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.
3,481 documents
This document introduces two self-balancing binary search trees and explains how their balancing rules support efficient lookup, insertion, and deletion. For AVL trees, it defines node height and balance factor, then describes single and double rotations and…
The article describes an architecture that combines iTick market feeds, other economic and alternative data, and DeepSeek-R1 to support quantitative research workflows. It proposes attention-based data fusion, automated strategy variant generation,…
The document summarizes U.S. retirement accounts, especially 401(k) plans and IRAs, as flexible account structures that offer a broad range of investments. It reports that mutual funds represent about half of assets and describes equity funds as the largest…
The document considers whether quantitative trading in China has a durable future as the field becomes more crowded. It argues that more participants using similar strategies can compress the limited excess returns available in an A-share market that is not…
This strategy ranks constituents of the CSI Dividend Index using a combined score based on dividend yield and return on equity (ROE). Dividend yield represents distributions relative to share price, while ROE is used as a measure of how effectively a company…
This report introduces intraday high-frequency trading in the Chinese securities market, framing it as an attempt to capture short-lived market inefficiencies caused by investor behavior or delayed reactions to information. It defines the strategies under…
This report studies whether rotating among industry and thematic exchange-traded funds can support an absolute-return portfolio. It selects 14 industries with matching funds from a broader set of Chinese industry groups, then compares a rotation portfolio…
This research report explains reinforcement learning concepts, including Markov decision processes, state and action values, reward design, and value-based and policy-based algorithms. Its applied example uses a deep Q-network to time long positions in the…
This broad practical guide introduces machine learning in quantitative investing, covering supervised, unsupervised, and reinforcement learning. It explains data preparation, missing values, categorical features, scaling, train and test splits,…
This brief indicator note defines the moving average, abbreviated MA, as a calculation using closing prices and a lag or lookback parameter. It gives a compact formula reference for producing an MA value from those inputs. The entry identifies the indicator…
The document describes a Chinese stock-selection example that represents daily buy signals as a factor value of one and sell signals as zero. It proposes processing those values and using them to filter and rank stocks. The stated data source is daily stock…
This assignment outlines a workflow for creating a visual machine learning stock strategy using BigQuant’s StockRanker course and strategy template. The participant is asked to modify the model’s input features by adding personal factors, then backtest the…
The post describes a feature-data issue in a quantitative research workflow. A request failed with an error indicating that no observations remained after missing values were removed. Changing the feature lookback period, as suggested by common…
The document introduces convex functions through their geometric shape and explains a first-order characterization for differentiable convex functions on an open convex domain. At any point, the function’s value at another point is at least as large as the…
The document examines how benchmark choice affects mutual fund performance estimates. It contrasts return-based evaluation, which is sensitive to benchmark selection, with holdings-based evaluation, which can be more precise but depends on infrequently…
This discretionary trend-following framework aims to join an established advance rather than predict a market bottom. It lists three possible entry cues: a bullish engulfing candle after a pullback to support, a retest near the short moving-average cluster…
This report outlines the first step in a multi-factor model research process: selecting which factors to combine. It proposes using a hedging perspective and benchmark factor exposures as part of the screening criteria, with the stated aim of reducing the…
This brief platform report describes a suspected stock-price data problem in a backtest using TWAP-based adjustment prices. A newly added position in Ping An Bank appeared to generate a sudden large profit because its recorded average holding price was…
This research summary presents labor leverage as a company characteristic that may help explain differences in expected stock returns. It defines labor leverage through the relative size and rigidity of labor expenses, which shape how operating profits…
This report summary presents an investment case for Chinese technology equities and the China Securities Technology 50 Strategy Index. It argues that sharp market declines may offer opportunities to add exposure, based on a claimed review of 13 major…
This tutorial outlines a workflow for modeling Chinese steel futures using high frequency market data. It starts with tick snapshots for the RB2305 contract, explains that futures may trade during night sessions, and resamples the raw observations into ten…
This introductory assignment defines quantitative investing as using computer programs and historical market data to identify assets and apply predefined buying and selling rules. It highlights the ability to scan a broad universe more quickly than a human…
This study examines whether active U.S. equity mutual funds that increase portfolio concentration subsequently achieve better risk-adjusted performance. Its rationale is a tradeoff: concentrating holdings can give greater weight to the manager’s strongest…
This document describes a competition built around Deep Alpha-CNN, an equity prediction model using a seven-layer one-dimensional convolutional neural network with residual connections. The report’s stated findings favor the deeper network over a two-layer…