike prices are normally spread much wider than the distance between the S/R 1 and S/R4 lines. Therefore, the typical Compression trade zone represented in these performance figures would be considered very narrow. Also this S/R trade zone is by nature self - adjusting to the underlying stock s volatility level, due to the characteristics of N Bands. N Bands themselves rise and fall with the underlying volatility. Also, the S/R lines are separated by a unit of ATR(40), which is another measurement of volatility.

9 The PIV Options Advantage 435 TABLE B.3 M eta S wing Compression Performance Relative to Random Market Definitions: Unit of Movement = (S/R 1 S/R 4) / 2 Center Line = S/R 4 + Unit of Movement Unit of Movement = Distance from Center Line Mean Unit of Movement: Compression Mean Unit of Movement: Random Market 12.4% of Price 23.6% of Price 2.5 Weeks 3.5 Weeks 4.5 Weeks Mean Unit of Movement: Compression Mean Unit of Movement: Random Market Mean Coe cient of Movement: Compression Mean Coe cient of Movement: Random Market Volatility Reduction of Compression vs Random 35.5% 36.3% 37.0% 1 Unit of Movement Relative to 12.4 percent of Price. 2 Unit of Movement Relative to 23.6 percent of Price. 3 Mean Unit of Movement multiplied by its associated percent of price. Trades were conducted on the S & P 500 stock population from July 2000 to June Compression Relative to Random Another method for determining a trade signal s performance is to compare it to random market movement. In this case, we compared the movement of price during the Compression trade period to the weekly movement that each stock exhibited. To avoid sampling discrepancies, we measured the same 2.5, 3.5, and 4.5 week price movement for each trading week during the test period. This was done for each stock in the S & P 500. Therefore we were measuring the entire population, not a sample of the population. The same unit of movement shown in Figure B.3 was used in all of these calculations. Table B.3 shows the Compression Unit of Movement performance relative to random market movement. However, one difference surfaces during this process. Specifically, we are measuring all market trading weeks, which includes many more weeks when price is not sitting between S/R 1 and S/R4. For most of the weeks, price is sitting at a level far lower than S/R 4. To compensate for this, we measure the mean price percentage between the S/R 1 and S/R 4 lines during a Compression signal, and the same mean for times when price is not in that zone. As you can see in the second section of Table B.3, the S/R zone is a much wider percentage of price during random market days; it is almost double. This makes the random price movement percentage more relevant than the actual price movement expressed in dollars. After adjusting for this unit of movement disparity, each condition is expressed as a mean coefficient of movement. In the middle section of Table B.3 are the results showing the results of decreased volatility subsequent to a Compression signal. At the end of the week trade duration, the Compression method shows a 36.3 percent decrease in price movement, that is, a decrease in short - term volatility, when compared with normal weekly price behavior. This also serves as empirical proof that N Band S/R

10 436 APPENDIX B lines function as price support and resistance, which is to say that volatility - based support and resistance works well. Additional Factors Affecting Compression Performance The study just presented does not consider two normal and important market occurrences. Anyone trading Butterfly or Condor option spreads would avoid holding such positions during a corporate earnings announcement of the underlying stock. The data collected for this study did not filter out hold periods when earnings announcements happened. Since earnings releases generally cause increases in volatility, it is safe to assume that the results were negatively affected by some quantity of those events. Therefore, the Compression performance just stated would likely be improved by implementing the avoidance of corporate earnings releases. Also, when a MetaSwing Correction signal occurs there is a higher likelihood that overall market downside volatility will happen in the short term. This study does not include exiting positions when Correction signals happen. Incorporating the proper response to those events would also be likely to improve the Compression method performance. See Chapter 13 for a complete understanding of the Correction signal and its market breadth ratio calculation. PIV FOR THE STRANGLE AND STRADDLE We explore in this section how PIV supports trading spreads that rely on an increase in volatility. Fundamentally based catalysts create immediate reactions in underlying stock price movements. Different sectors experience information releases that are unique to their specific industry. The most common catalyst is a corporate earnings report. You may recall the Microsoft (MSFT) chart examples in Chapter 4. We focus in this section on another earnings release scenario. If an option trader has reason to believe that the underlying stock price will move in response to an earnings release, she may choose to put on a straddle spread trade. If she thinks that price will make a large move, then a strangle may be the best choice. The more difficult part of the technical analysis is forecasting the extremes that price will reach. This is the task that PIV and N Bands can help you solve. Predicting the market s reaction, however, is not a scientific process because of the subjective nature of the severity of the news release, which is the largest contributing factor. Reaction: Immediate and Projected The N Band represents immediate support or resistance, while the S/R line represents future support and resistance. When a surprise earning release occurs, there is usually an immediate reaction: a price move up or down to a different level. In the days or weeks that follow, price will often continue on to an even more extreme level. Specifically, N Bands tend to identify the limits of the immediate reaction. S/R lines

