Software products from Piligrimm – DAO
As an example of use of indicators for creation of trading systems I shall result, in the simplified variant, strategy which I realized in trading system on the neural network described in section " the Example of construction of trading system ". After days off I shall start debugging this trading system, and on end I shall expose in this theme as an example of use of indicators. In the written trading system I used some indicators, as an example, I shall result strategy only on two of them, "Kristi_GrafModelWav", " Indicator Trend ". But in the beginning I shall tell some words about principles of construction of these indicators. "Kristi_GrafModelWav" - it is constructed on the basis of polynoms of a following kind:
GR1[i][0] = 0.6*(0.5*(-0.00000808108 +1.64312*(SD[i+5][4]-SD[i][4]) -0.387988*(SD[i][1]-SD[i][2]) +0.598535*(SD[i][2]-SD[i][3])
-0.468099*(SD[i+1][3]-SD[i][2])-0.461584*(SD[i+1][4]-SD[i][1])+SD[i][3])+0.5*((SD[i+2][1]-SD[i][3])-(SD[i+3][1]
-SD[i][4])+0.00000000861016*(SD[i][2]-SD[i][3])+(SD[i+19][3]-SD[i][4])+0.3*SD[i][3]+0.7*SD[i][4]))+0.4*((SD[i][1]
-SD[i+10][3])-(SD[i][1]-SD[i+10][4])+0.00000000861016*(SD[i][2]-SD[i+10][3])+(SD[i][3]-SD[i+10][4])+SD[i][4]);.
Signals SD are received by a principle of " a sliding window " scanning history of quotations on the set depth and changing the width. Training of polynoms was spent on signals of quotations EURUSD M1 received on the basis of history. Signals SD have been passed through group wavelet transformations adjusted on different parameters. Training was conducted on algorithms of linear regress and the formalized neural networks of polynoms led a kind. The purpose of training of polynoms was to strengthen information on entrance signal due to inclusion of late arguments from the last history, and a filtration of minor and stirring factors. One more of the purposes of training of polynoms, - to create group of the signals synthesized on the basis of quotations bearing in affiliated attributes of the basic signal, and its possible changes overlapping a spectrum. It is made with the purpose that neural networks on which input this group of signals will move, had an opportunity "to be hooked" for different harmonics entering in an entrance signal, and to pass qualitative training. Though training was spent on EURUSD M1, the indicator works on any tools, any periods, any markets. The indicator forms group of 40 synthesized signals.
" Indicator Trend " - uses in a basis of the work the algorithm of threshold digitization developed by me with reconstructed step to proportionally dynamics of change of quotations. As a result of such digitization readout corresponding only to extreme points of a signal turn out, there is a compression of a signal, the deep filtration due to noise elimination, insignificant fluctuations is carried out. Restoration of a signal on each bar it is carried out by interpolation between two next points of digitization. On the resulted pictures, a red line, you can see the restored signal on each bar, and points of crisis correspond to points of digitization. By the same principle indicators " Indicator Power " and " Indicator Channel " are constructed also. The similar compression of a signal allows to receive deep study of history of quotations at rather small amount of points of digitization, that essentially reduces time and increases accuracy of training of neural networks at submission by an input of sample of discrete readout.
The essence of strategy of construction of trading system on the basis of two indicators consists in the following. Indicator " Kristi_GrafModelWav " to use as a source of entrance signals, and the indicator " Indicator Trend ", as a source of a signal concerning which training a neural network will be made. The purpose of training - to make the forecast of the future point of digitization that will allow to define a direction of movement of the prices and their provisional level. For synchronization of work of these indicators we shall set identical length of entrance sample LengthSample. In adjustments of indicators we shall establish the sanction of preservation of data to a disk with a direction of indexation from the beginning of a file. In adjustments of the indicator " Indicator Trend " we shall establish a mode of preservation of discrete readout then price levels will be written to a file on a disk in points of digitization (points of crisis in figure) and a serial number of readout from the beginning of sample LengthSample, we shall set a static operating mode of the indicator. Thus on arrival of each new bar all history will be recalculated and to be formed a file covering all the length long samples LengthSample. In a file created by the indicator " Indicator Trend ", we have received readout which we shall use as data rather which the neural network will be trained. Now we should generate a file which we shall submit on an input of a neural network. Indicator " Kristi_GrafModelWav " writes on a disk a file consisting of 40 columns corresponding its target signals, and length LengthSample. We need to choose from this file lines digitizations of the indicator corresponding points " Indicator Trend ".
