Knowledge library

This article proposes ListFold, a listwise learning-to-rank loss for factor-based long-short equity strategies. Its central design treats the top and bottom of a predicted stock ranking as equally important, so one coherent ordering can select both long and…

This short commentary describes pressures facing quantitative investment firms: crowded strategy areas, convergence among approaches, thin profits, and greater barriers to entry for researchers. It argues that widespread use of machine learning can produce…

This article presents a four-stage account of quantitative trading. It describes early strategies built on financial theory, multi-factor stock selection, and linear regression; a later high-frequency phase centered on low-latency hardware and algorithms; an…

This overview explains how unsupervised learning finds structure in unlabeled data, contrasting it with supervised prediction. It presents K-means clustering and principal component analysis (PCA) through equity examples. K-means groups stocks using scaled…

This module fits a bivariate mixture of Clayton, Student-t, and Gumbel copulas, motivated by a mixed-copula pairs trading approach. It first maps each input series to empirical cumulative probabilities, then estimates component parameters and mixture weights…

This strategy turns changes in a spread series into long and short entry thresholds. It separates historical spread changes into positive and negative values, then calculates a chosen upper quantile of positive changes and a lower quantile of negative…

This report adapts hidden Markov models (HMMs), a technique associated with speech recognition, to rank stocks by the likelihood of an upward move. It assumes that rising and falling stocks follow distinct patterns and models those patterns separately. Six…

This career overview explains the work of quantitative traders and the mix of skills commonly needed to research and automate trading strategies. Quants analyze market data, develop models, test ideas, and implement trades using computer programs. The…

This research summary presents a framework for timing equity factors whose returns have become less reliable. It reviews indicators such as valuation spreads and pairwise correlations, and examines their relationship with future factor returns. A random…

This document outlines a long-horizon commodity futures strategy that uses machine learning to forecast the following week’s direction. It calculates signals at the end of each week and adjusts positions on the first trading day of the next week using TWAP…

This survey introduces regression methods used in financial modeling, from ordinary linear models to polynomial and logistic regression. It also explains quantile regression for examining different parts of a conditional outcome distribution, including its…

This practitioner note compares very high-return, moderate-return, and lower-return backtests for short-horizon equity models, focusing on whether their apparent performance persists out of sample. The author reports that aggressive timing filters and models…

This document introduces the goal of deriving gradients for Word2Vec’s cost function with respect to a target vector’s weights. It says the derivation uses a set of expressions and formulas, then works through the gradient calculation step by step. The…

This short forum exchange discusses how to construct market risk-control indicators, particularly for broad-market timing. It says such approaches commonly use an index together with custom technical indicators, such as MACD, to guide decisions. It also…

The article describes a risk-model framework for evaluating fund managers beyond headline returns, Sharpe ratio, or drawdown. It decomposes active returns into factor and stock-specific contributions, and examines risk contributions from a fund’s style…

The document describes an intraday strategy that combines classification and regression models built with XGBoost to rank stocks by predicted daily price change. Its features include overnight returns and opening-auction behavior, such as staged price moves,…

This article collects practical lessons from building supervised machine learning strategies for stock selection. It recommends keeping training and test data separate and splitting financial time series chronologically, since random splits can mix market…

This overview proposes using machine learning to support value investing by combining financial statements, market data, and macroeconomic information. It lists a broad set of candidate measures, spanning performance and risk statistics, valuation and…

This forum post describes a Chinese equity backtest comparison between standard regression models and a platform stock ranking model. The author trained XGBoost, LightGBM, and CatBoost regressors on five-day returns using rolling historical periods, then…

The Edge Drift Detector compares closing-deal outcomes from a live Expert Advisor with a backtest baseline to assess whether average per-deal performance has weakened. Its central method is a cumulative-sum alarm calibrated by resampling baseline deals, with…

This configuration defines a Qlib workflow for ranking CSI 500 equities with a LightGBM model and Alpha360 features. It uses cross-sectional rank normalization for labels and trains on data from 2008 through 2014, validates on 2015–2016, and evaluates a…

This Qlib configuration defines a Chinese equity research workflow using the CSI 300 universe and Alpha158 features. It trains a double ensemble of gradient-boosted models, with both sample reweighting and feature selection enabled, then evaluates signals…

The article compares a StockRanker stock-selection strategy with a Deepalpha deep neural network using the same 17 short-term factors, training data, and backtest setup. The strategy selects the top-ranked A-share stock each trading day and sells it the…

The document introduces generative adversarial networks (GANs), a framework that trains two models against each other. A generator creates samples intended to resemble the training data, while a discriminator estimates whether a sample is real or generated.…