| Home | My photos | Forex | My trading | Contacts |
|
|
Joe DiNapoli - Trading with DiNapoli Levels | ||||
|
free download links about online stock trading, forex, futures, stock investing, market, trading systems GENERAL DISCUSSION: Like most traders new to the Stochastic, I was befuddled for a time by its movement, and dismayed by my initial attempts to use it. Fortunately the equipment I used at that time to display the Stochastic was the CQG TQ20/20™. I say fortunately, because the TQ used a modified type of Stochastic, rather than what was said to be the standard Lane Stochastic. Some time later I learned there were differences in Stochastic formulas. Had I attempted to apply this indicator using a different Stochastic formula than what was programmed in the TQ20/20™, my learning curve would have been much steeper. This is because the indicator would have been much more difficult to apply and interpret. PROGRAMS, PROGRAMMERS, AND PROBLEMS: At the risk of having your eyes glaze over from boredom or creating a brain embolism from strain, we're going to digress a bit and discuss some significant issues we, as traders, encounter while attempting to utilize trading tools. The Stochastic, and to a lesser extent the MACD, give us a perfect setting for this discussion. We'll consider the Stochastic first. George Lane 1 , the originator of the Stochastics, observed that the closing price within the range of a bar had significance relative to future price development. After considerable diligent effort, he derived a formula that quantified this belief. That seems straight forward and simple enough, but in the real world of trading software it's anything but straight forward and simple. There are disturbing variations of Stochastic formulas floating around and specified in reference material. Even talking with George himself who is one of the most knowledgeable, generous, and gentlemanly figures associated with the business, I have not found a simple means of getting from the original formula to what we as traders are faced with using now. So here's how we'll proceed. I will try not to bury you in complex math. There are a variety of equations in Appendix E for the mathematicians and programmers among you. I will also give you the option to skip ahead a few pages to the Preferred Stochastic, if your only interest is in knowing what I use. By doing so you will eliminate going through a somewhat tortured discussion. However, you will miss out on some of the issues we face utilizing trading software to make our trading decisions. WILL THE RIGHT STOCHASTIC PLEASE STAND UP: When traders purchase a graphics package from let's say, Trade-Em-Quick Software Inc., we can see that they have "The Stochastic" available as one of their preprogrammed indicators. Great, we're happy, since that's one of the indicators that we've read about and want to use. But, which Stochastic is "The Stochastic"? If we don't know enough to ask some very pertinent questions to some hopefully informed and responsible sales people, we haven't any idea of what we are really getting! So, let's try to learn something about Stochastics and let's be worldly about software and how it is developed. LANE (RAW) STOCHASTIC: As far as I'm concerned, all Stochastics can rightfully be called the Lane Stochastic since they all emanated from George Lane's research. Lane Stochastics, i.e. all of the Stochastics we will discuss, have two lines: a fast moving line %K, and a slow moving line %D. There seems to be some agreement between the different Stochastics formulas for %K of the Fast Stochastic, sometimes referred to as the Raw Stochastic, so we'll start with it and I'll include the equation here. FAST STOCHASTICS: If we use the formula in Figure 5-1 for %K and smooth it by using a three period modified Moving Average (MAV), we have the %D line of the Fast Stochastic. In George Lane's Stochastics article 2 , he used an example from the TQ20/20™, which used this type of smoothing of %K to create the % D line. The TQ was also programmed to use the same type of smoothing to create the Slow Stochastic referred to in Figure 5-3. %D (of the Fast Stochastic) = 3 period MAV of %K (of the Fast Stochastic) SLOW ( PREFERRED) STOCHASTICS: Slow (Preferred) Stochastics are derived from Fast Stochastics. If we take the %D computed above and rename it to %K, and then smooth this line by using a three period modified Moving Average, we have the new slow, Slow Stochastic line, %D. These two lines make up the indicator called the Slow Stochastic, created by a modified Moving Average smoothing. This is what I use ("Preferred"). %K (of the Slow Stochastic) = %D (of the Fast Stochastic)) %D (ofthe Slow Stochastic) = 3 period MAV of %K (ofthe Slow Stochastic Some software companies will use other smoothing methods, however, and still call the indicator the Slow Stochastic. The formula for the modified Moving Average is shown below. The starting point (MAV t ) is calculated identically to that of a simple Moving Average. MODIFIED MOVING AVERAGE (MAV): 3 MA V t = MA V t -1 + (P t -MA V t- 1)/n where: MA V t is the current modified Moving Average value MA V t -1 is the previous modified Moving Average value P t is the current price n is the number of periods MODIFIED STOCHASTIC: If we start at the original agreed upon formula for the fast %K (Figure 5-1) and smooth that line by