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
Blood Inheritance Optimization (BIO) is a population-based method for searching parameter spaces. It assigns each candidate one of four blood-type labels and uses an inheritance table to choose a label for each new candidate based on its two parents. Parents…
The article introduces random forests as ensembles of decision trees trained on bootstrap samples and randomized feature subsets. It explains how tree voting can reduce variance, describes out-of-bag evaluation, and outlines strengths such as modest…
The article introduces perceptrons for forecasting Forex prices, describing supervised training on historical inputs and the adjustment of model weights to reduce prediction error. It discusses choices such as network size, activation functions, learning…
This article explores representing price behavior as geometric features for a perceptron rather than feeding it raw prices or oscillator values. The proposed inputs use two simple moving averages, with periods of one and twenty-four, and describe their…
The article proposes labeling machine-learning training examples from oscillator states available at the time of a trading decision. RSI, CCI, Stochastic, Bollinger Bands, and a Fourier-based method are among the examples. Values beyond overbought or…
This article describes a custom MQL5 CFileCSV class intended to support communication between MetaTrader 5 and Python machine-learning workflows. CSV files act as a simple handoff format for transferring data to Python and returning predictions to the…
The article introduces the Geweke–Porter–Hudak estimator for measuring the fractional differencing parameter d from the low-frequency slope of a log-periodogram. It distinguishes this estimate from the Hurst exponent: the measures are theoretically linked,…
The article presents the Duelist Algorithm as a population-based method for optimizing strategy parameters. Candidates compete in randomized pairwise duels; the better-performing candidate usually wins, while a luck term can affect the outcome. The algorithm…
The article explains Isolation Forest, which identifies unusual observations by repeatedly splitting feature space at random and measuring how quickly each point becomes isolated. It describes the anomaly score, the correction for tree branches that stop…
The article introduces PatchTST, a Transformer approach that groups consecutive time steps into patches before modeling a multivariate time series. It treats each variable as a separate univariate channel, shares model parameters across channels, and uses…
The article adapts a conditional generative adversarial network to forecast financial time series within an MQL5 expert advisor. It describes a generator and discriminator, both based on a multilayer perceptron: the generator uses prior close-price changes…
The article describes a MetaTrader 5 workflow for combining many optimized strategy parameter sets in one expert advisor. It focuses on EURUSD hourly trading and uses perceptrons built from TEMA slope inputs, with optimized weights and a threshold parameter…
The document presents the second-level path signature, especially its antisymmetric component called Levy area, as a rolling measure of ordering between two time series. Rather than scanning candidate lags as cross-correlation does, it treats paired…
The article develops the first two stages of the Mapper topological data analysis pipeline for price point clouds: assign each point a scalar lens value, then divide the lens range into overlapping intervals called a cover. It describes three lenses:…
The article explains symbolic regression as a way to model relationships between input data and a target without choosing a fixed equation form in advance. It represents candidate models as expression trees built from randomly selected coefficients,…
This tutorial uses a single sigmoid neuron trained on a small AND-gate dataset to explain the cost of numerical gradient estimation. In the initial version, the code estimates changes in the loss by repeatedly evaluating the cost function after perturbing…
This document introduces the CFTC’s Commitments of Traders and Traders in Financial Futures reports as sources of aggregated futures positioning. It describes commercial hedgers, non-commercial speculators, and smaller traders, and explains how TFF breaks…
This tutorial explains how to frame a trading problem for CatBoost gradient boosting without requiring Python or R expertise. It distinguishes manually written trading rules from supervised learning: predictors encode market information, while a target…
The document explains how to represent time in machine-learning features for forex. It argues that raw integer timestamps and hour values misrepresent cyclical adjacency, then describes encoding periodic variables with sine and cosine pairs. Adding Fourier…
This article begins a native MQL5 implementation of the Multifractal Model of Asset Returns (MMAR), focusing on partition analysis. For each moment order, the method calculates partition functions across multiple time scales, fits lines to their logarithms,…
This article proposes a continuous-space variant of Tabu Search for optimization problems. It partitions each parameter range into sectors and tracks successful and unsuccessful outcomes with separate white and black counters. New candidate coordinates are…
The article applies feedback control and linear system identification to a simple moving-average strategy. The benchmark buys when price is above the average and sells when it is below, with fixed strategy parameters and ATR-based trade levels. The…
The article explains Pointformer, a Transformer architecture for learning from unordered, irregular point clouds. Its hierarchy combines local attention within point neighborhoods, cross-attention between resolutions, and global attention across the scene.…
This article continues a practical implementation of the Chimera framework, a two-dimensional state space model that learns dependencies across both time and input features. It describes an MQL5/OpenCL neural network layer: input projections are formed…