It is easy for making, having taken advantage of a file received by means of the indicator " Indicator Trend ", its second column contains serial numbers of readout of points of digitization. How length of entrance sample for both these indicators identical, these numbers will correspond also to a file of indicator " Kristi_GrafModelWav ", having chosen lines of a file of indicator " Kristi_GrafModelWav " corresponding these numbers, we shall carry out that digitization of data of indicator " Kristi_GrafModelWav " synchronous to data of the indicator " Indicator Trend ". Thus we have received an entrance file for training a neural network. We spend training a neural network. Now to us it is necessary to spend calculation on the basis of the trained neural network and to give forecasts in a real mode of time. For this purpose in adjustments of indicator " Kristi_GrafModelWav " we shall establish LengthSample = 1. Thus on arrival of each new bar on a disk the file consisting of one line of 40 values of a signal containing all will be written.
We should do calculation by a neural network during the moments of formation of a new point of digitization of the indicator " Indicator Trend ", according to algorithm of training. For definition of these moments it is enough to us to take advantage of two global variables which are formed with the indicator " Indicator Trend ". As soon as value in these two global variables will change rather each other, it means there was a formation of a new point of digitization. During this moment we should consider data file of indicator " Kristi_GrafModelWav " from a disk, submit it on an input of the trained neural network, and to settle an invoice. We shall receive the forecast of a following point of digitization during the moment when last was generated. Here strategy of trading system on the basis of indicators offered by me in the most simple variant is those. It is possible to build and more complex strategy, with use of all indicators, it will raise accuracy and reliability of forecasts.
=============================================================================
The indicator " Indicator Power ".
The idea on the basis of which the indicator " Indicator Power " was developed, consists in dynamically to trace struggle of the bull and bear tendency which occurs in the market in a present situation. At domination of the bull tendency in process of receipt of new bars of value of the indicator on a zero bar to proportionally force of the tendency rise upwards, at domination bear - accordingly downwards. At insignificant fluctuations of the tendency that in one in other party of value of the indicator are reflected in the form of a direct line which on all length from last point of crisis to proportionally force of the dominating tendency slowly is displaced, and at essential increase any of tendencies jump moves to the corresponding party, forming a new point of crisis. " Indicator Power " - uses in a basis of the work the algorithm of threshold digitization developed by me with reconstructed step to proportionally dynamics of change of quotations. As a result of such digitization readout corresponding only to extreme points of a signal turn out. Thus the level of a signal which shows the indicator, is defined by force of the tendency dominating at present and slowly varies proportionally to force of the tendency, the following point of digitization will not be generated yet and will not fix this level.
The indicator " Indicator Channel ".
In the market there are two tendencies, ascending - bull, and descending - bear. Character of these tendencies is a little bit distinguished from each other, but, nevertheless, they are dependent among themselves, and influence one another. In a basis of this indicator the idea is incorporated, to allocate from the general stream of quotations these two tendencies, to formalize their character in the form of time of some, reflecting their development and mutual influence, and also result of their struggle among themselves. The indicator " Indicator Channel " creates the channel of a trend and reflects dynamics of its behaviour, extrapolating a direction of development of the tendency of the market. The top line reflects character of the bull tendency under influence bear, the bottom line reflects character of the bear tendency under influence bull, the average line - result of their struggle, and actually is one of harmonics of a trend of digitization of a stream of quotations corresponding a set level. The different level of a threshold of digitization allows to allocate different on amplitude - to frequency characteristics of a harmonic. " Indicator Channel " - uses in a basis of the work the algorithm of threshold digitization developed by me with reconstructed step to proportionally dynamics of change of quotations. As a result of such digitization readout corresponding only to extreme points of a signal for each of tendencies separately turn out. Thus the level of a signal, each of three lines of the indicator, is defined by dynamics of the corresponding tendency and slowly varies proportionally to force of the tendency, and character of prevalence of one above another, the following point of digitization will not be generated yet and will not fix this level. Informative components of the indicator are: narrowing and expansion of the channel, corners of a divergence of the top and bottom beams, points of an excess and the moments of a turn of the central line of the channel. At receipt of new bars, the indicator dynamically reflects a varying tendency of the market, reconstructing beams from last point of crisis up to a first bar.
