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
This introductory article presents ten practical points about machine learning for non-specialists. It emphasizes that machine learning finds patterns in data, so data quality, accurate labels, representative samples, and careful transformations often matter…
The post asks how to define groups for a learning-to-rank model in a stock selection setting. In ranking algorithms, each group contains observations that should be ranked against one another; the author proposes grouping stock samples by trading date so…
This document describes a trading-platform indicator for displaying statistics from closed trades. Users can optionally filter the history by a trade's magic number, allowing results to be isolated by strategy or identifier. A magic number can be entered…
This document describes a MetaTrader indicator that identifies bullish and bearish pinbar candlestick patterns using price action, without relying on standard technical indicators. It marks bullish patterns below their candles and bearish patterns above…
This brief entry describes a technical indicator intended to display the relative power of bullish and bearish market forces. It identifies the indicator as an MQL4-origin work later associated with a MetaTrader 5 implementation, and notes that the…
The document describes a stock screen for main-board companies with turnover between 3% and 12%, circulating market value between 5 billion and 10 billion yuan, and a positive price-to-earnings ratio. It frames the turnover and size ranges as initial…
This proposed A-share stock screen combines three conditions: daily amplitude above 1%, MACD above its zero line, and a product of price change and a large-order net-volume measure above a threshold. Its rationale is to combine a volatility filter, a…
This post describes a Chinese equity screen that selects stocks with turnover between 3% and 12%, stock codes beginning with 60, and a positive recent change in institutional ownership. Its stated rationale is that rising institutional holdings may indicate…
This Chinese-language post proposes screening listed Chinese stocks using price movement, the presence of an outstanding convertible bond name, and a minimum listing age. It explains the intended rationale as combining trading activity with indicators that…
This note describes a Swami-style stochastic indicator associated with John Ehlers’s discussion in a March 2012 issue of TASC. The indicator displays colored zones derived from stochastic calculations and adds a generic stochastic value, described as an…
This indicator gathers values from up to ten combinations of symbols, timeframes, and signal variants, synchronizing them to the symbol on which the indicator is installed. Its original purpose is to provide normalized inputs to a neural network, though the…
FactorVAE is described as a probabilistic dynamic factor model that applies a variational autoencoder to noisy market data to forecast cross-sectional stock returns. In training, an encoder-decoder uses future returns to identify posterior factors that…
FactorVAE is presented as a probabilistic dynamic factor model that uses a variational autoencoder to extract latent factors from noisy market observations and forecast cross-sectional stock returns. During training, an encoder and decoder use realized…
The document motivates adapting forecasting models to changing market conditions: financial data distributions can shift over time, so models trained on earlier periods may lose predictive strength. It compares two approaches, RR and DDG-DA, using linear and…
This Chinese-language strategy note proposes screening equities with turnover between 3% and 12%, seven consecutive declining sessions, and a shareholder concentration measure between 20% and 70%. It provides example expressions for a charting platform and a…
This indicator presents a configurable period’s standard deviation of closing prices as a histogram. Each bar is colored according to whether the calculated deviation is higher or lower than its value on the previous bar. The example uses a default lookback…
This extended Fibonacci channel indicator lets users choose how to define the reference distance that represents the full channel width on each bar. The available methods are the maximum, mean, or median price displacement from a moving average, or the…
This introductory article explains Matplotlib’s role in Python data visualization and outlines the Figure and Axes objects, along with its state-based pyplot and object-oriented interfaces. It notes that the library supports chart types such as lines,…
This indicator concept estimates seller pressure as the distance from a bar’s high to its close, and buyer pressure from its low to its close, with an option to take the absolute value of the buyer measure. It smooths these series using a configurable moving…
This Chinese-language post proposes screening metaverse-related stocks using the previous day's actual turnover rate, with a stated range above 3% and below 28%, and a 14-period RSI below 65. The stated rationale is to focus on stocks with some trading…
The report develops stock selection factors from abnormal operating cash flow as proxies for real earnings management. Its premise is that firms can alter real business activity to make reported performance look stronger, potentially harming future…
This research summary discusses how asset prices can reflect economic expectations before official data is released. It proposes comparing year-over-year changes in a basket of asset prices with economic indicators, and comparing asset-price changes with…
The Market Cycle Projection Engine turns a Wyckoff-inspired cycle model into a five-state indicator: Accumulation, Markup, Distribution, Markdown, and Transition. It classifies bars using normalized linear-regression slope, the ratio of fast to slow ATR, and…
This trading exercise translates a discretionary method into explicit stock-selection rules. A candidate must close above its five-day moving average for three consecutive days and remain within ten percent above its lowest price over the previous twenty…