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 outlines a custom MQL5 trailing module that uses a Bloom filter to screen repeated prices and a recurrent neural network (RNN) state to moderate stop adjustments. Price values are converted to point-based integers, hashed into a compact bit…
Explore, Discover and Learn (EDL) is an unsupervised skill-learning method for hierarchical reinforcement learning. It addresses limited state coverage in approaches that train skills from a prior distribution. EDL first gathers a broad sample of environment…
The Code Lock Algorithm (CLA) is a population-based optimization method that encodes each candidate parameter as a sequence of decimal digits. Digit sequences map to values within each parameter’s search range. Candidate solutions are modified by copying…
This experiment adds features from a fixed three-qubit circuit to a bidirectional LSTM that predicts the next EURUSD hourly candle’s direction. Recent-window mean return, volatility, and range determine rotation angles; linked qubits produce an eight-outcome…
The Artificial Tribe Algorithm (ATA) is a population-based optimization method inspired by tribal behavior. It represents candidate solutions as individuals with personal best positions and a shared best position. The method switches between two search…
This article describes an experimental machine-learning system for forecasting movement across eight currency pairs on a 15-minute timeframe. It combines four groups of inputs: 3D-bar features derived from price, time, volume, and volatility; a quantum…
The article presents Temporal Query Network (TQNet), a neural architecture for forecasting multiple related time series. Its attention layer forms queries from trainable vectors that are shifted periodically through time, while keys and values come from the…
The article models an open trade’s exit as a discrete-time survival problem. At each bar, a position can hit its take-profit, hit its stop-loss, or remain open; a time-stop position is treated as censored. It fits two cause-specific hazards with multinomial…
This article presents the Billiards Optimization Algorithm, a population-based method for numerical optimization inspired by balls moving toward pool pockets. It begins with candidate solutions, treats the best candidates as targets, and updates each agent…
This article explains how tensor frameworks can support a Soft Actor-Critic implementation, focusing on replay buffers and critic networks. It contrasts a manual NumPy buffer with a PyTorch tensor buffer that stores states, actions, rewards, next states, and…
The document explains how standard column reduction turns a boundary matrix over the two-element field into a persistence diagram. Each diagram pair records when a topological feature appears and disappears as the Vietoris-Rips scale grows. The article…
The article describes a permutation test for assessing whether an Expert Advisor’s performance may arise by chance. It proposes testing the EA on an original price sample, then repeating the test on many altered versions and comparing a selected performance…
The article introduces reinforcement learning through the interaction of an agent and an environment. The agent observes a state, chooses an action under a policy, and receives rewards as the environment changes. It contrasts this setup with supervised…
The article explains the Butterfly Optimization Algorithm (BOA), a population-based search method inspired by butterflies following fragrance signals. It describes fragrance as a function of solution quality and sensory parameters, with a switching…
The article adapts Relative Molecule Self-Attention Transformer ideas to financial forecasting. Its central method encodes relationships between pairs of input elements, including relative position, distance, and contextual or neighborhood information, then…
The article describes a trading decision architecture in which three language-model analysts review the same market briefing from different roles: bullish, bearish, and risk-focused. A fourth model acts as an arbiter. Its stated rules force a hold when the…
The article explains how to speed up neural network training in an MQL5 Expert Advisor by moving repeated neuron calculations to OpenCL kernels. It contrasts the terminal’s thread allocation with OpenCL’s ability to run work across supported CPUs and GPUs,…
This introductory article explains feedforward neural networks, focusing on multilayer perceptrons and their components. It describes how inputs are combined with weights and bias in hidden layers, then passed through activation functions. It distinguishes…
This article outlines a file-based messaging design for connecting Python machine learning models with MQL5 Expert Advisors during Strategy Tester runs. Because network requests are unavailable in the tester, the author proposes exchanging CSV files through…
The article develops a method for estimating whether signals from a moving average crossover are likely to be profitable. It builds on a more responsive crossover using separate moving averages on opening and closing prices, then compares prediction of the…
This article introduces Transformer architecture and focuses on the Temporal Fusion Transformer (TFT) for multi-horizon time-series forecasting. It describes attention, parallel training, positional information, and differences among encoder, decoder, and…
The article argues that a model’s forecasts may be difficult to compare directly with observed prices because predictions and targets can occupy differently transformed coordinate spaces. It explains this idea through a geometric interpretation of linear…
The article examines how uneven bullish and bearish labels can bias financial classification models toward the more common class. It illustrates the issue with daily candle direction counts across several instruments and a USDJPY Random Forest example.…
The article explains Symbolic Aggregate Approximation (SAX) for converting price windows into compact symbolic sequences. It first normalizes each window to emphasize shape, averages values within equal segments, then maps those averages to symbols using…