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 presents an implementation of John Ehlers’ Smoothed Relative Strength Index. It first applies a short weighted moving average to closing prices, giving more weight to the middle observations. Across a chosen lookback period, it separately totals…
The document examines how one-minute OHLC backtests can misclassify bracketed trades when a bar touches both the stop and target. Because OHLC records prices but not their sequence within the bar, a tester must assume an intrabar path; real tick data can…
This assignment analyzes how a stock selection strategy’s exposure to market styles changes over time, then outlines a framework for rotating among factor strategies. The initial analysis uses a small-capitalization selection factor and compares strategy…
The document defines a Butterworth Moving Average indicator as a smoothed moving average built with a Butterworth filter. It identifies two configurable inputs: the calculation period and the applied price. The applied price is first represented as a…
This brief description explains an indicator display that plots a selected indicator and its moving average in a separate chart window. The moving average can use any method, and the indicator line can optionally be shown together with that average. It also…
This stock-selection rule looks for shares with turnover between 3% and 12%, excludes Beijing-listed A-shares, and requires the day’s increase in position share to exceed 5%. The explanation treats the turnover band as a way to select reasonably active and…
The document summarizes research on shared movements across commodity markets. It reports that commodity return fluctuations often move together and that a common factor explains a substantial portion of their variation. The factor is closely associated with…
The document outlines a 30-day Trend Intensity Index calculation attributed to M. H. Pee. First, compute a 60-day simple moving average from closing prices. Then compare each of the latest 30 closes with that average and record the size of each deviation,…
The document describes a simple random-entry trading experiment based on the assumption that short-term upward and downward moves are equally likely. A script randomly opens a buy or sell order, while the user sets trade size and exit levels. With equally…
This article examines how the length of a training window affects an AI stock selection model. It compares long historical samples with shorter rolling samples and recommends checking each choice against a held-out backtest period. In the author's…
The article reviews research on salience theory in equity pricing. Investors are proposed to overweight unusually prominent past returns relative to the market: strong standout gains can attract demand and raise prices, while conspicuous declines can depress…
This forum exchange addresses how to choose a target metric when optimizing strategy parameters in VeighNa’s Jupyter-based workflow. A user shows a Sharpe ratio target and asks which field represents total profit, after an attempted profit-field name did not…
A user asks whether a futures position should be counted at its full market value or at a value adjusted for margin when a strategy calculates returns by adding asset value and cash. The reply emphasizes that futures use margin and daily settlement, so their…
The document outlines a market-mode indicator based on John Ehlers’ bandpass filter. It describes the original approach as estimating typical peaks and valleys, then determining market mode by comparing the filter reading with those reference levels. This…
The article distinguishes high-frequency trading firms from quantitative investment firms by their typical origins, time horizons, capital models, and technical priorities. It describes high-frequency trading as focused on rapid processing and efficient…
The document explains a simple chart indicator that marks bars whose size exceeds a user-defined minimum. Its description says the bar’s body is considered, using the opening and closing prices to determine size. The only setting described is the minimum bar…
This stock-selection screen combines three filters: amplitude above 1%, an opening price near the 10-day moving average, and turnover between 2% and 9%. The article describes the range filter as a way to find more volatile shares, the moving-average band as…
This outline maps a beginner’s path through quantitative trading. It starts with the purpose and requirements of quant trading and the components of a complete strategy, then introduces a trading platform, its configuration, common APIs, and strategy…
This code reference presents several ways to measure dependence or distance between financial data vectors and matrices. It defines angular distance from Pearson correlation, plus absolute and squared variants that alter how negative or strong correlations…
This research summary considers how the CSI 300 might change if Chinese companies listed overseas returned to mainland China. It notes that trade tensions and tighter US oversight were associated with weaker performance of US-listed Chinese stocks relative…
The document describes a minimal short-term trading approach that chooses long or short direction at random, such as by a coin toss. Its code example first checks that no positions are open, then uses a pseudo-random value to select a buy or sell order and…
This document explains a volume-adjusted version of MACD. It replaces ordinary moving-average smoothing with volume-adjusted moving averages, calculating each average from the sum of price multiplied by volume divided by total volume over the selected…
This report studies cases where Chinese listed companies’ annual earnings forecasts later change, either through a revision announcement or a discrepancy in the formal results. It uses a multinomial Logit model to assess whether company financial data,…
The document summarizes the ResNet paper’s approach to training deep convolutional neural networks. Instead of having stacked layers learn a full transformation directly, residual learning lets them learn a change relative to their input. The paper’s…