Hi Wayne, Collaboration is what it is all about! I am working through several candidate approaches to coding channels currently and have been dwelling on Jacks comment regarding landmarks and lookbacks. Very insightful comments there! As you can see, I am only building the "tapes" as of now and I am not applying any analysis or decision logic with them at this point. As Jack has mentioned, it becomes obvious how important channels are (on the course level) for staying in a trade for longer periods and not placing yourself on the wrong side (especially at or near the TL) - Witness the first several sloppy reversals after entry I am using the poor mans version for graphics... GDI+ I am out of town for the weekend but would love to touch base with you to review the LR logic you have. If I may, I will PM you next week and we can discuss further. P.S. Here is todays print. Have a great weekend!
Great stuff ! You can share some of your findings here: Software Used to Trade Jack Hershey Methods http://www.elitetrader.com/vb/showthread.php?s=&threadid=97684
Alright, here's the image of the ATS development based on the USD/JPY forex from 7:00 AM till 5:00 PM on September 4, 2008. This is still in the works (a lot of rules still to add) but makes 695 pips in one day using an variable number of contracts by sweeping the channels with 5 partial fills at each turn. So divide by 5 and it makes 100% of the ATR. Wow. It's easy to see how adding some more rules will set it over 2 to 3 times the ATR. Okay, Jack. Please help with the channeling and ANY other advice you have. Question follow after the image. (I'm trying to have the image actually show here, but can't get it to work. So it's attached.) Most importantly, I need to learn how to handle the BOs (Break outs). Specifically this ATS only has code to handle bounces off the channels. 1. I think one of your documents talks about forecasting the channel after the BO based on the velocity shown during the break out till it reverses. I'm going to go read up again, but any advice helps. That's a big how in this logic. 2. Also, how to avoid trading the wrong direction before a BO. Maybe that involves using the PV (price volume) relationship and maybe the DOM. As you see I have volume indicator even those this is Forex. I developed an algorithm to extrapolate volume from the DOM. 3. How to forecast the reversal point after the break out? My EYES have opened to how you mention planning several trades in advance. FYI: That shocked me at first and trigger me "to good to be true" bone like many others. But today I had an AH HA! and understand it. It's because you also said you're more concerned with timing than price. From a timing perspective, it's a lot easier to approximately forecast the reversal point. Any specifics on that will help. 4. Another area of BOs are squeezes. No squeeze trades are shown because that code is still working poorly. I'm guessing I need to create a "timing" modules that can forecase turning points in these cases. There's a laundry list of other stuff that can improve. NOTE: I do have Marketing Operating points coded. Only a few so far. In programming we call that "state" changes. Sincerely, Wayne
This area on the same chart above is zoomed to the period starting 6:00 am since it has good examples of appropriate reversals at highs and lows but ALSO examples of bad entries that need better rules to handle breakouts. Also, there's a sqeeze or two here where 2 channels overlap. I'm going to read up to understand how to handle these better and see how to turn it into code. Wayne
Yes. The look backs are critical. I have code dealing with a few of them but needs more states in my state engine. Jack calls them market operating points but in software we call that state changes. Example, what to do if long and at a 90% level inside a long trend channel? That depends on whether the last state was simply lower in the channel, previously above the channel, or just coming at of squeeze. So holding or reverse depends on what state you're coming from. In many information systems we use sophistaced workflow (to manage state) and rules engines. I'm only using something simplistic at the moment. Using encapsulation, etc to keep it simply as possible. The LR is working well. I found and fixed a bug where it choked on a perfectly horizontal line when calculation correlation. So it seems to be perfected now. Things maybe you can help me with. I see all show labels on your chart and JAcks' also. I don't know all of them. That might help me understand some more rules to add for handling the issues I asked Jack about earlier. Also, you're using the short term channels rather than intermediate. I have both on my chart but haven't figured out what to do with the short term channels yet. Have you? This ATS only trades off the intermediate channels so far. Thanks, it's cool discussing it. Wayne
eHorn, I was just studying the Channels to Build Wealth. So now I understand what you mean by tapes and why you're doing them. I also see that the LR won't be useful. Basically, I read so far the other day and got overwhelmed so did as he recommended and started "charting" using my software and of course fell in the comfort zone of stuff I already knew about trend lines. It seems the IT channels I created do fit the JH stuff and work well so I'll keep those and expound on them. But I'm throwing out the LR ST "tapes" as useless. Instead I'll create "tapes" similar to how you do it. I now, finally, fully understand what Jack refers to as PACE of the market. Many things Jack describes are FAR easier for a human being to do that to program into ATS. My mind is racing with ideas to figure out how to make a PACE algorithm. I have use some statistics to measure and it does seems possible to forecast when the next reversal will happen. [talking to myself] Let's be realistic, can a computer be programmed to drive a car? And would anyone feel safe riding in it? PACE might be such a fluid human concept that varies enough each day to be close to impossible to automate. It was a big AH HA for me to see the difference between TL and S/R using the ping pong paddle illustration. I used to think everything was an S/R but what he says is 100% and makes total sense. Anyway, discussing what I learn is therapeutic for me so I hope someone enjoys reading. Wayne
