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 explains ridge regression as a way to estimate linear model coefficients when predictors are correlated. It frames regularization as a bias–variance tradeoff: accepting some bias can reduce variance and overfitting. It also contrasts ridge with…
The article applies Gaussian Naïve Bayes to classify whether a bar closes above or below its open, using Bulls Power, Bears Power, RSI, tick volume, and Money Flow Index as features. It explains preparing a labeled matrix, splitting observations into…
The article turns topological features of a rolling price window into chart and Expert Advisor buffers. It defines persistence entropy separately for H0 connected-component bars and H1 loop bars: entropy is low when persistence is concentrated in a few…
The article turns rough-volatility theory into a rolling local Hurst estimate for an XAUUSD intraday trading system. It blocks short-horizon returns into realized-variance observations, takes their logarithms, and estimates roughness from the slope of log…
The article considers which data sources might help a multilayer perceptron forecast the next quarter’s direction for the SPDR XLV healthcare ETF. Candidate inputs include historical OHLC changes, volatility, volume, insurance claims, pharmaceutical sales,…
The article explains genetic algorithms as gradient-free methods for optimizing parametric models, including neural trading models that are not differentiable or are difficult to train with gradient descent. It describes evolving a population of agents…
This article describes Biogeography-Based Optimization (BBO), a population-based method in which each candidate solution is modeled as a habitat and its quality as habitat suitability. Better solutions have higher emigration rates and can share selected…
The article describes Attraos, a neural forecasting framework that draws on chaos theory. It frames market series as observations of a more complex dynamical system and uses phase-space reconstruction, with embedding dimension and time delay, to represent…
This article explains how to apply agglomerative hierarchical classification (AHC) and dendrograms to historical price bar ranges. It treats each range as an observation, groups similar observations into clusters, and uses their hierarchical relationships to…
This tutorial demonstrates connecting an MQL trading program to an external JSON data service, saving and parsing API responses, and using forecast data as a filter for a moving-average crossover. The example combines a directional moving-average signal with…
This article describes how to train and evaluate an MSFformer-based environmental state encoder for market time-series forecasting. It combines batch normalization, a feature extraction module that builds multiscale features, and stacked pyramid attention…
The article explains differential evolution (DE), a population-based metaheuristic for optimizing objectives when gradients are unavailable or the function is difficult to analyze. It represents candidate solutions as vectors, creates trial vectors by adding…
The article presents a modified population-based optimization algorithm inspired by dingo hunting behavior. It changes three parts of the original method: group attacks move toward the best solution by changing the sign and scaling of the attack…
The article offers a geometric interpretation of machine learning prediction error. It argues that a model maps supplied features into a representation that may not align with the space of the true target, so predictions can differ from reality even beyond…
The article describes Extremal Optimization (EO), a population-based metaheuristic inspired by the Bak–Sneppen ecosystem model. Rather than concentrating updates on the best solution, EO ranks agents and their components, then selects poorly ranked elements…
The document introduces gating as a way to adapt how an ensemble uses specialist models according to contextual variables or model outputs. It distinguishes selecting one model for each case from combining several models with context-dependent weights.…
The document explains COMBI, a single-layer form of the Group Method of Data Handling. It generates candidate linear models from combinations of input variables, estimates their coefficients on a training sample, and uses performance on a separate test…
The document explains online learning as a way to update predictive models as new market observations arrive. It contrasts this approach with batch training and discusses adaptability to changing patterns, lower memory demands for incremental methods, and…
The article discusses TEMPO, a time-series forecasting approach that adapts ideas from pretrained language models. Its general method decomposes a series into trend, seasonal, and residual components, then encodes those components for attention-based…
The article applies a beta variational autoencoder to binary event features built from the Stochastic Oscillator and Fractal Adaptive Moving Average (FrAMA). It focuses on three patterns that had underperformed in earlier tests, encoding indicator conditions…
GoertzelBrain combines Goertzel frequency analysis with an ensemble of online-trained neural networks to turn detected cycle structure into a directional confirmation signal. It extracts cycle period, amplitude, spectral confidence and their changes,…
This article adapts Directional Diffusion Models, originally developed for graph representation learning, to financial market data. Standard diffusion adds isotropic Gaussian noise, which can erase structure quickly in anisotropic data. The proposed method…
The document presents a workflow for labeling financial time-series data by trend. It retrieves bars from a MetaTrader 5 terminal with Python, converts the returned data into a pandas DataFrame, converts timestamps, and selects the close-price series for…
The document explains FITS, a compact neural-network approach to time-series forecasting. It transforms an input window with the fast Fourier transform, interpolates its complex frequency representation using a complex linear layer, then applies the inverse…