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 document describes a demonstration of a backpropagation neural network (BPNN) for training and testing time-series predictions in MQL5. It says the implementation was ported from a C++ library and is presented as a coding demo rather than a real-world…
The article introduces least squares as a way to fit an empirical relationship to observed data. It illustrates a straight-line model relating ship length to width, first using two observations to determine a line and then explaining why that line need not…
This research summary examines whether intraday stock price behavior predicts future returns. It uses five-minute data and calculates volatility, skewness, and kurtosis from returns residualized with a Fama-French regression to reduce market and style…
This document describes a price-difference indicator that compares current prices with prices from a selected number of periods earlier. It offers two calculation modes: one based on the difference between open and close, and another based on the difference…
This article introduces unsupervised clustering and contrasts it with supervised classification: cluster labels are not known in advance, so judging whether a grouping is correct is difficult. The goal is to group observations so that members of each cluster…
This note describes a chart indicator that marks local price highs and lows using a single period setting. An upper extremum is identified when the prior bar’s high reaches the maximum for the calculation window and exceeds the current bar’s high; a lower…
This document describes an expert advisor that compares price changes across two adjacent historical windows. It calculates the start-to-end price difference within each window, compares the resulting values, and uses their relative size to predict the…
This research summary describes stock-selection factors derived from tick-level trades. Trades are aggregated to minute intervals to estimate active buy and active sell amounts, which are used to form active-buy share and active-buy intensity measures. It…
This indicator description explains how least-squares regression can create price channels from recent highs and lows. For a selected lookback period, it fits separate lines to the high and low observations, balancing the distances of the points around each…
The document presents an extended version of a smoothed momentum MACD indicator with floating levels. It describes four ways the indicator can change color: when price crosses outer levels, crosses a middle level, changes slope, or crosses the zero level.…
This reference describes a collection of statistical functions for analyzing time series. The listed tools cover basic summaries such as means and standard deviations, relationships between series through correlation and linear regression, and…
The document summarizes research on whether financial data vendors act as information intermediaries. The study examines quarterly earnings announcements and the timing of their distribution by First Call, comparing immediate dissemination with delays of at…
The document summarizes research on whether financial data vendors act as information intermediaries. The study examines quarterly earnings announcements and the timing of their distribution by First Call, comparing immediate dissemination with delays of at…
The Congestion Identifier (CI) measures how many bars in a chosen lookback period closed within a price corridor around the current bar’s close. The corridor is set by adding and subtracting a configurable point distance, Delta. Its output is the count of…
The article summarizes research on how trading volume interacts with equity mispricing. Using monthly US stock portfolios sorted jointly by a mispricing score and turnover, the study finds that volume and subsequent returns move in opposite directions…
This text compares two approaches to changing trade size after wins and losses. Under the classical Martingale approach, a trader begins with a minimum position and increases size after each loss, aiming for a later winning trade to recover the losses in the…
The document introduces an indicator that displays a chart’s price in normalized form. It suggests looking for divergence between the normalized series and market behavior, and placing normalized series for several currency pairs in the same window to…
The document describes a feature for counting how many limit-up sessions occurred in a recent period. Its example creates a five-session count by checking each of five lagged limit-status fields and adding one for every session marked with the limit-up…
The report reviews two major historical gold rallies and argues that the drivers of gold prices vary over time. It examines two short-term and four long-term factors, using rolling correlations to show that relationships between these factors and gold are…
This essay contrasts two views of investment risk. Modern portfolio theory uses volatility, standard deviation, and beta as measurable proxies, treating diversifiable company-specific risk separately from market-wide risk. The value investing perspective…
The article presents parameter-plane analysis as a way to visualize how combinations of strategy settings affect backtest results. Its example is a Chinese equities strategy combining valuation and float market capitalization ranks. It varies the number of…
This brief forum exchange concerns how to use the previous day’s moving average in a quantitative feature expression. The original question compares a five-day mean of closing prices with a twenty-day mean and asks how to shift the five-day mean back by one…
This note outlines the Abstract Factory software design pattern, which creates families of related products through a common factory interface. It contrasts that role with Builder: a builder assembles a complex object through multiple steps and returns it…
Track Cyclical is an oscillator that expresses the applied price’s percentage deviation from a moving average. Its calculation compares the selected price with a simple moving average over a configurable period, divides the difference by that average, and…