Jack, I read your word doc about A/D (it was deleted but I was able to download it before that.) That was very useful. Also, you reiterated the point about multidimensional operating points. You are very kind to always press towards the "real" problem to solve which is rarely the question we students ask at the moment. Therefore, I started a new thread under automated trading specifically dealing with the process of automating your methods. After an introduction, it starts by discussing how to implement the market operating points. Please realize that your expertise, Jack, clearly lies in training yourself and other humans to consistently trade profitably. The drills, sensory input, sweeps, MADA and debriefing all work for a discretionary trader to learn this effectively over time. I used a similar process when I day traded before. I logged every trade with a printed chart, debriefed at the end of each day by writing on trades done according to the plan and an explanation of what went wrong (in my head) on bad trades. It has a Pavlovian effect. And it made me consistently profitable within 2 weeks. So I understand from experience and deeply admire your ability stick with that very demanding task for 50 years. It staggers my imagination how much knowledge you built up during that time. Again, those methods work fabulously to train the human brain. HOWEVER, as you know, this is not a human brain we're trying to teach at this moment. This is a very, very stupid computer. It requires spelling out every detail. So my dilemma is figuring out the process to gradually teach it the market like a human would learn because it's unrealistic to teach it everyone all at once. It need iterative refinement just like you recommend for a human. We can do it rather quickly but it must be iterative. So in the other thread, I mention the need for a "State Machine". That's the programming pattern commonly used is software engineering to solve problems like you describe with the market operating points. State machines are frequently use to build the engines for controlling robots or characters in video games. By the way, I keep re-reading your excellent works on the traderuniverse.info website. Each time it brings a clearer picture. We'll discuss the finite state machine more in the other thread. NOTE: "Tums", a moderator, graciously offered to immediately remove any posts by detractors to your work in that thread since that will be off topic. That topic is focused on experienced traders (especially you) interacting with ATS developers (me to start) to automate what you traders do. It'll be open to other ATS or even traders to contribute. But I'm focusing on automating your methods. Hopefully you'll find it interesting since I have extensive experience in automating a variety of human business processes in similar but different industries--and of course, I understand trading. Most important to me at the moment is your valid point that "every bar matters" and the multi-dimensional states the market has. Please read, my posts there about this, you can skip the introductory material meant for others. Well, let's continue this in the thread called "How to Build an Automated Trading System" in the automated trading forum. Sincerely, Wayne
I was just having a little joke at your expense. Hope you don't mind. What is worrying me is that you seem to have gone from independent thinker originating and testing his own ideas to working on someone elses ideas without questioning them. You come across as a smart person, much smarter than Jack. Your only advisors should be mathematics and logic. Suggestions? Well you might want to consider what you are trying to measure with those channels. Why does price appear to follow channels? Is there a better way?
Interesting. Well so far the things I have learned from JH like about the DOM have been verified and very valuable. However price and channels? I have previously worked hard on automated traded of S/R. But never got it working very well. However, Jack taught something very logical and helpful. He pointed how that the trendline which always to the right of the price action. Right means that if the trend is up then the TL is diagonally beneath it. For a down trend, it is diagonally above. It's much easier to consider both of these at to the Right or price data while thinking symmetrically. Okay, now is the crucial detail. That TL's are rather accurate in predicting the next point at which the market rebounds if it retraces from hi/lo of the trend. However, the S/R is defined by Jack as the opposite side of the trend. Rather inconsistent on how far it will go before rebounding to the trend. His analogy was a ping pong ball and paddle. The TL is the paddle that keeps hitting the back several times and then it bounces back from S/R and gets hit again. That was very helpful and useful. I'm still learning so don't trust me, I may be leaving out or missing some details. What I like about Jack's stuff, especially at traderuniverse.info is that it clearly all comes from PRACTICE and experience of what really makes money hand over fist AND NOT from theory that you hear and read so much about. Ever hear people ask, if that guys strategy he wants to sell works so well to make millions of dollars, why does want to sell it to others? Well it's a valid point. Many of the strategies for sale don't work for a variety of reasons even though they may work for some people some of the time. Jack on the other hand is teach all he knows without any suggestion of getting paid or even saying he'll accept donations. I know another awesome trader who gives out signals and other advice but always includes a request for donations. Ask yourself, why ask people to pay back if your trading generates so much much. Jack only asks you to pay it forward proving he has no need of our money. He has made enough. If people can't GET what he's teaching. Fine. Different strategies work better for different personalities. some could NEVER day trade. Either too stressful or too boring. I suffer from the boring periods when nothing is happening. I can handle the fast moving periods find. That's why I want a trading system to sit through the boredom and make money the rest of the time. NOTE: People criticize Jack for his criptic posts. It's mainly because he uses terminology and acronyms that are over their heads. That was my feeling at first. Well, at traderuniverse.info it was very refreshing. All the acronyms were explained all the terms clearly discussed with diagrams. Oh, it was a heavenly experience. After reading that stuff which altered how I think about trading entirely, I know understand the other posts Jack rights. So the main problem, in my opinion isn't Jack. It's that others don't put the effort to understand what he's saying. He's already going WAY out his way to teach, why should he repetitively spell out the meaning of every term or acronym in every email. Maybe he should put a reference to traderuniverse.info in every email. Wait. He might be disallowed from doing that due to terms of the site. I don't know. Didn't occur to me before. Anyway, I appreciate you recognize I'm not a naive idiot. Thanks. I am a humble and very willing learning. But I VERIFY and TEST everything I learn to ... separate the wheat from the chaff. So far, Jack is all wheat. Wayne