Articles on machine learning in trading

In the previous article, we introduced the DDPG method, which allows training models in a continuous action space. However, like other Q-learning methods, DDPG is prone to overestimating Q-function values. This problem often results in training an agent with a suboptimal strategy. In this article, we will look at some approaches to overcome the mentioned issue.

Did you know that we can gain more accuracy forecasting certain technical indicators than predicting the underlying price of a traded symbol? Join us to explore how to leverage this insight for better trading strategies.

When working with machine learning models, it’s essential to ensure consistency in the data used for training, validation, and testing. In this article, we will create our own version of the Pandas library in MQL5 to ensure a unified approach for handling machine learning data, for ensuring the same data is applied inside and outside MQL5, where most of the training occurs.

In this article, we will create an arbitration system that remains legal in the eyes of brokers, creates thousands of synthetic prices on the Forex market, analyzes them, and successfully trades for profit.

In the previous article, we discussed a simple RNN which despite its inability to understand long-term dependencies in the data, was able to make a profitable strategy. In this article, we are discussing both the Long-Short Term Memory(LSTM) and the Gated Recurrent Unit(GRU). These two were introduced to overcome the shortcomings of a simple RNN and to outsmart it.

The article considers the theoretical application of quantization in the construction of tree models and showcases the implemented quantization methods in CatBoost. No complex mathematical equations are used.

We continue studying distributed Q-learning algorithms. In previous articles, we have considered distributed and quantile Q-learning algorithms. In the first algorithm, we trained the probabilities of given ranges of values. In the second algorithm, we trained ranges with a given probability. In both of them, we used a priori knowledge of one distribution and trained another one. In this article, we will consider an algorithm which allows the model to train for both distributions.

The authors of the FreDF method experimentally confirmed the advantage of combined forecasting in the frequency and time domains. However, the use of the weight hyperparameter is not optimal for non-stationary time series. In this article, we will get acquainted with the method of adaptive combination of forecasts in frequency and time domains.

In this article, I will consider the Bat Algorithm (BA), which shows good convergence on smooth functions.

The article explores the development of a quantum-inspired trading system, transitioning from a Python prototype to an MQL5 implementation for real-world trading. The system uses quantum computing principles like superposition and entanglement to analyze market states, though it runs on classical computers using quantum simulators. Key features include a three-qubit system for analyzing eight market states simultaneously, 24-hour lookback periods, and seven technical indicators for market analysis. While the accuracy rates might seem modest, they provide a significant edge when combined with proper risk management strategies.

Long Short-Term Memory (LSTM) is a type of recurrent neural network (RNN) designed to model sequential data by effectively capturing long-term dependencies and addressing the vanishing gradient problem. In this article, we will explore how to utilize LSTM to predict future trends, enhancing the performance of trend-following strategies. The article will cover the introduction of key concepts and the motivation behind development, fetching data from MetaTrader 5, using that data to train the model in Python, integrating the machine learning model into MQL5, and reflecting on the results and future aspirations based on statistical backtesting.

Explore the world of MQL5 arrays in Part 5, designed for absolute beginners. Simplifying complex coding concepts, this article focuses on clarity and inclusivity. Join our community of learners, where questions are embraced, and knowledge is shared!

Since the first articles devoted to reinforcement learning, we have in one way or another touched upon 2 problems: exploring the environment and determining the reward function. Recent articles have been devoted to the problem of exploration in offline learning. In this article, I would like to introduce you to an algorithm whose authors completely eliminated the reward function.

These advanced gradient-boosted decision tree techniques offer superior performance and flexibility, making them ideal for financial modeling and algorithmic trading. Learn how to leverage these tools to optimize your trading strategies, improve predictive accuracy, and gain a competitive edge in the financial markets.

Forecasting plays an important role in time series analysis. In the new article, we will talk about the benefits of time series patching.

The last two articles were devoted to the Decision Transformer method, which models action sequences in the context of an autoregressive model of desired rewards. In this article, we will look at another optimization algorithm for this method.

This project explores the fusion of deep learning and technical analysis to test trading strategies in forex. A Python script is used for rapid experimentation, employing an ONNX model alongside traditional indicators like PSAR, SMA, and RSI to predict EUR/USD movements. A MetaTrader 5 script then brings this strategy into a live environment, using historical data and technical analysis to make informed trading decisions. The backtesting results indicate a cautious yet consistent approach, with a focus on risk management and steady growth rather than aggressive profit-seeking.

The article implements a fast strategy tester for machine learning models using Numba. It is 50 times faster than the pure Python strategy tester. The author recommends using this library to speed up mathematical calculations, especially the ones involving loops.

In this series article, we will empirically analyze classic trading strategies to see if we can improve them using AI. In today's discussion, we tried to predict higher highs and lower lows using the Linear Discriminant Analysis model.

Perceptrons, single hidden layer networks, can be a good segue for anyone familiar with basic automated trading and is looking to dip into neural networks. We take a step by step look at how this could be realized in a signal class assembly that is part of the MQL5 Wizard classes for expert advisors.

Do the positions of planets and stars affect financial markets? Let's arm ourselves with statistics and big data, and embark on an exciting journey into the world where stars and stock charts intersect.

We continue studying the environment in reinforcement learning models. And in this article we will look at another algorithm – Go-Explore, which allows you to effectively explore the environment at the model training stage.

We continue our acquaintance with the TEMPO method. In this article we will evaluate the actual effectiveness of the proposed approaches on real historical data.

This is a unique optimization algorithm inspired by the evolution of the turtle shell. The TSEA algorithm emulates the gradual formation of keratinized skin areas, which represent optimal solutions to a problem. The best solutions become "harder" and are located closer to the outer surface, while the less successful solutions remain "softer" and are located inside. The algorithm uses clustering of solutions by quality and distance, allowing to preserve less successful options and providing flexibility and adaptability.

In this article, we will have a look at yet another reinforcement learning approach. It is called goal-conditioned reinforcement learning (GCRL). In this approach, an agent is trained to achieve different goals in specific scenarios.

Category Theory is a diverse and expanding branch of Mathematics which is only recently getting some coverage in the MQL5 community. These series of articles look to explore and examine some of its concepts & axioms with the overall goal of establishing an open library that provides insight while also hopefully furthering the use of this remarkable field in Traders' strategy development.

The last two articles considered the Soft Actor-Critic algorithm, which incorporates entropy regularization into the reward function. This approach balances environmental exploration and model exploitation, but it is only applicable to stochastic models. The current article proposes an alternative approach that is applicable to both stochastic and deterministic models.