================================================================================
On e-mail to me letters with the request act to send a demo of the version of indicators, to look their work. Unfortunately, I do not have demo of versions. For partial indemnification of it to give representation about work of 3 indicators - " Indicator Trend ", " Indicator Power " and " Indicator Channel " in dynamics, I have decided to spend in current of several days demonstration of work in the form of slides removed with the terminal working on a demo. To not overload a theme with figures, I shall show only slides reflect any changes in a market situation and change of indications of indicators. A difference in time of the terminal and a forum 2 hours. The first figure:
The Second slide, I little has enlarged the threshold of the indicator "Indicator Trend", so better, nothing else correct not shall.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Software products from Piligrimm – DAO
_______________________________________
In this theme I wish to begin presentation of programs developed by me for trading. Unfortunately, I do not know English language, therefore is compelled to use the electronic translator, and its translation is not always exact. If who from reading this theme also not so well knows English, for the best understanding it is possible to take advantage of the version of a site in Russian https://www.mql5.com/ru/forum/108505, discussion on this theme there is conducted. Translation can be made, having taken advantage online the translator http://www.translate.ru/ the translator works with 5 languages. With its help it is possible to take part in discussion and to esteem other themes on the Russian version of a site. For the beginning I shall present five indicators, in the further I shall expose other development of indicators, expert and trading systems. Before to pass to the description of indicators, I wish to explain briefly the principles put in a basis of my development. I have started to be engaged in modelling and forecasting of time numbers in 1978. As a basis for the works used the Method of the Group Account of Arguments, it something close to modern genetic algorithms. For 10 years of work in this direction has fulfilled the approaches to the analysis and forecasting of time numbers in conditions strong noise, and the big percent of a casual component in an initial signal, and also synthesis of additional signals on the basis of entrance, with the purpose of reinforcement to information on an entrance signal and improvement of quality of the forecast. In 1988 has left works in this direction and has left in business, was engaged in it till 2000. Since 2000 Forex, and development of programs for its analysis and trade has borrowed in research of the market. As a base package basically I use Matlab, one year ago has started to study MQL4, the truth, till now used it a little, only in a writing of indicators collecting initial data for calculation, and displaying schedules of the terminal, all calculations I spend in Matlab. After long-term studying the market Forex and a plenty of experiments, has come to firm belief that creation of effective predicting expert systems for its analysis and acceptance of trading decisions is quite possible. Most effectively to do it it is possible on the basis of the multifactorial analysis of data by means of neural networks. As entrance data for neural networks, in my opinion, most effectively to use the data synthesized or filtered from a multiple currency signal which have been last preliminary processing for reinforcement to information on an entrance signal. Expert systems should be under construction by a principle self-trained the systems continuously adapting a changing market situation. In this direction I also conduct the development. The additional information on my approach to research of the market and development of programs can be received from a theme on the Russian version of a site: " Effective trading strategy based on the multiple currency analysis of several by dealing centers" https://www.mql5.com/ru/forum/103138, and also from my clause: whether " Can the Forex Market Be Forecasted? How to Create One's Own Trading Strategy?" https://www.mql5.com/en/articles/1477, The Russian version of clause: 'Можно ли прогнозировать рынок Форекс? Как создать собственную торговую стратегию?'. However, I have signed this clause pseudonym DAO. It is connected by that one and a half year ago I for the first time have left on this forum and, having seen themes discussed here, has decided to write clause and to share the experience, hoping, that someone it can appear useful, that discussion will begin, and others will state the vision and understanding of the market Forex. But discussions has not followed, and soon I have ceased to watch this theme. Under name Piligrimm I at a forum asked questions concerning problems connected with mastering MQL4, but was gradually involved and have started to participate in discussion of other themes.
________________________________________
The indicator on wavelet - transformations "Kristi_GrafWav".
Presentation of indicators I wish to begin with indicator "Kristi_GrafWav". Indicator "Kristi_GrafWav" is intended for the multiple currency analysis on three tools. Groups of currencies, a time frames steal up on your discretion. Schedules will be deduced in one window in one scale, it is possible to choose any of 13 schedules who are created both on separate Symbol, and at their combined use. The indicator is intended for use, both for manual trade, and together with advisers for creation of Trading Systems, but developed as the Symbol in a greater degree focused on manual trade, and also on research of the market with the purpose of selection of the most effective group of tools of multiple currency systems supplementing each other at designing. In the first version of the indicator it was used ехе - a file written and compiled in Matlab. I already exposed it on sale, but experience of sales has shown, that many of users do not wish to load on the computer the unknown person ехе - a file with a lot of auxiliary libraries necessary for its work. I have processed the indicator, having transferred all calculation in the indicator written on MQL4, and вейвлет transformations are delivered in the complete set in the form of a source code written on Matlab, and the user himself loads it in Matlab and compiles on the computer, creating necessary for work of the indicator ехе - a file. The complete set of delivery includes the detailed instruction by means of which, even not not having an operational experience with Matlab the user, can easily make all.
The example of work of the indicator can be seen on below resulted figure drawing 1, tools EURUSD, EURGBP, EURJPY are used. Acceptance of trading decisions on the basis of this indicator can be carried out in points of change of a direction or crossing of some alarm lines.
