Despite the volumetric description of the product, we still receive many fair questions. And rightly so, as it should be, because no matter what description you made, all the nuances cannot be described, especially when it comes to complex product.
With this article I would like to reveal more nuances, give some pieces of advices and talk about the future of the project "PriceTension".
About the concept
No doubt, that the machine cannot compete with people in the breadth of vision and in range of acquired experience, as well as in the ability to think deeply in psychology of studied processes and no one except human beings can see the sense where it's absent. Despite the fact that the market is formed primarily because of the behavior of the human psyche, it still remains the chaos to the human eye. This is quite understandable when you consider that the market behavior is influenced by many (if not by an infinite number of) other factors, that are not subject to human decisions, be sure, the human psychology is a quite unstable thing. So what turns out? Billions of minds, which are effected by everything: a third-party opinion, the clock, the weather outside, the cat sitting next to us, emotions, especially emotions mixing influence the train of thought. What the logic can be here? In result we have chaos with a very high order which has its own logic, with very confused cause and effect. But what is hard to argue about is the causal nature of the market, at least, because our real world obeys the laws of physics and the people are dependent on the real world. And if it's not possible to calculate everything, can we calculate at least something? Obviously, it's necessary to identify deep connections (deep regularities), so deep, that the human brain with its high-level analytically developed thinking, by the way developed very biased, is ineffective. It's not intended for fast low-level analysis (yet?), not developed to such an extent and sometimes we have to take a hand calculator. So you need a fast low-level analyzer. In this sphere the machine is out of competition.
Of course, there are people who can show phenomenal things with the only help of their brain, but what is their number? And how much time and effort will it take for an average person to train the brain to solve such a specific task, without losing an ability to solve vital problems?
The advantages of the machine will always be lack of emotions (which they simply don't need), the speed of low-level analysis for the search of deep connections and, certainly, no fatigue, that ultimately gives a significant performance in the market compared to the man.
The need to configure the product always leads to difficulties on the way to obtaining desired result. Especially, if it's required to do by the end user. Well, it is not a problem if the average user is in the know and has sufficient professional knowledge. But in reality it's not so, and even savvy people, experts in the field, are not always capable of error-free solution of this problem. But, I'm confident, if the machine is designed in such a way to make it possible for it to adjust itself, the set of potential problems, related to the human factor will simply disappear.
"PriceTension" expert was developed to exclude the necessity to adjust or prepare something, or to reduce this factor to a minimum. Therefore, in this regard, the expert is self-sufficient (or almost, but I aspire to it).
In the full version to customize, need money management: to select the type of mm, lot size etc. In the free version, do not need to do, because the expert will always open with the minimum lot size. Only if this is necessary, you need to specify your magic number (can be any, as long as it was unique) or leave standard.
Tips
It is very desirable not to run at the same time with this expert other experts on the same trading instrument, this will eliminate conflicts when trading. If you want, you can run another, but only my expert and mode enabled "LOC" (in all experts).
There is another important point that I didn't notice immediately (one interested person gave me a hint, so I'm really grateful to him). Expert is greedy in terms of resources, so it needs not second-class processor cores. If you decide to launch it on a budget PC, you may get a mild reaction from an expert, because at low frequencies it will show low concentration and it, consequently, will result , in the best case, in poor trade. All this, as it turned out, happened because the expert leveled up its experience on my PC (core frequency: 2.7 GHz) and accustomed to it :) So you will need a processor with at least 2.3 GHz (normal frequency) per one core. The faster is the CPU, the faster the expert will concentrate and this, in turn, directly affect the quality, both of training and trade.
Last update was on August 12th, 2014. Time passed after that and you shouldn't hope that after the first launch a miracle will happen and you will get insane profit. It will require time for "training up", analysis, study of the new (changed) history. To do this, you need to forbid it to trade (disable auto trade) and to allow time for self-study. How much? I will pretend to be an expert and say that at least one week. But to be honest, it needs much more time. The free version has a limiter of concentration level, so you have to cheat, when the concentration level comes to its maximum you should restart the expert, certainly, it's not convenient and it will reduce the concentration rate, nevertheless it will be effective. Full version doesn't require rebooting, because it's not equipped with any limiters. If your want to speed up the process, than for the beginning you should launch the expert in testing mode (at least 5 times), it will significantly speed up the process, however, it can reduce the quality, i.e. the virtual space won't be fully involved.
Graphic panel shows important information: the process and its status. Description of some of the statuses:
- "Developing a plan of trading..." - the EA invents and develops a plan for the new week. The EA only trades on the plan!
- "Trading + Learning" - the EA has a trading plan, now it trades under the plan, and at the same time he is learning in the virtual space.
- "Trading" - the EA has a trading plan, now it trades under the plan. He's not learning in the virtual space.
- "Learning" - the EA does not trade (maybe he doesn't have a trading plan), and only is learning in the virtual space.
