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 document outlines a book about learning neural networks and incorporating them into algorithmic trading systems built with MQL5. Its stated coverage includes machine learning fundamentals, convolutional and recurrent networks, more complex architectures,…
This document explains how to install Qlib, a Python library used in quantitative research. It outlines two routes: installing the published package with pip, or obtaining the source and running its setup process after installing prerequisite packages. It…
This source code implements components of a Temporal Fusion Transformer, a neural network architecture for time-series forecasting. The visible sections define feed-forward layers, gated linear units, gated residual networks, skip connections with layer…
The article evaluates a deep-learning stock-ranking model across broad-market and filtered Chinese equity universes. It uses rolling windows of three years for training and one year for evaluation, then compares portfolio sizes, rebalance intervals,…
The document compares machine-learning approaches to strategy development with more subjective manual parameter tuning. It recommends measuring overfitting by comparing performance on training and test data, using cross-validation where appropriate. For…
A brief troubleshooting note explains a model training error in a stock ranking workflow. The stated cause is that all input factor values were either zero or one. Since the ranking model needs variation in its inputs to distinguish and order securities, a…
This note describes two stock selection approaches, XGBoost and StockRank, using the same five inputs: ranked dividend yield, market capitalization, closing price, a 20-period price ratio, and average turnover. It also gives example hyperparameter…
The report describes a sector-rotation framework that pairs quantitative clustering with macroeconomic reasoning. It first groups industries using features such as returns across market conditions, return on equity, and revenue growth, then makes limited…
The document explains FinRL as a modular framework organized into market environments, deep reinforcement learning agents, and trading applications. The environment layer supplies interfaces to the agent layer; agents interact with simulated markets through…
This BigQuant forum thread discusses whether rolling parameter optimization can run in simulated trading and whether newly optimized parameters are automatically applied to a strategy. Replies say that rolling optimization can work in simulated trading,…
The author describes a stock reversal approach based on a prior limit-up move followed by a pullback, with a bullish candle engulfing a bearish one as a defining pattern. A proposed variation focuses on a leading stock's first bearish session and subsequent…
This document introduces a lecture on deep reinforcement learning by researcher Matteo Hessel. It says the session discusses practical considerations and algorithms, along with implementing them using automatic differentiation tools such as JAX. The subject…
This Expert Advisor pairs the Larry Williams Outside Bar setup with an ONNX machine-learning filter. At each new bar, it identifies a candle whose high exceeds the prior candle’s high and whose low falls below the prior low. It then supplies the model with…
This Q&A compilation collects practitioner questions about developing, combining, and deploying quantitative strategies. Topics include variation between XGBoost rolling-training backtests, selecting and evaluating factors and models, overlapping holdings…
This configuration defines a Qlib workflow that trains an LSTM on Alpha360 features to rank CSI 300 stocks. It normalizes feature data with robust z-scores, fills missing feature values, drops missing labels, and cross-sectionally ranks labels. The…
This Qlib configuration sets up a time-series forecasting and portfolio backtest workflow for the CSI 300 universe, using the Shanghai-Shenzhen 300 index as its benchmark. The dataset uses Alpha158 features, applies robust feature normalization and…
This Qlib configuration defines a Chinese-equity ranking workflow using the Alpha158 feature handler and a ridge linear model. It assigns CSI 300 instruments and the related index benchmark, with historical data split into training, validation, and test…
This tutorial explains the skip-gram model as a neural network trained to predict a nearby word from a chosen input word. Training examples are word pairs drawn from a configurable context window. The input word is represented as a one-hot vector, and a…
The author describes why converting discretionary trading into fixed rules can be difficult: decisions may combine many implicit features and weights, and traders adapt as market conditions change. The post compares this challenge to recognizing a familiar…
This indicator applies a nearest-neighbor method to Change of Character (CHoCH) breaks in market structure. It detects swings with centered pivots, records a break when price crosses the last relevant swing in a direction that changes the trend state, and…
The document gives a brief overview of artificial intelligence, from the field’s origins to its wider adoption as data, computing hardware, and algorithms advanced. It describes AI as systems that model aspects of human reasoning and notes its use in…
The article outlines a workflow for reconstructing and simulating a Chinese public fund’s equity holdings. It scrapes disclosed stock codes and portfolio weights for a China Merchants CSI 300 enhanced index fund, then assumes quarterly rebalancing: the…
This note warns that several strategies in a folder contain lookahead bias and presents them as exercises for identifying the problem. It points to normalization procedures that calculate minimum and maximum values using an entire dataset. When those…
This Freqtrade strategy creates long entry and exit signals from ratios among three simple moving averages. It computes separate buy-side and sell-side averages, then checks whether adjacent-average ratios fall between configurable minimum and maximum…