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 paper proposes a Transformer model for predicting stock price movements, with enhancements intended to capture both local structure and longer-range relationships in financial time series. A multi-scale Gaussian prior is added to encourage the attention…
This announcement outlines a VeighNa community event focused on applying large language models and machine learning to quantitative research. Its agenda covers a progression from conversational use of language models to prompt design and integration into…
The article argues that greater differences among stocks can improve the opportunity set for systematic equity strategies. When company returns, sectors, and factors diverge, models can rank securities using valuation, momentum, quality, earnings revisions,…
The document outlines a classic radial basis function neural network with a hidden layer of radially symmetric neurons and a linear or sigmoid-style output layer. It describes choosing the output activation from the training target range: hyperbolic tangent…
This guide describes connecting OctoBot to Reddit so it can monitor selected subreddits for posts about traded cryptocurrencies. When its forum evaluator is enabled, OctoBot applies VADER sentiment analysis to posts and creates coin-level summaries for use…
This paper summary describes a reinforcement-learning approach to continuous-time mean-variance portfolio selection. It frames the problem as entropy-regularized relaxed stochastic control, aiming to balance theoretical tractability with practical…
This course-page excerpt outlines a study of stocks that reach their daily upper price limit. It says the material first explains the limit-up mechanism, classifies limit-up stocks, and examines returns after the limit event, then applies a support vector…
This Chinese-language post collects questions about factor research and strategy implementation, but does not provide answers. Topics include adding data factors, scheduling entries and exits, evaluating and combining factors, handling collinearity,…
The document introduces a Nadaraya–Watson kernel regression indicator that applies Gaussian weighting to historical prices. It describes how the method gives more influence to nearby price observations and uses the resulting smoothed estimate to form a…
This configuration describes a Qlib workflow that trains a CatBoost model on Alpha360 features for China’s CSI 500 universe. Its label is a forward close-to-close return, normalized cross-sectionally after missing labels are dropped. The data is divided into…
This document explains how to build fuzzy inference systems in MQL4 with the FuzzyNet library. It covers Mamdani and Sugeno model types, fuzzy input and output variables, membership functions, terms, and rule sets. Mamdani rules map fuzzy conditions to fuzzy…
This research summary examines whether factor crowding measures can help time factor returns or adjust multi-factor portfolio weights in Chinese equities. It constructs eight relative crowding indicators across valuation spreads, pairwise correlations,…
This practice report compares a DeepAlpha deep neural network with LightGBM in a daily A-share stock-selection strategy. Both models use the same 32 short-term price and volume factors and five-year training sample. The portfolio buys the ten highest-ranked…
This experiment runner configures repeated, rolling train-and-test evaluations for LSTM and Temporal Fusion Transformer models on a multi-asset Quandl dataset. It offers variants with different input sequence lengths and optional changepoint feature…
The document discusses the growth of electronic trading and presents algorithmic execution as a tool for handling large orders in volatile markets. Its core methods include splitting a parent order into smaller trades, aligning execution with market-volume…
This platform discussion addresses a reporting difference between older and newer versions of a StockRanker workflow. A user says the older version displayed feature importance and NDCG after a backtest, while those results were not visible in the newer…
This discussion explains how a quantitative platform interprets factor weights and describes its StockRanker as a learning-to-rank model based on a gradient-boosted decision tree structure. The response says the model does not require independent factors,…
This research summary tests whether machine-generated stock-selection factors add value to established manually designed factor libraries. It uses genetic programming to evolve candidate formulas from market data, then applies random forests to turn factor…
This strategy concept uses gradient-boosted regression to score stocks for leadership over a chosen evaluation window. Its proposed label rewards stocks that gain more, experience smaller drawdowns, and break out earlier. A model trained to predict this…
The document reports survey responses about U.S. crypto holders’ views of AI tokens, including expectations for sector performance, long-term optimism, investment interest, and perceived barriers. It describes differences by age and investing experience, and…
This forum post asks how a portfolio-sizing expression allocates cash for purchases and sales. It defines an average cash amount by dividing portfolio value by the holding period, caps buy cash at available cash and a staging-dependent multiple of that…
This short introductory note defines quantitative investing as using numerical and mathematical methods to describe market behavior, identify patterns, and inform investment decisions. It presents the approach as an attempt to make decisions more objective…
This research summary compares five classifiers for predicting stock returns from historical equity factors: multinomial logistic regression, linear support vector machines, random forests, XGBoost, and deep neural networks. It distinguishes the linear…
The Sampler indicator is described as a way to derive training targets from future price changes. One output is an analog value normalized between negative and positive one, representing the balance of upward and downward price movement over a specified…