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 trading bot that connects an MT5 Expert Advisor to a Python server over a socket. The EA sends a symbol and Twitter API credentials, requests sentiment data at regular intervals, and receives a sentiment value. The described trading…
The article describes the Blue Monkey algorithm as a population-based method for optimizing discrete trading robot parameters. It divides adult candidates into groups, moves each toward its group’s best candidate, and separately moves a smaller offspring…
The article argues that common regression losses such as RMSE, MSE, and MAE can favor a forecast near the target’s historical mean. In noisy financial data, a model that predicts the average return may score well without learning a signal useful for trading.…
The article presents a workflow for using CatBoost to filter trades from a moving average crossover trend-following strategy. The model is trained to classify trade outcomes as wins or losses, using market features recorded for each trade. The proposed role…
This article investigates why an automated optimization pipeline failed to produce a final Expert Advisor database for one project. It traces the project through its stages by inspecting SQLite records for stages, jobs, tasks, and completed passes. Short…
This article continues a machine learning trading project by choosing binary classification and training an XGBoost model on engineered features. It describes separating labels from numeric inputs, fitting the classifier, and evaluating predictions on a…
This article adds Golden Cross and Death Cross logic to a multi-strategy MetaTrader 5 Expert Advisor. The signals are based on a faster moving average crossing above or below a slower one, with the commonly cited long-period pairing used as an example. The…
This article surveys data preparation for neural network classifiers, focusing on filtering mislabeled training examples and reducing input dimensions. It compares data-level noise filtering with methods that try to make the classifier itself more robust,…
This reference article surveys MQL5's matrix and vector types and their built-in methods for mathematical and data-processing work. It covers object creation and initialization, copying arrays and timeseries, matrix and vector arithmetic, transformations,…
This article explains PointNet++, a neural network architecture that learns local and global structure from unordered point sets. It contrasts the method with PointNet, which aggregates points into a global representation without capturing local…
The article proposes replacing sequential trailing-stop updates with a probabilistic skip-list search, paired with a Hopfield network that checks whether recent price moves still fit a broader market pattern. The skip list uses sparse upper levels to search…
The article presents Adaptive Social Behavior Optimization (ASBO), a population-based method for optimizing an objective function. In its first phase, several populations evolve independently using movements influenced by a global leader, each agent's…
The article outlines an automated trading approach that uses changes in trading volume and their acceleration as signals of rising or fading participation. It proposes combining those measures with an LSTM forecast of future volume, with ADX or OBV available…
This article describes Traj-LLM, a framework originally designed to predict autonomous vehicle trajectories, and adapts its components to financial time-series forecasting. It combines sparse joint encoding of agent and scene features, high-level interaction…
This article presents a caching architecture for speeding up repeated financial machine-learning computations during feature development, parameter searches, and backtests. It explains why a basic in-memory cache is insufficient for Pandas and NumPy data:…
The article introduces Client, a multivariate forecasting architecture that shifts Transformer attention from relationships across time steps to relationships between variables. It combines this attention block with a linear module intended to capture…
This article explains how regression metrics assess model predictions and how they differ from loss functions used during training. It reviews MAE, MSE, RMSE, R-squared, MAPE, MSPE, and RMSLE, describing their units, sensitivity to large errors,…
The article explains SparseTSF, a compact approach to long-horizon time-series forecasting that assumes a known period in the input data. It splits a sequence into period-aligned subsequences, forecasts those subsequences with shared linear parameters, and…
The article explains Pretrained Decision Transformer (PDT), a reinforcement learning approach that first trains on offline trajectories without reward labels and later adapts to a reward-defined task. During pretraining, an actor predicts actions from past…
The article explains InjectTST, a Transformer method for forecasting multivariate time series that combines independent channel processing with selectively shared global context. Separate channel pathways use a shared encoder, positional information, and…
The article proposes encoding EURUSD market features through an eight-qubit simulated quantum circuit, then using statistics from its measured state distribution as inputs to a CatBoost forecasting workflow. Returns, volatility, and RSI are normalized into…
The document explains Mind Evolutionary Computation (MEC), a population optimizer modeled on competition and information exchange among groups. In the basic method, agents form groups, search around each group’s best candidate, and use group comparisons to…
This article presents an MQL5 routine that takes recent closing prices, ordered with the newest first, and returns a one-step-ahead close forecast. It estimates differencing order with a variance-based heuristic, differences the series, fits ARMA candidates…
The article describes a two-stage forecasting system for financial time series. A linear autoregressive model uses a 25-feature input set, including price lags, transformed values, indicators, and cyclical time features. A U-Transformer then models the…