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 workflow for adding AI-generated commentary to an MQL5 news headline Expert Advisor. It describes downloading a quantized language model, running it locally through a Python inference service, and having the EA send prompts to a web…
The article describes ways to transform existing variables and create or select predictors for a neural-network classification task. It discusses logarithms and roots for changing skew, discretization for grouping values, and normalization, alongside outlier…
This article presents an MQL5 implementation of a TQNet temporal query model, focusing on a multi-head attention module that combines source time-series data with accumulated global correlation parameters. The module extends a cross-attention layer, uses a…
This article explains how a Kohonen self-organizing map (SOM) can organize multidimensional market features into a two-dimensional grid. Each input pattern is matched to its closest neuron using Euclidean distance, then the winning neuron and nearby neurons…
This article describes a workflow for turning single-candle patterns into machine-learning inputs. It defines patterns using rules on open, high, low and close prices, including thresholds for doji bodies, shadow lengths and body size. Those rules feed an…
The article explains propensity scores as a way to reduce the dimensionality problem in causal matching. Rather than matching observations across many covariates directly, researchers estimate each observation’s probability of receiving treatment and match…
The article describes a pipeline for adapting a pretrained language model to forecast currency-pair direction from historical MetaTrader data. Its dataset method labels whether price is higher or lower after 24 hours, samples balanced UP and DOWN examples,…
This article compares forecasting price direction with predicting whether a future close will exceed the current high, fall below the current low, or remain between those levels. The latter target formalizes a classic price action idea: traders look for…
Alexander Topchylo describes managing Forex accounts after winning the Automated Trading Championship and explains his preference for giving investors access to a detailed, public trading history. He discusses moving toward stocks and futures and argues that…
This article describes a workflow for recognizing potential liquidity zones with a Graph Neural Network. It represents detected swing highs and lows as graph nodes, uses price, tick volume, and swing type as node features, and links neighboring swings.…
The article presents NOA2, a population optimization method that combines boid-style group movement with individual neural networks. Agents use cohesion, separation, and alignment to explore a search space, while their networks adjust movement and the…
This article introduces association rule mining as an unsupervised method for finding item combinations that recur in transaction data, then explains how the ideas may be applied to trading. It defines support as the frequency of an itemset in the data and…
The article argues that decentralized broker feeds can produce materially different histories for the same forex symbol, making trading research and machine learning results difficult to reproduce across brokers. It compares four years of daily EURUSD data…
This article develops an automated workflow for optimizing trading-strategy parameters and assembling selected results into a standalone expert advisor. It defines a hierarchy of projects, stages, jobs, and tasks, each with queued, processing, and completed…
The article describes a trading adaptation of the Hidformer multivariate time-series architecture. Its dual towers analyze temporal patterns and frequency-domain information separately, using recursive attention for sequence dependencies and linear attention…
The article presents a feature pipeline built around Bill Williams' fractals, defined as swing highs or lows confirmed by bars on both sides. It shows how these patterns can feed strength and validity measures, support and resistance levels, breakout flags,…
This article lays out a practical pipeline for deciding whether to develop and evaluate a reinforcement-learning trading system for gold. It begins with a supervised LightGBM baseline using triple-barrier labels and purged walk-forward folds to check whether…
The article explains ASHA, a metaheuristic optimizer that represents candidate solutions as water units moving across an objective-function landscape. It describes initialization, probabilistic moves toward random lower positions or the current best, an…
This article proposes representing a bar’s open and close prices as Cartesian coordinates, then converting them into polar coordinates: a radial distance and an angle. It explains how trigonometric transformations and derivatives can be calculated in MQL5,…
Artificial Multi-Social Search (MSO) is a population optimization method in which groups of particles move among sectors of a multidimensional search space. Each group tracks its center and best solutions, while particles and groups retain memories of…
The article explains how feature selection can reduce redundant or noisy inputs, simplify machine learning models, and potentially improve training efficiency and predictive performance. It groups techniques into filter methods, wrapper methods, embedded…
The article explains DADA, a deep learning framework for detecting unusual patterns in time series, and begins an MQL5 interpretation of its architecture. DADA masks portions of input sequences and trains an encoder to reconstruct them. Its adaptive…
This article introduces local deployment of open-source large language models as a foundation for later integration with algorithmic trading systems. It explains the proposed benefits of keeping data under local control, reducing network latency, and…
This article describes a proposed S&P 500 Expert Advisor that combines a linear regression forecast with trend-following technical filters. It uses historical prices from a selected group of large index constituents as inputs to an ordinary least squares…