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.
21,023 documents
The document briefly introduces a multi-factor investment platform, focusing on how to estimate stock and portfolio factor exposures. Stock exposures are derived from database values such as price-to-earnings and price-to-book ratios, then cleaned and…
This document presents an implementation of John Ehlers’ Adaptive Commodity Channel Index. It estimates a dominant cycle period from price data by smoothing prices, deriving in-phase and quadrature components, and applying a homodyne discriminator. The…
This document presents a QuantStats tear sheet for a strategy labeled “buffett-plain,” compared with SPY over January 4–15, 2026. It lists returns, drawdowns, risk-adjusted statistics, market exposure, daily outcomes, and two drawdown episodes. The reported…
This report compares hourly Bitcoin returns, volatility, and trading volume across Binance, Bybit, Coinbase, HTX, Kraken, and OKX. It groups activity by the trading hours of several regions, using standard deviation of hourly returns as a volatility proxy…
This weekly competition report presents diagnostic summaries for the ten highest-ranked submitted factors and tracks a China 1000 index-enhancement portfolio. Factor diagnostics include average exposures to BARRA style factors and sector-level average Rank…
This introductory overview explains quantitative trading as the conversion of market observations and decision rules into mathematical models and computer programs. It outlines a typical process: analyze historical price, volume, and timing data; design a…
This note proposes a simple screen for Chinese stocks associated with the metaverse theme. It selects companies with a circulating share count no greater than 5.5 billion and a share price around 18.5 yuan; the sample code uses a price interval from 18 to 19…
This note describes a Chinese equity screen for stocks associated with the metaverse theme. It combines positive institutional activity with a ranking by capital strength, which the article also describes as selecting larger stocks. The proposed process…
This tutorial describes assembling a matrix of daily closing prices for Hong Kong stocks using BigQuant data sources and pandas. It first retrieves unique instrument identifiers, then loops through them to collect date and close fields over a specified…
This example implements a daily mean-reversion strategy for a Shanghai Futures Exchange gold contract. It calculates a Z-score from recent closing prices, enters long when the score falls below a negative entry threshold and short when it rises above a…
This document presents a simplified ZigZag indicator design that tracks alternating price highs and lows without arrays. It maintains the current extreme, then confirms a reversal after price moves away from that extreme by a configurable threshold. The…
This indicator plots separate volatility measures for bullish and bearish price movement. For the bearish line, it measures the prior close’s distance from the current low; for the bullish line, it measures the current high’s distance from the prior close.…
This indicator displays Stochastic Oscillator readings for a user-selected period across timeframes in a text block. Users can adjust the block's size and font, and hide timeframes that are not in use. The display supports up to ten indicators at once. Its…
The document introduces reinforcement learning through the interaction of an agent and an environment, where actions produce feedback and the objective is to maximize discounted future rewards. It frames the problem as a Markov decision process, explaining…
A BigQuant forum question asks how to select the ten stocks with the strongest return over the previous calendar year on the first trading day of a new year. The user’s current approximation compares prices about 250 trading days apart, but they want a…
A BigQuant forum exchange discusses why feature extraction can return multiple records for a stock on a given day, even when one record is expected. The suggested cause is that the feature input stage reads across tables; joining data with multiple matching…
This brief introduction contrasts linear regression of factor data with decision trees. Its central point is that market returns may not relate to factors in a simple linear way, so a tree-based method is worth considering when studying how factors…
The document outlines a Breakout RSI (BRSI), an oscillator inspired by a published technical-analysis article. It names three inputs: the calculation period and overbought and oversold levels. Its calculation resembles the relative-strength index, using…
The document explains a WILL VAL oscillator attributed to Larry Williams. It measures the difference between two exponential moving averages applied to a relative price series, then rescales that difference against its recent maximum and minimum. The…
The document explains how BigQuant’s factor research platform handles user-created factors and downstream dependencies. A researcher can combine platform factors A and B into a derived factor C, submit C to the platform, and let other users with access clone…
This research summary compares common stock-selection factors in the U.S. and Chinese markets, with a test of seven factor groups and 24 signals in the Russell 3000 universe. It reports that profitability, valuation, and growth measures were among the…
This Chinese stock-screening note combines a price-amplitude filter with a volume and turnover condition. It describes selecting shares whose amplitude exceeds 1 and whose prior-day turnover multiplied by the ratio of current auction volume to prior-day…
This stock-selection recipe begins with shares whose 9:25 a.m. price gain is below 6%, circulating market capitalization is above 10 billion yuan, and volume ratio ranks among the top 100. The explanation frames the volume measure as a way to find actively…
The document explains the Hull Moving Average as a method intended to make a moving average more responsive while retaining smoothness. Its calculation applies a weighted moving average to a weighted combination: twice the half-period average minus the…