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 tests whether five-year U.S. Treasury yield data can improve forecasts of the S&P 500, motivated by the idea that stocks and government bonds may reflect shifts in risk appetite. It compares models predicting a future S&P 500 close using index…
The article explains multi-head attention as parallel self-attention operations with separate learned weights. Their outputs are concatenated and projected into a shared representation, allowing a model to represent different relationships among sequence…
The article presents a development workflow for using MATLAB 2018 computations in an MQL5 indicator. Its example applies a seasonal autoregressive integrated moving average model to price time series and combines prediction with Kalman filtering. MATLAB…
This article explains CryStAl, a population-based metaheuristic for optimizing a bounded, multidimensional search problem. It initializes a population of candidate solutions, then repeatedly updates each candidate using one of four randomized strategies…
This article explains DA-CG-LSTM, a neural architecture intended to forecast multivariate time series such as prices, volume, indicators, and news-related features. It outlines two attention stages: input attention weights features and time positions before…
The article explains a reinforcement-learning approach in which action values are updated after an episode ends, using accumulated returns rather than updating after every individual step. An episode groups a configurable number of trading cycles, and…
The document introduces HypDiff, a graph-generation approach that uses hyperbolic geometry to represent hierarchical structure and address problems encountered when applying ordinary diffusion models to discrete, sparse graphs. It outlines a two-stage…
This article explains why an MQL5 trading application needs to serialize requests to an AI service and parse JSON responses. It outlines a parser design centered on a value class that represents primitive types, arrays, and nested objects, with methods for…
The document introduces SAGDFN, a graph neural network framework for forecasting multivariate time series when relationships among assets change over time. It describes learning connections from data and using Significant Neighbors Sampling to focus on a…
The article describes a modified Whale Optimization Algorithm, a population-based method inspired by humpback whale hunting behavior. Candidate solutions move in relation to the current best solution through exploration and exploitation steps, including…
The article describes Fish School Search, a population-based metaheuristic for continuous optimization. Each candidate solution is represented as a fish with a weight reflecting search success. The method combines individual movement, a shared instinctive…
The document outlines an ARIMA training method implemented as an MQL5 class. It reviews autoregressive, moving-average, and differencing components, then describes fitting coefficients by generating predictions over known observations and minimizing their…
The article describes an implementation of HiSSD, a hierarchical skill-discovery framework, adapted for algorithmic trading. Its central design separates common skills that can apply across situations from task-specific skills derived from local…
The article introduces N-BEATS, a neural network architecture for time-series forecasting, and explains how stacked blocks iteratively produce forecasts and backcasts. Each block transforms a lookback window, estimates its contribution to the forecast, and…
This article sketches an MQL5 neural-network implementation for market data, combining a state-space model, transformer attention, multi-horizon context memories, and a verification module. It describes short-, medium-, and long-term memories alongside…
This article presents an MQL5 implementation of a trading model built around DA-CG-LSTM, which combines feature attention, temporal attention, and a modified recurrent block. The attention stages weight input variables and historical intervals; the recurrent…
The article introduces supervised decision trees and explains how the ID3 algorithm builds a classification tree. ID3 starts at the root and greedily selects a feature for each split, aiming to make the resulting groups more homogeneous. The tutorial focuses…
This article describes integrating the K²VAE framework into an Actor–Director–Critic trading agent. The model combines a Koopman representation for approximating nonlinear dynamics in latent space, a stabilized Kalman filter to update estimates with new…
The article presents Contrastive Intrinsic Control (CIC), an unsupervised reinforcement learning approach for discovering reusable agent skills, especially in continuous-action environments. It motivates CIC as an alternative to competency-based methods that…
The article describes a population optimizer built around a classroom metaphor. Candidate solutions are divided into classes; each class’s best member acts as its teacher. Students update their positions by learning from that teacher, later drawing on their…
This article explains ATFNet, a neural forecasting architecture that combines time-domain and frequency-domain models. Its time block uses segmented input to capture local patterns, while a complex-valued frequency block models global spectral structure.…
The article tests whether gold and palladium prices help forecast USDCAD. It explains the economic rationale for comparing the markets, then describes gathering price bars from MetaTrader, aligning observations by shared timestamps, and labeling a future…
The article demonstrates a computer-vision workflow for detecting chart patterns in screenshots from MetaTrader 5. It describes capturing overlapping chart images with a script, cleaning chart displays to reduce visual clutter, and applying a pretrained…
This tutorial describes a workflow for exporting historical MetaTrader data, loading it into Jupyter, and preparing it for statistical analysis. It covers inspecting columns and delimiters, converting price data to numeric values, sorting observations,…