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 Quino Channel indicator estimates horizontal support and resistance from the distribution of closing prices over a chosen history. It sorts closes into equal-width price classes and builds a frequency histogram; classes with prominent peaks represent…
This stock screen combines three conditions: daily high-to-low amplitude above one percent, a newly formed KDJ crossover, and a positive return relative to the prior close. The article treats a larger daily range as a sign of opportunity, the crossover as a…
This stock-selection idea screens for companies in the Metaverse concept group, with positive net buying attributed to major participants during the auction and a circulating share count no greater than 5.5 billion. The document frames concept membership as…
The document describes an RSI variant that first smooths price with a configurable moving average, then measures changes across a specified step interval. It sums positive and negative changes and uses those components to calculate RSI. Users can set the RSI…
The document presents a cross-sectional stock factor built from daily Chinese stock data. It calculates a rolling correlation over 20 observations between the close-to-close price ratio and the logarithm of the volume ratio plus one, then reverses the…
The document describes a minimal expert advisor that opens market orders in a direction chosen by a random-number generator. Its stated principle is equivalent to choosing a side by a coin flip. This makes the strategy a simple example of randomized trade…
The article questions the use of deep learning to forecast stock or futures returns from historical indicators. Its author says they tried both classification and regression approaches, including recurrent networks, and found the results poor. They also…
The document explains a Volume Spread Analysis indicator that compares trading effort, represented by volume, with price result, represented by bar range. It ranks both measures against a rolling window of 100 bars, converts the ranks into deciles, and…
This research summary examines why Chinese A-share returns show short-term reversal even though medium- and long-term raw returns do not show the momentum commonly reported in overseas markets. It decomposes stock returns into overnight returns and intraday…
The article reviews research comparing factor momentum with industry and stock momentum using historical U.S. equity data and a broad set of return factors. The central finding is that industry-neutral factor momentum contains the predictive information…
The document describes a script for day traders that examines a strategy’s historical results by time of day and day of week. It proposes using win–loss and reward–risk ratios to identify the periods when the strategy has performed best, comparing the…
This collection presents statistics on cryptocurrency ownership, trading activity, investor concerns, geographic adoption, and industry structure as of 2024. It reports estimates of market value and trading volume, survey findings on ownership and investor…
The screen selects stocks with turnover rates between 3% and 12%, a rising KDJ K value below 100, and a listing date earlier than a chosen cutoff. The KDJ condition is implemented by checking whether the current K value exceeds the prior value, while the…
The document summarizes Barra CNE6 as a risk and factor model with an estimation universe filtered for size and liquidity. It describes three versions: two long-term variants using slower style factors and a trading version combining slower and faster…
This document describes a Chinese equity screen requiring daily turnover between 3% and 12%, year-over-year growth in net profit attributable to parent-company shareholders above 20% and no greater than 100%, and a positive return from the day’s open. It…
This document describes a Chinese equity screen combining daily turnover, listing age, price movement, and large-order net volume. It selects stocks with turnover between 3% and 12%, listed in 2021, whose daily return multiplied by a large-order net-volume…
The document reviews limitations of Bollinger Bands, which place standard-deviation ranges above and below a simple moving average. It challenges the probabilistic interpretation of those bands: observed prices do not necessarily fall within the bands at…
The document introduces a running median indicator for MetaTrader 4. It explains the median as the middle value of an ordered sample and presents it as a smoothing method that is less affected by outliers than the mean. The indicator uses an odd-sized span…
This reference surveys machine-learning methods and relates them to investment research tasks. It covers linear and logistic regression, naive Bayes, nearest neighbors, support vector machines, decision trees and ensembles, neural networks, sequence models,…
The report summarizes research on whether the asymmetry of stock returns predicts expected returns, a relationship for which prior findings have conflicted. It presents two distribution-based measures beyond conventional skewness: the difference between the…
This guide describes an FMZ tool for evaluating trading ideas with time-series expressions modeled partly on publicly available Alpha101 calculations. It lists arithmetic and logical operators, transformations such as moving averages and differences,…
This 2018 analysis assesses a sharp decline in Chinese equities and treats the U.S. market as the main near-term risk trigger. It proposes two possible U.S. paths: continued weakness without a meaningful rebound, or stabilization followed by months of…
This reference introduces NumPy as a Python library for array-based numerical work and explains why it is useful for handling financial data. It describes multidimensional arrays, broadcasting, optimized computation, linear algebra, and random number…
The document introduces a variant of Williams’ Percent Range that uses the difference between the current high and low in its calculation. It defines the indicator from that high-low value relative to the highest and lowest prices over a configurable…