back start next
[start] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [ 23 ] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]
23 6 Volatility Trading There are some things You leam best in calm, and some in storm. Willa Gather Most human beings are conditioned from childhood that if we can buy something at a cheaper price, then it must be a great deal, and we feel good about the purchase. Since stocks trade in prices, we take the mental leap and assume that a cheaper stock is a bargain, expecting those good feelings in retum for a higher stock price. When the stock continues lower, more shares are purchased because the price is an even better bargain, and the buyer is wondering what these sellers must be thinking. This "buy the dip" strategy worked well during the bull market ofthe 1990s but fell apart in the early 2000s. As with any strategy, the efficacy of a system depends on where and when it is applied, as shown by the bottomfishing example. The Acme V system is selfchecking because it takes a long position only within the context of what it defines as bullish conditions. For example, the system requires the stock to be trading above its 50day moving averagea simple yet effective filter. The Acme V System is the most unorthodox system of all. Even with the moving average filter, it still breaks most ofthe rules because it is a countertrend system for volatile stocks. It is the only Acme system that does not take short positions because it has a few other tricks under its sleeve, and one of those tricks is the socalled Tuesday Turnaround effect [16]. The key to the system is the weekend because Saturday and Sunday do matter. Before the weekend, a given stock or index has established its weekly low and high. If this stock or index finishes the week near its weekly low, then this weakness creates weekend anxiety for the buyers. When trading resumes the following week, the stock will continue its descent, triggering a further selloff in the stock. While the panicked investors bail out ofthe stock, the V system steps in, absent of any news that is causing the decline. This is where the professional traders drink from the pool of liquidity and hunt in the land of volatility, where the trader without a plan is
The V System is a strategj especially suited for options because ofthe extreme voliflitj. Bujing a stock in the VZone is daierous because there may be no apparent reason for the dechne, especially if it is buckii the maiket and sector trend. This strategy works best durii general maiket dechnes and is tailored to the sector indices. When a sector signal is generated, find the bestperfomring stocks in that seaor, and buy a basket ofthem. This sjstem does not work well for industiisl and financial stocks. Because oftheir cjchcal nature and tendency to trend, these stocks tend not to midweek reversals In contrast, the strategy works well for both biotechnolcgy and technologj stocks. 6.1 Linear Regression The basis for file V system IS a statisticsl method known as >¹( ?1 1 20]. Linear regression analyzes past data to project future values using least squcavs fining, computing a fomiula for a line drawn through these data. For a stock chart, the regression line can reference any bar price inthe fomiula open, high, low, or close. Since we are attempting to pick a bottom, we use the low prices, so the regression line is drawn through the lows ofthe data, as shown in Figure 6.1. The rectangle contains the projection ofthe previous four lows to Low 5. We select a regressionlength of five bars because the V sjstem works on the weekly CLAST4)riy 12/1«2001 Mov Aval inefOoeeOJ)) 47.B1S Proceedii to the next bar, we calculate the linear regression for the previous four bars to project a line towards Low 5, highhghted by the rectaie on the chart in Figure 6.2. Now, compare the slope ofthe regression line in Figure 6.1 with the line in Figure 6.2. The former slope is at a steeper downward aie, wlule the latter is more horizontsl. This is the basis ofthe V sjstem. As soon as the slope starts to flatten out, we want to consider along enir. 1\ . I 6.2. Linear Regression Line, Point 2 The easiest way to detect the chaiit slope is to connect the dots for each linearregressionprojection. The lesaitis the linearregivssion showninFigure 6.3. Note how the curve descends and then ticks up at the point where the long entris taken. The astute trader will speculate about how the V sjstem went long when the slope ofthe regression curve was down on the previous bar, given that the value ofthe curve is not known until the end ofthe tradii day. The answer is that the linear regression value is projected one bar into the future, givii us a statistical jump on the other traders (refer to the discussion on resltime trade entrversus endofday entrj in Chapter 5). Instead ofparticipating injust a followthrough day, the traderis able to capitalize on the first day as well. Unlike the other sjstems, the V sjstem enters on a stop order above the previous days close instead of the high, i.e., when the stock goes green plus the ATR percentage. In the following section, we discuss the other conditions that make the V system more r
6  Trading Figure 6,3. Linear Regression Curve 6.2 Volatility Trading System (Acme V) Since V bottoms are tricky, the sjstem has strict requirements for entrj. The sjstem takes entries only on Tuesdays or Wednesdays. In the past several years, Monday has been a relatively bullish day as weh [16], so the trader may wish to change the code to accommodate Mondays. A stock that has not reversed by late Wednesday or Thursday win tend to clcee on the low of the week The second condition is that the stock must be aUyve its 50day mo\Tng a\erage. We are trjing to simulate bullish conditions and to fdter out eh stocks and indices trading below their 50day mo\Tng aerage. Duiii the prolonged huh maiket, we did not need this fdter but learned quickly once the market turned down m the s\mm of2000. The third condition is that the low ofthe current bar is greater than the projected low ofthe regression curve. When a stock is fahit sharply, it tends to outpace the regression curve, i.e., the lows are below the curve. As soon as the projected low is aboe the curve, thenthis conditionis satisfied. The other entry conditions are that the high ofthe current bar must be the lowest high ofthe regression raiige, the current bars range must be less than a given percentage ofthe ATR, and the high ofthe current bar must be greater than yesterdays low (no gap down). 6.2 Volatility Trading Sjstem (Acme V) Calculations 1. Calculate the Average True Range for the past 20 bars (ATRo). 2. Calculate the Moving Average for the past 50 bars ( ) 3. Calculate the Linear Regression value ofthe projected low from the last 5 bars (LRj). 4. Get the Lowest Hig of the previous 4 bars (LH4). 6.2.1 Long Signal I Entry Rules 1. Today is Monday or Tuesday 2. High<LH4 3. Range < RangeFactor * ATR20 4. Low>LR5 5. Low>MA5o 6. High > Yesterdays Low 7. Bi the next bar at or above the Close + (EntiyFactor * ATRo) I Exit Rules: Profit Target 1. Sell half ofthe position at or above the High + (ProfitFactor "ATRjo) 2. Sell half of the position at or above the High of ProfitBars a (2 * ProfitFactor * ATR20) [ Exit Rules: Stop Loss 1. Sell all shares at or below the Lowest Low for StopBars (ExitFactor *ATR2o) 6.22 Short Signal The V Sjstem does not have a corresponding short entrj. The design ofthis strategj is "an exercise left to the reader." Our recommendation is that the short entry be symmetric to the long entry. Use the linear regression ofthe high and a stockbelow its 50dsy moving average. The EasyLanguage code for the Acme V System is shown in Example 6.1. Since the stop order is being triggered above the close, and not above the 1. ,;h as in the othei systems, the ExitFactor may be sethighei
[start] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [ 23 ] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]
