Yevgeniy Koshtenko / Profile

The article examines three key obstacles to integrating LLMs with MetaTrader 5: the lack of direct access, strict rate limits, and API key security given the architectural limitations of MQL5. A configuration is proposed that uses a local Python server as a bridge between the Expert Advisor and OpenRouter. The article covers WebSocket and fallback to TCP, storing the key on the server, batch processing of multiple symbols, and constructing a technical prompt. Readers get a ready-made architecture that reduces latency and costs.

Статья показывает, как формализовать интуитивно замеченные ценовые паттерны и превратить их в статистически проверенные торговые сигналы. Советник кодирует последовательности баров в бинарные строки U/D и для каждого паттерна вычисляет пять независимых метрик: поддержку, уверенность, лифт, хи-квадрат и байесовскую вероятность. Позиция открывается только тогда, когда текущий паттерн совпадает с историческим правилом и все фильтры пройдены — динамический лот масштабируется по силе сигнала, стоп и тейк рассчитываются через дневной ATR.

The article provides a critical analysis of an LLM strategy in which forecasting the direction is separated from trading decisions, and demonstrates why this leads to a disconnect between metrics and PnL. We will describe procedures for dataset balancing, feature engineering, prompt and response preparation, fine-tuning configuration in Ollama, and reliable parsing. Backtesting and forward testing reveal systematic degradation. The practical conclusion is that the problem must be formulated as a direct optimization of trading outcomes.

A hybrid architecture based on Llama 3.2 and SEAL is being tested on eight currency pairs (M15), with forward-period data isolation and information leakage control. The methodology combines adversarial self-play, curriculum learning, and class balancing to ensure stable training. The experiments confirm the gap between forecast accuracy and actual returns, providing readers with practical guidelines for testing strategies and accurately assessing their generalizability.

The article proposes a hybrid approach to algorithmic trading based on quantum encoding of market states, Double DQN with a prioritized experience replay buffer, and an LLM acting as a contextual EA. The SEAL methodology enables asynchronous continued training of the agent without halting trading. A lightweight Q-learning filter (USE/SKIP/REDUCE) controls signal execution at the meta-level. Practical details are provided on integrating the system with the MetaTrader 5 trading platform, along with a scheme for adapting it to market regime shifts.

SEAL (Self-Evolving Adaptive Learning) is a system for the continuous adaptation of large language models (LLMs) for algorithmic trading, designed to address the problem of rapid model degradation in changing markets. Instead of periodic retraining, which takes hours and erases old patterns, SEAL learns from every closed trade, maintains priority memory for important examples, and automatically initiates incremental fine-tuning when accuracy drops or a market regime change occurs.

The article presents a reproducible implementation of a hybrid quantum-neural network model for algorithmic trading on Forex without using real quantum hardware. A fixed three-qubit quantum circuit in IBM Qiskit converts sliding-window statistics (mean returns, volatility, and range) into a probability distribution, from which seven quantum metrics are calculated. These features are integrated into a bidirectional LSTM architecture with regularization and mechanisms to address class imbalance, including focal loss and a sampler.

The article presents the full integration of the 3D-bar module into a quantum-enhanced trading system for forecasting the movement of currency pairs. The system combines stationary four-dimensional features, an 8-qubit quantum encoder, and CatBoost gradient boosting with 52+ features. The system is implemented in Python using MetaTrader 5, Qiskit, CatBoost, and optional integration with the Llama 3.2 LLM for interpreting forecasts.

The article proposes a synthesis of new technologies to overcome the limitations of classical indicators in market data analytics. It shows how language models and quantum encoding can reveal hidden market patterns that traditional methods overlook. The experiment confirms the value of new technologies and proposes an updated analysis methodology aligned with the current state of computational innovation.

The article describes the practical implementation of a hybrid algorithmic trading system that combines quantum computing (IBM Qiskit) and gradient boosting (CatBoost) to predict movements in the EURUSD pair on the hourly time frame. The system extracts four unique quantum features from a probability distribution across 256 states using eight qubits and, in combination with classical indicators and delta encoding of time categories, achieves 62% accuracy on 15,000 candlesticks.

The article describes the process of fine-tuning a language model for trading based on real historical data from MetaTrader 5. The base model, which has only theoretical knowledge of technical analysis, is trained on a thousand examples of the real behavior of currency pairs (EURUSD, GBPUSD, USDCHF, USDCAD) over 180 days. After being trained using Ollama, the model begins to understand the specific characteristics of each instrument.