Cost of indicator "Kristi_GrafWav": $85 US Dollar.
____________________________________________
drawing 1
____________________________________________
Indicator "Kristi_GrafModelWav"
Indicator "Kristi_GrafModelWav" I begin presentation of group of 4 indicators supplementing each other, and developed both for manual trade, and for creation of trading systems. Indicators are realized on MQL4 as standard user indicators. Indicator "Kristi_GrafModelWav" is intended for formation of group of 40 signals synthesized on the basis of polynoms, and reinforcement to information on an entrance signal the Symbol on which trade is conducted. Polynoms are trained to model work wavelet transformations, the algorithm of training was set so that to create a spectrum of signals with various a phase - frequency characteristics, but being derivative of the Symbol on which the indicator is established, and bearing in itself its attributes. Such variety of signals allows at their submission on an input neural networks to receive informative enough entrance sample for modelling various processes, and also forecasting of a direction of movement of a trend. And though some of signals are visually similar against each other, polynoms on the basis of which they are received, differ the characteristics and were trained on different algorithms, and these internal differences are perceived neural networks. By development of trading system on the basis of this indicator together with a neural network, I have decided to change parameters of one of two very similar against each other, to make their more excellent, as a result, accuracy of training of a neural network has fallen to the order. I have returned all to an initial kind, including, that accuracy, than visual perception of a variety is better above.
In the indicator there is an opportunity to normalize target signals in a wide range, setting in adjustments factor of normalization. If necessary, having set the corresponding sanction in adjustments, the file with the synthesized signals can be kept on a disk for its further processing by other programs. Training of polynoms was conducted on quotations of currency pairs, but the indicator works and on other tools, not only the market Forex though in other markets I did not spend detailed research of its work. The example of work of the indicator is resulted in figure drawing 2.
Cost of indicator "Kristi_GrafModelWav": $150 US Dollar.
_____________________________________________
drawing 2
_____________________________________________
The Indicator "Indicator Trend"
The Given indicator was developed as a source of a signal for neural networks concerning which training a network is conducted. Its indications can be considered as modelling of one of harmonics of a trend. Principle of work of the indicator similar to a principle of work ZigZag, but unlike it besides local minima and maxima intermediate points of crisis of a trend to respective alteration of intensity of its movement are created. Such decision considerably raises reinforcement to information on a signal of the indicator "Indicator Trend" in comparison with ZigZag, and it essentially affects increase of accuracy of forecasts of neural networks at use of this indicator. Use of this indicator effectively not only together with neural networks, but also is direct together with the adviser, and also for manual trade. Preservation of the values of a signal calculated by the indicator on a disk, at corresponding sanctions in adjustments, is possible in two variants:
1). Preservation of values of a signal on each bar;
2). Preservation of values of a signal only in points of crisis of a trend.
Values of a signal in two last points of crisis are kept also in global variables, names of global variables can be changed, it allows to establish any quantity of indicators with different levels of thresholds on one tool and through global variables to organize communication with the adviser or other indicators. At work of the indicator, probably to optimize its parameters, reconstructing factor of a threshold. The choice of concrete value of a threshold is defined by on what Symbol and Period is established the indicator, and also what strategy of trade you wish to realize on the basis of this indicator. The indicator works in two modes, static and dynamic. In a static mode the indicator builds model of a trend for all length of the set sample of history and recalculates it on all length on arrival of a new bar, creating patterns of the set length which it is possible to show as training sample for a neural network. In a dynamic mode the indicator counts value only on again acted bar, gradually accumulating the calculated values in process of receipt of new data. The example of work of the indicator is resulted on below located figure drawing 3, to the indicator "Indicator Trend" there corresponds a red line of the schedule.
Cost of the indicator " Indicator Trend ": $150 US Dollar.
