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 code example demonstrates a data workflow for a FinRL stock-trading project. It fetches a single stock’s price history through both yfinance and FinRL’s Yahoo downloader, then downloads a Dow 30 universe over configured training and trading periods. A…
This retrospective contrasts rule-based stock selection with machine-learning ranking and describes backtesting as a way to evaluate a strategy on historical market data. Its central caution is that a strong fit on a small sample can reflect an irrelevant…
This configuration defines a Qlib experiment that trains an IGMTF model on Alpha360 features for CSI 300 stocks. It uses historical data from 2008 through 2020, with training through 2014, validation in 2015–2016, and a held-out test period beginning in…
This assignment turns a discretionary idea—finding concentrated holdings in recent hot industries—into a proposed equity research process. It suggests first identifying strong sectors with a sector momentum factor, then ranking stocks within those sectors…
This release note describes changes in OctoBot 1.0.2. ChatGPT strategy profiles can now be backtested for some settings using prompts precomputed from historical exchange-pair data downloaded by the service. The ChatGPT profile also shifts from daily trading…
This overview explains how deep learning models can be assembled from input, intermediate, and output layers in a visual strategy-building platform. It surveys layer families including convolution, pooling, recurrent networks, embeddings, noise and dropout,…
The document contrasts selecting stocks by relative rank with selecting them by an absolute model score. A ranking method can always choose the highest-ranked names in a universe, even when their scores are weak. The proposed alternative sets a minimum score…
The article proposes reducing noise in stock prices with wavelet decomposition, fitting ARMA models to the resulting coefficient series, forecasting those coefficients, and reconstructing a one-step-ahead price estimate. Its denoising explanation relies on…
This post describes a revised rolling machine-learning training workflow, reporting that its code was reorganized for clarity, model parameters were adjusted, and memory monitoring was added. The author says the parameter changes increased backtest speed…
The article argues that individual investors should not expect consumer AI tools to compete with professional high-frequency trading. It points to differences in computing location, market data access, and technical resources, and describes an alleged…
This introductory page presents FinRL as a framework for applying deep reinforcement learning to stock trading. It directs readers to a sequence of example notebooks covering data preparation, model training, and backtesting, framing them as a way to follow…
This example organizes daily decisions across leveraged sector and broad-market ETFs using separate AI agents for technology, financials, healthcare, energy, and consumer-related groups. Each sector pod is instructed to consult recent news and macroeconomic…
The document describes how to build daily return data for level-two industries and use it in stock selection. It proposes joining stock industry classifications with daily returns and float market capitalizations, then grouping by industry and date. Each…
This overview explains active learning as a way to reduce the cost of building supervised or semi-supervised models when expert labels are scarce. A model repeatedly identifies candidate examples for human review, incorporates the resulting labels through…
This research summary describes using machine learning to predict equity returns from alpha factors. It compares LASSO, support vector machines, boosted decision trees, and random forests, selecting random forests for their relatively simple structure,…
This document describes an automated trading system framework that combines a two-layer neural network with MACD when assessing whether to open positions. The neural network and indicator signals are analyzed together; MACD is not merely a fallback used only…
The article surveys possible uses of artificial intelligence in crypto trading and decentralized finance. It discusses robo-advisory, automated bots, strategy development and backtesting, risk assessment, arbitrage monitoring, sentiment analysis, predictive…
The document surveys a Chinese securities research team’s work on applying artificial intelligence to quantitative investing. It organizes that research around model evaluation, factor discovery, overfitting controls, synthetic data, and methods intended to…
This document describes an automated trading bot that connects machine-learning models trained in Python and converted to ONNX format. It presents configurable controls for risk-based or fixed lot sizing, order limits, stop loss, and take profit, and notes…
The document gives a brief historical overview of quantitative investing. It describes how advances in computing made it practical to store and process large amounts of historical data, supporting the use of statistical and mathematical models in investment…
The document describes an indicator that applies K-means clustering to historical price points. It groups prices by density and treats each cluster’s center, or centroid, as a potential area of concentrated liquidity. The intended use is to plot support and…
This educational article introduces support vector machines (SVMs) as classification models, with examples framed around separating two classes using features. It explains the maximum-margin objective: choose a decision boundary that stays as far as possible…
This repository overview describes a Python framework for building rule-based strategies, AI-assisted decision systems, and combinations of the two. Its central workflow is to test strategy decisions on historical data, inspect simulated orders and reports,…
This reference describes several candidate fitness objectives for genetic programming that produces equity factors. It defines IC information ratio as the mean information coefficient divided by its standard deviation, with the information coefficient…