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 indicator estimates buying and selling pressure from each bar’s price range and volume. A bull-bear balance calculation allocates volume between the two sides, then smoothed averages produce pressure series. The indicator compares buying pressure with…
This short forum exchange asks how to count trading days since a condition last became true. Its example starts counting on the session after the trigger: if the condition occurs on one date, the next trading day returns one and the following session returns…
This document explains how to build fuzzy inference systems in MQL4 with the FuzzyNet library. It covers Mamdani and Sugeno model types, fuzzy input and output variables, membership functions, terms, and rule sets. Mamdani rules map fuzzy conditions to fuzzy…
This research summary examines whether factor crowding measures can help time factor returns or adjust multi-factor portfolio weights in Chinese equities. It constructs eight relative crowding indicators across valuation spreads, pairwise correlations,…
The range-weighted exponential moving average is defined as the exponential moving average of price multiplied by a range-based weight, divided by the exponential moving average of that weight. This construction differs from a regular range-weighted average…
This stock-selection proposal combines an RSI below 65 with a positive price-to-earnings ratio and a historical filter tied to 2021. The final description specifies stocks in the upper half of that year's gain ranking, while the earlier explanation only says…
This document explains how to display Expert Advisor optimization results while a MetaTrader 5 Strategy Tester run is in progress. It describes using tester event handlers and frame functions to send per-pass information from test agents to the terminal. In…
This document describes a matrix utility for MQL4 that stores a matrix in a one-dimensional array and provides operations for setting dimensions and values, copying data, arithmetic, transposition, determinants, rank, minors, and inversion. It also lists…
This Chinese equity screen selects stocks with turnover between 3% and 12%, a reversal pattern described as a reversal-covering signal, and year-over-year growth in net profit attributable to parent-company shareholders above 20% and at or below 100%. It…
This article outlines an A-share selection screen that ranks stocks by trading activity, limits share prices to below 12 yuan, and seeks positive gains over ten days that remain below 35 percent. The stated idea is to favor stocks attracting market attention…
The document presents a hand-written Average True Range calculation and asks for an equivalent implementation using pandas without a technical-analysis library. For the first record, true range is the high minus the low. For later records, it is the greatest…
This research summary describes two equity factors, jump beta and continuous beta, constructed from five-minute market data over the prior year. It reports that both factors showed stock-selection power across sample universes, including after industry…
This short description presents a price-band indicator that adapts the general idea of Bollinger Bands and Envelopes. Its bands are based on two moving averages calculated with the least-squares linear regression method, also called LSMA. The only stated…
This A-share stock screen combines three conditions: turnover between 3% and 12%, best bid volume greater than best ask volume, and revenue in 2021 more than 1.1 times revenue in 2018. The article provides a plain-language rationale and example formulas and…
This indicator description explains an Advances-Declines measure built from the instruments shown in a platform’s Market Watch. It counts instruments with bullish and bearish candlesticks, then forms a normalized advance-decline ratio by subtracting the…
This Chinese-language research note analyzes recent trading value as a liquidity factor for stock selection. It describes changing the evaluation horizon to five-day returns, excluding stocks at their daily upper price limit to reflect execution difficulty…
This indicator description explains a way to smooth an RSI time series by filtering higher-order harmonics. The stated aim is to reduce roughness while avoiding the delay associated with ordinary smoothing, and the author says the same approach can be…
This stock-screening rule selects Chinese A-share companies using daily turnover, circulating market value, and region. It specifies turnover between 3% and 12%, circulating value between 5 billion and 10 billion yuan, and excludes Beijing-listed companies.…
This stock-selection proposal combines a volatility filter, a weekly moving-average crossover, and historical revenue growth. It seeks stocks with an amplitude above one, a weekly five-period moving average crossing above the ten-period average, and 2021…
This experiment runner configures repeated, rolling train-and-test evaluations for LSTM and Temporal Fusion Transformer models on a multi-asset Quandl dataset. It offers variants with different input sequence lengths and optional changepoint feature…
This report introduction frames research into how factor and stock weights affect style exposure in multi-factor equity portfolios. It focuses on reducing unintended market-cap, industry, and other style biases, which the authors associate with past setbacks…
This indicator method searches for bullish and bearish divergences between price swings and RSX, an RSI-derived oscillator. It calculates an RSX-like series from changes in a price measure smoothed with Wilder averages, then compares that series with a short…
The document presents the Brinson framework for explaining portfolio returns through the benchmark, active asset allocation or timing, security selection, and their interaction. The benchmark reflects long-term asset-class weights and passive returns. Timing…
This forum post reports that two BigQuant notebooks produce different numbers of extracted features even when no filters are applied. The author observes that one notebook appears to include Beijing Stock Exchange data while another does not, and questions…