any means whatsoever, we have %K of the modified Stochastic. If we then take that %K line and smooth it by any means whatsoever, and call the result %D, we have the slow line of the modified Stochastic. It is likely you will find other definitions of the Modified Stochastic in reference material or software users manuals. THE STOCHASTIC: Display Trade-Em-Quick Software or Aspen Graphics™ or CIS TRADING PACKAGE or TradeStation® on your screen and there you have it, an indicator that is a Stochastic. What it is or how useful it is, is anyone's guess. Without diligent research, I wouldn't hope to define this term as the study it suggests can have such vastly different appearance, applicability and usefulness, depending on how the equations are manipulated in the software ofyour choice! THE PREFERRED STOCHASTIC: This is a new term. It is meant to reflect what I use and find of benefit. The equations cited above to produce the Slow Stochastic and the modified Moving Average achieve what I want. Other equations and references relating to the Stochastic are in Appendix E so as not to confuse this issue further. The last time I looked, my Preferred Stochastic was called the Slow Stochastic in CQG, Inc., Aspen™, and our own CIS TRADING PACKAGE. It was not available in TradeStation® as a preprogrammed indicator, but could be created by inputting the proper equations in so-called "Easy Language™"(Appendix D). MetaStock™ defaults do not produce the Preferred Stochastic. You can change the defaults in MetaStock™ without inputting equations, to create the Preferred Stochastic. When you study the examples of Stochastics on charts in this book, using the Aspen Graphics™ software, you will see the name Modified Stochastic rather than Slow Stochastic, even though Slow Stochastic is my Preferred Stochastic. Why? When I set up the study initially, I didn't trust that the programmers had computed the Slow Stochastics correctly. I therefore went to the Modified Stochastic study and set up the study myself, to duplicate what I knew were correct inputs. I then compared these values with what I • knew were correct: our own CIS TRADING PACKAGE. After I did that, I compared the specific Modified Stochastic I set up in Aspen with what Aspen called the Slow Stochastic, and found out their programmers did get it right. When we discuss "The Stochastic" in the body ofthis book, it will be my Preferred Stochastic. As trading software evolves, it is likely that the Modified Stochastic will supplant all other forms of Stochastics, since by definition the Modified Stochastic can be adjusted to simulate all others. In that case, to simulate our Preferred Stochastic, the user would input four variables: • eight periods for consideration (eight days, eight hours, etc.) • three periods of smoothing for the fast line • three periods of the smoothing for the slow line • modifiedfor the type ofMoving Average to accomplish the desired smoothing As if this level of detail does not complicate things enough, let me tell you about two other aspects you need to watch out for when selecting software packages and trading with the indicators they produce. MARKET-ALIGNED VS TIME-ALIGNED BARS: It is much easier to program time-aligned bars but they are not as good as market-aligned bars for analysis. Let's look at bonds as an example. Even though the bond market starts trading at 8:20 AM and ends its actual half hour at 8:50 AM, time-aligned bars would begin this market at 8:00 AM, and end the first bar at 8:30 AM. In this example, the first 1/2 hour (8:00-8:30) will have 10 minutes of actual data in it. The second half hour bar will have only 20 minutes of the first half hour's data in it, and 10 minutes of the second half hour's trading data in it. Another example of time-aligned bars producing "erroneous" high, low, last data, would be an hourly S&P. In this case, an hourly S&P's first bar would contain data from 9:00 AM to 10:00 AM although data is not flowing until 9:30 AM! The second hour starts at 10:00 AM and goes until 11:00 AM, instead of correctly starting at 10:30 AM and going until 11:30 AM. With the high, low, last for these intraday charts being recorded "incorrectly," all studies calculated from them are obviously also incorrect. Don't let complacency put you to sleep on this one. Many traders have used studies generated from time-aligned bars for years, with substandard results. Many of these traders are totally unaware of how these studies are being calculated. I suggest to you that the poor performance of indicators may be the result of the improper basis from which they are calculated, rather than the quality of the indicators, or the trader's understanding of their use! THE DATA SAMPLE: The second, and again, not so obvious malady certain graphics packages suffer from, derives from the data sample they select to calculate studies. Let's say you reduce the horizontal (time) axis from 140 to 40 bars. If the studies you are using require in excess of 40 bars to produce accurate values, some vendor programming will be inadequate in that they will calculate only from the sample shown on the screen. Whether you look at 20, 40 or 400 bars on the screen, good graphics programs will give you the same values on the studies. These values should not depend on the number of days shown on the screen assuming, of course, that you have adequate data