Metatrader 5 Overview

Principal Component Analysis (PCA) is the focusing on only the ‘principal components’ among the many dimensions of a data set, such that one is reducing the dimensions of that data set by ignoring the ‘non-principal’ parts.

PCA though, with eigen values & vectors, take on a slightly deeper approach. Typically, data sets that are handled under PCA are in a matrix format and the principal components, that are sought from a matrix would be a single vector column (or row) that is most significant among the other matrix vectors and would suffice as a representative of the entire matrix. As alluded in the intro above, this vector alone would hold the main components of the entire matrix, hence the name PCA. Identifying this vector though does not necessarily have to be done by eigen vectors & values, as Singular Value Decomposition (SVD) and the Power Iteration are other alternatives.

more...

This article continues the series on how the MQL5 wizard is ideal for rapid testing and prototyping ideas for traders. For a lot of people developing expert advisers and trade systems, the need to keep learning and be abreast with trends in not just machine learning but trade & risk management in general is important. We therefore consider within these series how the MQL5 IDE is useful in this regard by not only saving time but also minimizing coding errors.

more...

In this article series, we'll explore how MQL5 indicators are created, customized, and utilized to enhance trading strategies in MetaTrader 5. From basic indicator logic to advanced customization options, we'll cover the basics and gradually move deeper into the more advanced concepts of indicator development as we progress with the article series. The main aim of this article series is to empower you to create your own MQL5 custom indicators tailored to your trading preferences and goals.

more...

We continue the series on MQL5 wizard implementation by looking into Neural Architecture Search while specifically dwelling on the role Eigen Vectors can play in making this process, of expediting network training, more efficient. Neural networks are arguably the fitting of a curve to a set of data in that they help come up with a formulaic expression that, when applied to input data (x), provides a target value (y) just like a quadratic equation does with a curve. The x and y data points though can be, and in fact are often, multidimensional, which is why neural networks have gained a lot of popularity. Nonetheless, the principle of coming up with a formulaic expression does remain, which is why neural networks are simply a means of arriving at this but not the only way of doing so.

more...

In this article, we will consider the multi-population ESG algorithm that I created specifically for this article. We will look at the basic principles of such algorithms. In addition, we will consider the results of comparative studies that will allow us to evaluate the effectiveness of these algorithms in comparison with mono-population optimization methods.

more...

Introducing two innovative portfolio optimization programs designed to revolutionize trading strategies and maximize returns while minimizing risk The first a Python-based solution leverages the power of MetaTrader 5 integration alongside advanced libraries such as pandas Numpy and cvxpy to analyze historical data optimize asset allocation and visualize results with Matplotlib. The second a similar implementation crafted in MQL5 harnesses the native capabilities of the MetaTrader 5 platform offering traders a seamless experience directly within their preferred trading environment. Both programs exemplify the cutting-edge intersection of quantitative finance and technology empowering traders with sophisticated tools to make data-driven decisions in an ever-evolving market landscape.

more...

We revisit and conclude our look at the role different formats of learning rates have on Expert Advisor performance by examining the adaptive learning rates and the one cycle learning rate. The format for this article will follow the approach we had in the last article by having test reports at each learning rate format section rather than at the end of the article.

more...

This article, like all in this series, highlights the use of wizard assembled Expert Advisors in testing out new ideas. Introductions on how this is done can be got from here and here for new readers, with those 2 articles providing some guidance on how to use the code attached at the end of this article. For this piece, we are using quite a few custom enumerations of data as optimizable inputs. The MQL5 inbuilt enumerations can be declared in the custom signal file’s header, and they will be automatically indicated as inputs and initialized as part of the signal filter. When the enumerations are custom though, placing them in the header will prevent the file from being visible (or recognizable) in the MQL5 wizard, meaning you cannot do the wizard assembly. The work around this we have, for now, is omitting them from the custom signal class header but having the parameters and their assignment functions declared within the signal class, as is always the case with any input parameter. Once the wizard assembly is completed, we then make manual changes to the roster of input parameters and also the initialization of the signal class to add this custom enumeration parameters.

more...

MQL5 wizard can be a test bed for a wide variety of ideas, as we have covered so far in these series. And every once in a while, one is presented with a custom signal that has more than one way of being implemented. We looked at this scenario in the 2 articles about learning rates, as well as the last article on batch normalization. Each of those aspects to machine learning presented more than one potential custom signal, as was discussed. The loss , also by virtue of having multiple formats, is in a similar situation.

more...

Regularization is another facet of machine learning algorithms that brings some sensitivity to the performance of neural networks. In the process of a network, there is often a tendency to over assign weighting to some parameters at the expense of others. This ‘biasing’ towards particular parameters (network weights) can come to hinder the network’s performance when testing is performed on out of sample data. This is why regularization was developed.

more...