11 The PIV Options Advantage 437 are usually better at forecasting price movement to its highs or lows, before any significant retracement. The part of the analysis that requires the most judgment is in choosing the proper chart periodicity. As a general rule a daily chart is best for looking at an immediate reaction, and the three - day and weekly charts are best for short - to intermediate - term targets. Figure B.7 shows us a daily and a weekly chart just before an earnings report by Google (GOOG) on April 17, Let s now assume that we are looking at the Google opportunity as a strangle option spread. Our decision for the strangle selection could be based on Google s history of making large moves after earnings release days. This daily and weekly chart combination is what s known to us at the close of April 17th. Your task now is to predict where price will jump to, and ultimately reach, before your particular choice of option expiration date. On the daily chart at left are two small circles. Upon a significant earnings surprise, these are usually where price will jump to just after the earnings release, often in after FIGURE B.7 The daily and weekly charts of Google (GOOG) help us to forecast an immediate and intermediate - term reaction to a scheduled earnings announcement. Chart created in MetaStock. Chart uses the MetaSwing Add - on, by Northington Trading, LLC. All rights reserved.

12 438 APPENDIX B hours or premarket trading. Going forward for a few days, or weeks, the next level of support or resistance is frequently where price will trade to. The weekly chart is usually the best place to find the intermediate - term support and resistance. In this case resistance is S/R 4, which is at Weekly support is at S/R 8, which is These two levels are marked with rectangles on both charts. You may be wondering why I am choosing a support level that is so much closer to price than the resistance level. In this case, that s happening for two reasons. First, support is just plain closer, so as they say, It is what it is. The second reason is that price had undergone a serious decline in recent months. Negative price moves tend to be less severe when a stock has already been recently beaten down. On the daily chart, the strike price target for the call leg of the spread looks quite high. It is important to never underestimate the weekly time frame for projected implied volatility. Based on this analysis, your job is to buy a call and a put further out of the money than most other market participants think price will reach. What will work in your favor is that you know it can reach a given extreme while most others believe it will not. This makes the deep out of the money call and put much less expensive. We can see in Figure B.8 the outcome of the strangle trade. Google does indeed issue an earnings surprise. Price actually gaps up to close just above the upper N Band on the daily chart, landing at the high end of a typical immediate reaction area. Closing above the upper N Band is a very bullish sign; it signifies the market s longer - term interest in owning the stock. Price then continues to rise to just past the weekly S/R 4 line. While this is a teaching example, it is very typical. The single most important factor for a strangle option spread is the strength of the earnings surprise or disappointment. That can be difficult prediction to make. Those that follow fundamental news events closely can develop a sense for getting the probability correct. Studying stock chart histories will help to find underlying issues that have a history of movement. Finding stocks with a tendency toward excessive sudden movement could mean that many of those movement events are probably due to earnings announcements. Mining through large quantities of charts looking for this type of high price movement event can be very time consuming. Another way to approach it is to let your software do the work for you. The following MetaStock code is written to run inside the MetaStock Explorer. It searches through a large stock database and identifies the largest moving issues in a particular month. Here is the code: Column A: Jan %

13 The PIV Options Advantage 439 FIGURE B.8 After an earnings surprise, Google (GOOG) displays price movement that conforms to daily extreme resistance and weekly PIV levels (S/R lines). Chart created in MetaStock. Chart uses the MetaSwing Add - on, by Northington Trading, LLC. All rights reserved. Column B: Jan ATR <> >

14 440 APPENDIX B <> > <> > <> > The code in column A quantifies the largest three - period percentage price change, on a daily periodicity, for the month of January. It then averages this change for the years of 2005 through 2008 to produce a single price change percentage. The code in column B does the same thing, but instead of using a percentage of price, it returns a value in units of ATR(40). This column is probably more useful because the ATR measurement adjusts the measurement for volatility. You can change the code to return results for any month by changing the desired month number in the code. When you run this exploration, be sure to load at least 300 records per year of history desired. PIV FOR OPTION SPREADS The market will always reward options traders that act on knowledge that other traders don t possess. As I write this, volatility - based support and resistance levels are generally known only to larger, resource - rich organizations. We have explored two important applications of PIV in this appendix, but there are many more. Options trading and PIV go together like salt and pepper. The significance of Projected Implied Volatility will consequently only grow. You need not fit the profile of a large institutional investment firm to exploit it, though. If you can trade with a greater ability to accurately forecast support and resistance, then the nonlinear relationships between underlying price movement, risk, and reward found in the option markets will be your ally.