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 presents an Expert Advisor design that combines a Hidden Markov Model with regime-specific Gated Recurrent Units. The HMM estimates whether conditions are ranging, trending, or unusually volatile from trend-strength and relative-volatility…
The document explains Mantis, a foundation model for classifying time-series patterns and market regimes. It describes splitting sequences into patches, combining local convolutional features with global attention, and using separate streams for normalized…
The article evaluates a two-sided CUSUM indicator that accumulates standardized log returns and flags a breakpoint when either accumulator crosses a threshold. It tests the Siegmund average run length formula, commonly used to estimate the spacing of false…
The article reviews TEMPO, a forecasting architecture that adapts a pretrained GPT-style model to time series. TEMPO first separates a series into trend, seasonal, and residual components, maps them into embeddings, and combines the components with learned…
This research compares population-based optimization algorithms on test functions designed to assess escape from local extrema and progress toward global optima. It reports normalized scores for Hilly, Forest, and Megacity functions at several sizes, with…
The article introduces the conventional restricted Boltzmann machine (RBM) as a network that learns hidden representations through positive and negative passes, with weights adjusted by contrastive divergence. It sketches an MQL5 implementation using Gibbs…
The article proposes a custom trailing stop for momentum trading that combines a price boundary with a sequential momentum filter. A rolling convex hull is used to locate structural price extremes, while a compact gated recurrent unit (GRU) evaluates recent…
The Neuroboids Optimization Algorithm (NOA) is a population-based method in which each candidate solution has its own small neural network. Agents begin at random positions, train their networks with Adam using the best solution found as a target, and use…
The article introduces Neural ODEs, which describe hidden-state changes with a neural network and use a differential-equation solver to evolve the state through continuous time. It explains the adjoint sensitivity method for training: gradients are obtained…
The article explains goal-conditioned reinforcement learning (GCRL) as a way to make an agent’s action depend on both market state and a specified subtask, such as opening or closing a position. It contrasts this with skill conditioning and describes…
This article addresses a Soft Actor-Critic trading model that had not trained profitably in an earlier installment. Its central change replaces action perturbations chosen independently of the policy with stochastic sampling from the actor’s learned action…
The article extends a quarterly indicator-ranking pipeline into a predictive model for the XLF ETF. It uses FrAMA, Parabolic SAR, Alligator lines, and TRIX as aligned features, then trains a multilayer perceptron to classify future returns or predict them as…
This article explains how to rebuild trading-system objects from parameter strings stored in a database. It distinguishes strategy parameters from tester inputs, noting that the stored strategy representation contains properties needed to recreate an object,…
This tutorial demonstrates a basic feed-forward neural network using the FANN library through MQL. The example is deliberately independent of market data: it trains on triples of numbers and assigns a binary label according to whether the first and third…
The article discusses using ChatGPT as an assistant for MQL4 and MQL5 development, including generating, refactoring, commenting on, and debugging code, as well as translating mathematical models into implementations. It recommends treating the tool as a way…
This article details a practical token-generation stage for a LightGTS-style forecasting model. It first estimates each input series’ dominant frequency with a Fourier transform, then uses that estimate to set patch length and overlap while keeping the…
The article describes a compact support vector machine classification approach for two-dimensional trading data, implemented in MQL5 without an external library. It outlines polynomial kernels and support vectors, then uses Newton polynomial interpolation to…
The article presents an MQL5 trading system based on Quantum Reservoir Computing. A fixed four-qubit reservoir maps market features into a larger state representation, while learning updates the output layer using a buffer of recent examples. The proposed…
This article explains why fixed-time bars can assign equal weight to periods with very different trading activity, then presents activity-based alternatives inspired by AFML. It covers time, tick, volume, and dollar bars, plus imbalance bars that close when…
This article describes a self-optimizing Expert Advisor that uses a genetic algorithm to choose among predefined strategies, financial instruments, deposit allocation, and indicator parameters. Candidate settings are evaluated by simulating trades on…
The article outlines a EURUSD mean-reversion system that uses weekly moving averages of highs and lows to form a channel. A move above the upper average is treated as an overbought condition and a short candidate; a move below the lower average is treated as…
The article presents a workflow for connecting MetaTrader 5 to the ENCOG machine-learning framework through a C# bridge. Its example uses a feedforward neural network trained with resilient propagation to forecast financial time series. Stochastic %K,…
The article proposes forecasting SPDR’s XLY consumer discretionary ETF with quarterly macroeconomic and consumer data rather than relying mainly on the ETF’s own price and volume history. It reviews candidate datasets covering economic indicators, interest…
The article introduces the Artificial Coronary Circulation System, a population-based optimization method inspired by coronary growth. Candidate solutions are treated as capillaries, and their fitness informs global movement, local refinement, branching, and…