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
This article explains principal component analysis (PCA) as a preprocessing method for compressing correlated inputs before they enter a trading decision model. It frames the problem around large sets of indicators, instruments, and timeframes: reducing…
The article explores whether independent components analysis (ICA) can separate structure from noise in market data already smoothed by moving averages. It constructs a dataset from raw and SMA-filtered open, high, low, and close prices, along with lagged…
This article adapts closed-loop control theory to a moving average channel strategy. The baseline buys when price breaks above a channel formed from moving averages of highs and lows, and sells on a break below. A controller first observes 90 days of market…
The article introduces autoencoders as unsupervised neural networks that compress input features into a latent representation and reconstruct them. It describes using this approach to reduce noise in forex data and create alternative candlestick series for…
The article presents an MQL5 workflow for preprocessing training data and selecting quantization tables for CatBoost predictors. A script loads a sample, tests candidate tables by measuring reconstruction error for each predictor, and records the best…
This article introduces receiver operating characteristic curves as a way to evaluate binary classifiers across decision thresholds. It defines the confusion-matrix outcomes and uses them to explain the true positive rate and false positive rate. Plotting…
The article surveys fractal ideas for describing financial time series whose behavior includes heavy tails, volatility clustering, and patterns that vary across scales. It explains self-similarity and the Hurst exponent as indicators of persistence, mean…
The article introduces multi-horizon forecasting and compares three approaches: training a separate model for each forecast step, iteratively feeding predictions from a single model back into itself, and predicting several horizons together with a…
The Comet Tail Algorithm is a population-based optimization method inspired by comets and their tails. Candidate solutions are treated as comet nuclei, while clouds of particles sample nearby regions. The best solution acts as the sun: tail size and…
This tutorial refreshes the structure and implementation of decision trees for classification and regression. It explains internal decision nodes, leaf predictions, feature thresholds, and recursive tree construction, then contrasts CART, ID3, and C4.5. The…
This article describes Artificial Ecosystem-based Optimization (AEO), a population-based search method inspired by production, consumption, and decomposition in ecosystems. It represents candidate solutions as organisms with different roles. In the…
This article explains Node-Adaptive Feature Smoothing (NAFS), a graph representation method that combines node features with information from neighborhoods at multiple smoothing scales. Rather than applying the same amount of smoothing to every node, it…
The article presents CCMR, a coarse-to-fine optical-flow method adapted from computer vision, and describes an MQL5 implementation for analyzing financial data. The model extracts image and context features at several resolutions, computes global context…
The article introduces unsupervised clustering and implements k-means in MQL5. It walks through selecting initial centroids, measuring each observation's distance to each centroid, assigning observations to the nearest cluster, grouping observations, and…
This article evaluates a trading robot that combines a discrete Fourier transform with a leaky integrate-and-fire spiking neural network. The Fourier component identifies the strongest frequency in a rolling window and maps its phase to a directional score;…
This article explains how to collect government bond yields from web pages and use them to derive interest rate differentials for forex analysis. It describes finding the data source in a page or its linked resources, extracting the quoted yield, and…
This tutorial explains a workflow for using a neural network trained in NeuroSolutions 5 from an MQL5 Expert Advisor. Its example strategy predicts a daily bar’s close at the bar’s opening, using OHLC data from five previous bars represented as changes…
This article outlines a foreign-exchange forecasting system that combines a fine-tuned Llama language model with a Self-Evolving Adversarial Learning agent. It describes building labeled examples from MetaTrader data, balancing upward and downward classes,…
The article introduces Nelder–Mead as a derivative-free method for minimizing an objective function by moving a simplex through parameter space. It describes sorting simplex vertices by objective value, calculating the centroid excluding the worst vertex,…
The article explains ST-Expert, a neural forecasting framework intended to handle changing relationships among financial assets. It frames market data as a sequence of regimes, each with its own correlation structure, and proposes combining learned…
This article walks through a single-layer neural network intended to predict a continuous market price value. It explains the forward pass as matrix multiplication of inputs by weights, addition of biases, and application of an activation function. It…
This tutorial explains how long short-term memory (LSTM) and gated recurrent unit (GRU) networks process sequences. It describes their gates and state updates, and contrasts them with simple recurrent networks, which can struggle to learn long-range patterns…
The article introduces logistic regression as a binary classification method. It explains how a linear combination of features is passed through a sigmoid function to produce a probability between zero and one, then describes using a threshold to assign a…
The article introduces Soft Actor-Critic (SAC) as a reinforcement-learning method for trading systems. It describes an actor network that outputs a Gaussian action distribution and two critic networks that estimate action values. Using the lower critic…