Tactics is a component of the strategy. Development of a trading plan is a very important step. Tactical moves for the week are thought out. Its strategic development is reflected in experience, so the more experience the expert has, the more strategically he thinks. Expert tries to develop the plan on weekends (Saturday and Sunday), when the market is closed, so try not to turn off the expert these days, otherwise it will develop plan on weekdays in stress mode (less time for the task will be allocated), thus it will be reflected in the quality of trade not in the best way.
It's worth to say about the experience. With the experience of less than 500 (parrots), the expert is unlikely to perform well, so if you decide run it using trading instrument with zero experience, be sensible and do not expect that it will quickly figure everything out and will start trading adequately, you will have to wait a lot. In this case you have to be realist and to understand that even people with their incredibly strong intellect takes time, very, very many time for analysis and interpretation of causal relationship, it's not difficult to learn to trade quickly, it is complicated to learn to trade profitably and particularly stable. And even if the expert gains decent experience, you have no guarantee that the market will present no gift, be always ready to unexpected things. Integrated memory was prepared with my assistance, so don't think that the clean memory will be leveled as effectively as integrated one, more precisely it can do it even more effectively but the probability is quite low. In any case, his experience is the real indicator.
About development
Due to the catastrophic lack of time, the development of this project was suspended, i.e. because of its complexity expert needs a lot of time and attention. I didn't have an aim to forget about it. I simply had other tasks, and experienced lack of resources, other projects also demanded attention. In part, because of this, in the new year, I lowered the price from 700 to 400 and set the goal to start developing a cool mod for it in spring (yes, I'm already doing this). Naturally, the price will rise after the update.
During all this time I invented a lot of ideas for its updating. The following major ones I will try to gradually implement:
- Trading strategy. Leveling, mainly, will change its trading strategy (it learns to control it), because it's weak, and came from distant times when it was just a low-budget expert for 40 bucks, without any frills. It turns out that now it has a weak engine and cool electronics. I want to make it amorphous, modularly extendable.
- Computing resources. I will seek to minimize the dependence on processor speed by avoiding the time-dependence. At least, it will be obvious what its concentration aspires to, and how much to expect. I will try to add a manual concentration speed settings in the panel and settings, or rather its CPU usage.
- Universal memory. It would be cool. This will solve a lot of problems with trading tools. Now without integrated memory (the one with initially upgraded experience) expert with experience of less than 500 is almost unable to show sane results. It's practically impossible for me to create integrated memory for each trading instrument. And users, with high probability, will run on the entire spectrum of various characters, memory for which is completely absent. It's not good to start with zero experience. This feature will be implemented mainly for the full version.
- Intelligence increasing. For example, a self-assessment of the environment: if the expert has not been used for a long time, then when being launched it will upgrade his experience to the acceptable level, to fill the gap of new information and thereby reduce its incompetence, because of the fact that now it's not in the know (not used to) of the market nature changes, so it will reduce the risk of erroneous trading decisions as well as inadequate behavior to the minimum and the user will not have to scratch his head, see year-old date of the last expert update.
- Assimilation. This means that if your expert gained additional experience (over integrated or not for integrated trading instrument), then running a new version, the expert will assimilate the old memory with a new, so old experience won't be lost.
- News. I have not yet decide to make a deep analysis of news, but I think, that it's good to allow user to choose whether to trade or not before the release of important news. In any case, I think it will be necessary to focus on the subject.
- Smart MM + RM (money management + risk management). Now the expert is limited in trading decisions only by stop orders and the side of opening. In future, I'll add probabilistic assessment of trading decisions that will affect the size of the opened lot. After all, if the expert is 100% sure in its trading decisions, then why not to allow it to open with a bigger lot, right?
- Mode "Adviser". In this mode, the expert will not trade, it will only give advice on trading with the help of graphic objects (or by mail, or other notifications): when, where and for how much to enter the market. I really like this idea, as it will also make available clever semi-automatic system and create a comprehensive tool.
- GPU acceleration. Architecture of expert is not trivial and it is extremely difficult to parallelize. Nevertheless, I'll address this situation. In the future I will try to use the GPU using OpenCL, which should speed up the calculations.
It will take time, but I'm sure it's worth it and I'm extremely curious to see how it will eventually work.
Prospective
In fact, from the very beginning, I launched this project as a long-term one, in terms of development. The first version was the most primitive, it was a simple algorithm, moreover, using a harmful method of averaging, respectively the cost was minimal. Then I expanded functionality. But finally, in early 2014, I completed the implementation of version 4.0, it became real evolution of the expert, having created an experiment with an attempt to answer the question whether the strong tying of a weak algorithm (trading strategy) can give a stretch to the hole system in terms of prospects. Then, I continued developing the tying system. But since for now the answer to the question is exhausted, the engine of the expert (trading strategy) will also undergo a facelift to become amorphous, flexible, and modularly extensible.
By this, I also want to note that this product is worth more to consider as an investment, based on the new and improved versions. Keep this in mind making a decision.
