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
This tutorial walks through assembling a quantitative equity research workflow with Qlib. It covers retrieving and inspecting market data, interpreting adjusted prices, working with dynamic universes and point-in-time fundamentals, and constructing features.…
This brief troubleshooting note addresses an error in a stock LSTM strategy template involving the training epoch parameter. It advises changing the model-fitting argument from the older `nb_epoch` name to `epochs`, while leaving the batch size and…
The document walks through a basic LSTM workflow for forecasting Bitcoin prices from daily market data. It loads open, high, low, close, volume, and weighted-price fields; checks for missing and zero values; replaces zeros with missing values and carries the…
The document outlines an adaptive expert advisor that records market patterns and evaluates them through simulated orders with stop-loss and take-profit levels. Its stated process is to collect examples first, then use the observed outcomes to select…
The document describes a framework for computing quantitative stock factors as recursive transformations of numerical indicators. It says factor logic can be defined with programming languages or XML: procedural languages suit more complex structures, while…
The author describes a China A-share swing strategy that first narrows the universe to selected high-quality companies and blue chips, then uses a machine-learning model to identify timing patterns. The training labels are based on relative performance…
This article explains a structure inference network (SIN) that augments object detection with scene context and relationships among candidate objects. It represents a scene with global image features, candidate regions as graph nodes, and pairwise…
The document describes a MetaTrader expert advisor that combines a simple neural network with the Laguerre_PlusDi and SilverTrend_Signal indicators. Its notable operational feature is a configurable daily trading window with start and end times. It outlines…
This overview explains a supervised learning workflow for ranking stocks by their expected future returns. It introduces features, a fitted function, model parameters, and a loss function, then describes labeling return targets, splitting observations into…
This Chinese-language post proposes a stock screen combining price amplitude, a rounded price pattern, and sustained return on equity above a stated threshold. Its example formula adds further technical conditions: recent price position within a range, a…
This example builds a macro ETF portfolio through several AI roles: agents assess growth, inflation and rates, debt and liquidity, and challenge one another’s conclusions. A trading agent then combines their views into a diversified basket drawn from equity,…
This Chinese-language forum post asks how BigQuant’s StockRanker model works and whether its gradient-boosted decision tree component could be replaced with another ensemble method, such as XGBoost, while retaining a similar ranking strategy. It describes…
This article outlines a high-frequency strategy research framework for China’s order-driven markets, using Level 1 order book data and support vector machines (SVMs). It proposes extracting features such as best bid and ask prices and sizes, derived depth…
This FAQ explains implementation details for quantitative strategies built with FMZ workflows. It covers supported runtime requirements, JavaScript code nodes, sequential node execution, trigger timing, passing outputs between connected nodes, and using…
The document describes a Chinese equity screening rule that combines a 14-period RSI below 65, the product of daily percentage change and super-large-order net inflow, and large-order net inflow above 0.05 over three consecutive days. Its stated aim is to…
This example builds a daily direction classifier for a rubber futures contract. Near a scheduled end-of-day cutoff, it uses recent closing prices to calculate three technical features, aligns each historical feature row with the following session's…
This Qlib configuration describes a machine-learning workflow for ranking CSI 300 stocks. It uses the Alpha360 data handler and a double-ensemble model built from gradient-boosted trees. The configured label is a forward close-price return, while the…
The author investigates whether the highest-ranked stocks from a convolutional neural network strategy deliver the best results. The experiments apply a model trained on a small number of stocks to broader selections, then backtest separate groups of ranked…
This strategy-sharing article describes an active China A-share approach that blends model-based stock ranking with technical timing. It starts from a manually selected China Securities 150-related universe, described as roughly 100–300 larger, liquid…
This architecture overview explains how OctoBot organizes a modular crypto trading bot. Framework packages define shared contracts for data flow, evaluation, trading, and backtesting, while strategy logic, evaluators, and exchange connectors are supplied as…
This Chinese-language post describes an optimization of a machine-learning strategy template for the BigQuant platform. The author reports reducing a two-year backtest from 65 minutes to 26 seconds by using parallel computation and delayed execution. The…
This learning note outlines a stock-price forecasting pipeline that uses an LSTM as a sequence generator and a CNN as a discriminator in a generative adversarial network. It describes a broad feature set: related assets and rates, technical indicators, news…
The document explains the expected input, initial state, and output shapes for PyTorch’s LSTM layer. With the default layout, inputs are arranged as sequence length, batch size, and feature count; the hidden and cell states also include layer and direction…
This introductory explanation builds from the Markov chain assumption: the next state depends on the current state rather than the full history. It then describes a hidden Markov model (HMM), where observable data are probabilistically linked to an…