available on your hard drive to make the calculations accurate. Loose talk about "this Stochastic" or "that Oscillator," without researching the formulas that are used to create them or the programming behind the creation of the bars from which they are calculated, can lead to the most disappointing results, with no hint ofwhere the problem lies! PROGRAMMERS AND UPGRADES: On the subject of programmers, let's talk about what happens in the real world of software development. Let's assume you're the president of a software company and you're also a trader. You have a very stable piece of software you use daily for trading decisions but... it has a teensy, little bug in it. The number 8 in 1998 shows up a little too far to the right of the screen. You go to your programmers and say, "That's kind of annoying; can you fix it?" "Sure!" is the answer. You get the software back two months later and you find out they fixed the number 8, but they also "fixed" a "problem" one of them noticed in the Stochastic equation. Of course, they didn't tell you about this "fix. " I'd like to establish an industry-wide standard to ensure that when programmers do something and don't tell you about it, they lose a toenail. If you think that's harsh, consider a trade you've been planning for weeks, that should have netted you $20,000, but lost $10,000 instead. Why? Because the calculation of an indicator in your trading plan had been changed without your knowledge or approval. You tell me, as a trader, would you personally like to find the pliers, or would you just smile and say, "Please don't do that again?" If I had my way, there would be a lot of limping going on in trading rooms, at least initially! The same goes for upgrades. The software company that has produced your trading software says they have this fantastic new thing, this wonderful study in their upgrade you couldn't give a gaseous burp for. You'reforced to "upgrade," however, since they're no longer going to support that old software version you're working with . You find out later that in the "upgrade," they screwed up the continuous contract generator, created a bug in the cursor window, and your charts won't print correctly any more! When you tell them about this they say, "Don't worry, there's a new upgrade on the way - for only $195.00." Besides undocumented changes, so-called upgrades can wreak havoc with your trading plan in other ways. Often, a default setting on files contained in the upgrade can overwrite settings you may have painstakingly inputted. I'll give you one example which may or may not apply to the software you use. Bid and ask prices are routinely reflected on your quote page. Most traders want this feature. Most traders, however, do not -want bid and asked prices charted. If you select the "bid and asked not charted" option and this is overwritten by your upgrade, it may take months before you realize your charts are wrong! Meanwhile your indicators, D-Levels™, your highs, lows and lasts, will all be incorrect. You may even think you're due a fill by looking at your chart, when there was only a bid or offer beyond that price! If you are new to this game take heed. I've been involved with trading software generation and use for 15 years and these issues are real problems. While I am awed by the talent of programmers, I am equally dismayed by some of their actions and the actions of the managers who direct their efforts. These individuals who may know as much about trading as squirrels do about Fourier analysis, can easily take it upon themselves to "improve" or inadvertently destroy our critical decision-making tools! Notwithstanding the above, were it not for the incredible talent, persistent, dogged work of programmers, I would not have had an opportunity to develop the way I have as a trader. Without programmers, the Displaced Moving Average research, the Oscillator Predictor™ and the striking advantages of my FibNodes™ program would have remained as unfulfilled dreams. So, recognize that the advantages and costs of utilizing computers and human beings to program them, offer awesome benefits, as well as serious challenges. If you keep in mind that software engineers require the same level of strict management and diligent oversight as traders, the benefits should far outweigh the costs. USING THE STOCHASTIC In the early days, I started out with inputs of 14, 3, 3, but later settled on 8, 3, 3. Many of the initial problems I encountered with this study were solved when I internalized the concept of Trend being dependent on Time Frame. There is absolutely no inconsistency with the five minute Stochastic showing a "buy" while the half hour shows a "sell." The inconsistency, if any, is in the user's mind, by not knowing what Time Frame he is trading in, or lacking the experience to handle the speed of variations inherent in very short term, intraday trading. In my early days trading futures, I used only the Stochastic to determine intraday Trend in the traditional manner. When the fast line crossed the slow line from below 25, and broke above 25, it signaled an up trend. When the fast line crossed the slow line from above 75, and broke below 75, it signaled a down trend. See Chart 5-1 |
||
Smarter trading The art of day trading Trading Chaos Sane Investing In An Insane World |
| ©2007 Olesia | Home • My photos • Forex • My trading • Contacts |