_____________________________________________
The Indicator "Indicator Power"
The idea on the basis of which the indicator "Indicator Power" was developed, consists in dynamically to trace struggle of the bull and bear tendency which occurs in the market in a present situation. At domination of the bull tendency in process of receipt of new bars of value of the indicator on a zero bar to proportionally force of the tendency rise upwards, at domination bear - accordingly downwards. At insignificant fluctuations of the tendency that in one in other party of value of the indicator are reflected in the form of a direct line which on all length from last point of crisis to proportionally force of the dominating tendency slowly is displaced, and at essential increase any of tendencies jump moves to the corresponding party, forming a new point of crisis. Indications of this indicator give objective representation, that the trend now dominates over the market, or absence the trend. If in process of receipt of new bars the indicator draws a direct line or small on a level and rare steps on time it speaks about a condition absence the trend if transitions from one level on another become frequent and dynamical is speaks about transition to a trend. What to consider as a trend, and that absence the trend, everyone defines for itself(himself) in conformity with the representations about the market, and can adjust in the indicator change of a level of a threshold in adjustments. As well as in the previous indicator, in the indicator "Indicator Power " there is an opportunity to keep the calculated values on a disk on all bars on length of sample LengthSample which is set in adjustments, and values of last two points of crisis - in global variables. At greater levels of a threshold the indicator "Indicator Power", can be used as the indicator of change of phases of the market. By development of trading systems many face a problem, that the trading system on different phases of the market shows frequently opposite results, making a profit or completely merging the deposit. In this connection the question on definition of the moment of change of these phases becomes very actual. The practical operational experience with this indicator has shown, that at correct selection of length of sample and a level of a threshold, it reflects change of a phase of the market in the form of change of a sign between a difference of two values of global variables given out by this indicator. By analogy to the indicator "Indicator Trend", the indicator "Indicator Power" works in two modes, static and dynamic. The example of work of the indicator is resulted on below located figure drawing 3, to the indicator "Indicator Power" there corresponds a pink line of the schedule.
Cost of the indicator "Indicator Power ": $150 US Dollar.
_____________________________________________
The Indicator "Indicator Channel"
The indicator "Indicator Channel" creates the channel of a trend and reflects dynamics of its behaviour, extrapolating a direction of development of the tendency of the market. Informative components of the indicator are: narrowing and expansion of the channel, corners of a divergence of the top and bottom beams, points of an excess and the moments of a turn of the central line of the channel. At receipt of new bars, the indicator dynamically reflects a varying tendency of the market, reconstructing beams from last point of crisis up to a zero bar. As well as in the previous indicator, in the indicator "Indicator Channel" there is an opportunity to keep the calculated values on a disk on all bars on length of sample LengthSample, and values of last two points of crisis for the top, average and bottom line - in global variables. The example of work of the indicator is resulted on below located figure drawing 3, to the indicator "Indicator Channel" there correspond blue and yellow lines of the schedule.
Cost of the indicator "Indicator Channel ": $150 US Dollar.
_____________________________________________
drawing 3
___________________________________________
_____________________________________________
The Example of construction of Trading system
In a greater degree offered indicators were developed for use in trading systems. As an example of use of four indicators: "Kristi_GrafModelWav", "Indicator Trend", "Indicator Power", "Indicator Channel", I have written trading system, expecting to show its work with indicators in a real mode on a demo the account. In trading system, indicators using a different combination two different strategy were applied to management of two warrants. It has been made with the purpose of overlapping of possible losses of one warrant another. The trading system was created for work with EURUSD H1. Unfortunately, the trading system has turned out very braked, made on the average 1.5 trading operations in a week. It was necessary strongly загрублять parameters of system to eliminate false operations. Results of testing of trading system for a floor of year - in the attached archive. Joint testing under two warrants - Strategy Tester_, separate testing under the first warrant - Strategy Tester_1, separate testing under the second warrant - Strategy Tester_2. What optimization of trading system it was not spent. At such slow work of evident demonstration of this trading system it would not turn out.
I have decided to make other trading system for EURUSD M1. Having adjusted work of trading system originally on Sell for one warrant, I have received about the same parameters of trade, as well as for the first trading system. About 50 % of net profit a month with drawdown about 8 % for one trading signal Sell at trade to constants Lots. But feature of this trading system was that on a favorable phase of the market it made a series of trading operations, at approach not a favorable phase of the market it entered into a condition of expectation and could not trade 1 - 2 weeks if it to resolve trade on not to a favorable phase of the market there was a series of unprofitable transactions. On the average it turned out on one trading operation in day, that, considering possible greater breaks in trade, it would be again not so evident for demonstration on a demo the account. In this connection I did not begin to complete this trading system. Debugging of these two trading systems has even more strengthened my representation that the trading system should be self-trained and during work to reconstruct the parameters under a changing situation in the market. I have decided to make one more trading system for demonstration of opportunities of indicators, but already on the basis of a neural network with elements of self-training, and operative tracking of a varying market situation. Yesterday I have written this trading system, but its debugging will demand some time, and, having wasted time on debugging of first two trading systems, I have too tightened preparation for presentation of indicators. Therefore, I have solved, not waiting end of debugging of trading system to lead presentation of indicators and when the trading system will be ready, to show and its work.
____________________________________________
Wishing to buy indicators submit applications for the address: piligrimm_fx@mail.ru