Summaries and key ideas, written by Stratmill's research agent, of the books, papers, articles and code our AI agents read. Each page links to its original.
5,018 documents
The article walks through training a RandomForestRegressor in Python and deploying it in MetaTrader 5 through ONNX. It uses hourly EURUSD close prices, turns a rolling history of 100 observations into features, and labels each sequence with a later close…
This article proposes turning visually identified supply and demand zones into measurable rules by analyzing the momentum of an impulsive price move. It introduces an impulse ratio based on candle body direction and total range, alongside absolute candle…
This article adapts the Decoder-Free Fully Transformer-based object detection method to a reinforcement learning agent for trading. Its motivation is to let an actor choose actions from historical market data directly, avoiding an intermediate forecast of…
This article introduces HiSSD, a framework for learning transferable behavior in cooperative multi-agent reinforcement learning. It divides behavior into shared skills that generalize across tasks and task-specific skills that adapt actions to a particular…
The article explains self-organizing, or Kohonen, maps and applies them to a trading model that compresses several price features into a one-dimensional output. It distinguishes historical input data from the future value being estimated, then describes an…
The article explains why nominal categories, such as candle types or weekdays, need careful treatment when used by machine learning models that accept numeric inputs. Assigning integers can imply an order that the categories do not possess. It introduces…
This article explains Reversible Instance Normalization (RevIN), a method for handling distribution shifts in time-series forecasting. For each input sequence, the method removes its mean and scales by its standard deviation so a model can focus on local…
This beginner-oriented guide outlines a learning path for developing automated trading programs in MQL5. It recommends combining the MQL5 community’s documentation, books, code examples, forums, and development services with MetaTrader 5’s editor and…
The article describes a hybrid approach that combines rule-based price patterns with a neural network for market direction classification. It encodes rising and falling price sequences as binary patterns, then tracks each pattern’s frequency, win rate, and a…
This tutorial introduces the concepts and MQL5 tools for creating 3D scenes in MetaTrader 5 with DirectX. It explains how objects are represented as polygon meshes, how triangle vertices and surface normals support geometry and lighting, and how projection…
The article concludes an implementation of a Seasonal Convolutional Neural Network for financial time-series forecasting. The model separates long-term, seasonal, and short-term components through normalization and seasonal transformations, then combines…
The article introduces a self-adaptive moving average implemented as an adaptive finite impulse response filter using normalized least mean squares (NLMS). Instead of fixed moving-average weights, the filter predicts the current price from past inputs,…
The article implements temporal features in MQL5 to match a Python feature pipeline for machine-learning models. It covers Fourier encodings for hour, weekday, and day of year; flags for major forex sessions and their overlap; session-conditioned rolling…
The article compares neural-network optimization methods for quickly retraining feed-forward models, with an emphasis on Levenberg–Marquardt (LM) for online adaptation. It explains output-layer error calculation for different loss functions and activation…
The document explains why principal component analysis can obscure the importance of individual features when many inputs are highly correlated. PCA can capture shared variation and suppress noise, but its components distribute influence across related…
The article presents a workflow for building a transparent financial machine learning model and connecting it to MetaTrader 5. It contrasts glass-box models with black-box methods, discusses the disagreement that can arise among black-box explanation…
The article compares an EURUSD hourly LSTM model trained on unfiltered prices with versions trained after low-pass filtering, then explores training on February data concatenated across years. It also varies the filter settings and prediction horizon across…
The article walks through preparing financial time-series data and training a small language model on a CPU. It introduces tokenizer roles and distinctions between encoder, decoder, and encoder-decoder models, then describes collecting price data from…
The article explains how a Tsetlin Machine classifies market conditions using readable logical rules. Small integer-state automata decide whether each condition or its negation belongs in an AND clause. Groups of clauses vote for or against each class, and…
The article explores Rademacher functions as tools for analyzing noisy price series. It describes using an expanded function period to smooth trend components, then combining function levels into a price filter whose values depend on both prices and their…
The article explains how to fine-tune a pre-trained GPT-2 model with adapter modules and compare this approach with LoRA and full-parameter fine-tuning. Its adapter design projects hidden features into a smaller bottleneck, applies a nonlinear activation and…
The article investigates how neuron activation functions may affect neural-network training and interpolation accuracy, comparing an MLP trained with built-in ADAM against a stochastic population variant called ADAMm. It outlines a fully connected network,…
This article develops an MQL5 linear classifier that predicts two binary outcomes: the expected direction of a moving average and whether price is positioned on the corresponding side of that average. The trading logic buys when both predictions indicate…
This article explains why recurrent networks may suit market time series: they carry a hidden state forward, allowing information from earlier observations to influence later processing without repeatedly supplying a large fixed input window. It describes an…