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 describes FREL, a feature weighting and selection method based on regularized energy-based learning. It frames prediction as weighted nearest-neighbor classification: candidate predictors receive weights, and Manhattan distance determines how…
The article explains the Gravitational Search Algorithm (GSA), a population-based optimizer inspired by gravitational attraction. Candidate solutions act as agents whose fitness determines their relative masses. Agents attract one another, update their…
The article presents an implementation of MASAAT, an attention-based multi-agent framework for portfolio optimization. Agents analyze price sequences at multiple scales, using directional movement thresholds to represent shifts of varying size.…
The article surveys three normalization approaches for neural networks used in trading systems: standard scaling, feature scaling, and robust scaling. It discusses commonly proposed benefits of normalization, including steadier training and improved…
The article describes a custom MQL5 breakout signal that combines Disjoint Set Union (DSU) clustering with a Deep Belief Network (DBN). DSU groups adjacent high-volatility bars into time-bounded clusters using ATR expansion and Bollinger Band width as…
The article presents an integrated research pipeline intended to make trading machine learning experiments reproducible and auditable, from tick data through model export for MetaTrader 5. Its data architecture keeps tick data in a session-level RAM cache…
The article introduces Shannon entropy as a measure of uncertainty in a set of outcomes, illustrating how greater variety in event probabilities produces higher entropy. It then describes decision trees and random forests, including bootstrap sampling and…
The article presents a spatio-temporal neural network for multi-step forecasting from short multivariate time series. Its central idea is a spatio-temporal information transformation that recasts relationships among input variables as temporal information…
The article describes a framework for screening candidate trading strategies by estimating their historical returns. It builds market and indicator data in MetaTrader 5, assigns directional signals to moving average, RSI, and stochastic rules, and compares…
The article introduces convolutional neural networks and explains how convolution filters, stride, padding, activation functions, pooling, dense layers, and dropout contribute to feature extraction and prediction. It emphasizes one-dimensional convolutions…
The article describes elastic net regression as a way to regularize linear predictive models used in trading. Its penalty blends the lasso’s absolute-coefficient penalty, which can suppress predictors, with ridge regression’s squared-coefficient penalty,…
The article shows how a conventional LightGBM model can be adapted to forex time-series forecasting by encoding temporal information as features. It discusses lagged OHLC values, rolling statistics, date and calendar attributes, differencing, and…
The article introduces kernel density estimation as a way to estimate an unknown probability distribution, with particular interest in short and medium-length sequences that arise in market analysis. It contrasts the method with histograms, which can be…
The document explains Circle Search Algorithm (CSA), a population-based optimization method that moves candidate solutions in relation to the current best point using circle tangents and an adjustable angle. It describes exploration and exploitation phases,…
The article outlines an architecture linking an MQL5 Expert Advisor to a Python service for model predictions and post-trade feedback. The EA sends market features for inference, receives a prediction, and reports trade outcomes and related execution…
The article describes a machine learning extension to an earlier Expert Advisor using Fractal Adaptive Moving Average (FrAMA) and Force Index signals. It outlines Python implementations of both indicators: FrAMA adapts EMA smoothing using a simplified…
The article evaluates whether moving-average crossovers are easier to forecast than short-horizon price direction. Using EURUSD 15-minute data, it labels future outcomes based on the relative position of short and longer simple moving averages, then compares…
The article builds an hourly EURUSD breakout strategy around a channel formed from an earlier high and low. It waits for a breakout and subsequent price confirmation, then uses a fast and slow exponential moving average to filter entries. Average True Range…
The article describes a practical MQL5 implementation inspired by the GSM++ graph sequence framework. Its pipeline combines trainable mixture tokenization for bars, neighboring edges and subgraphs; adaptive local node encoding; and a global hybrid encoder.…
This article introduces population-based metaheuristic optimization and discusses how such algorithms can be assessed for convergence, speed, repeatability, and scalability as the number of variables grows. It contrasts population searches, which evaluate…
In this interview, trader Leonid Velichkovsky reflects on using neural networks in automated trading and challenges the idea that they produce exceptional profits by themselves. He describes them as trading systems whose adaptive capacity can help detect…
This article completes an MQL5 implementation of a hyperbolic latent diffusion framework for financial time series. It describes projecting inputs into hyperbolic space, mapping embeddings onto tangent planes, and generating centroids and curvature…
This article explains BFO-GA, a population optimization method combining bacterial foraging with genetic operators. Bacteria represent candidate solutions and move through the search space using gradient-guided and random movements. Selection, crossover, and…
This article explains HiVT, a hierarchical Transformer approach originally designed to forecast trajectories of multiple road users in autonomous-driving scenes. It represents agent paths and lane segments as vectors, using relative positions to preserve…