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.
16,761 documents
This paper compares machine learning methods for forecasting equity returns across the market time series and the cross section of stocks. It frames risk premium measurement as a prediction problem and describes how high-dimensional predictors,…
This document describes an Expert Advisor built around the XCCI_Histogram_Vol indicator. It generates a signal when a bar closes after crossing an overbought or oversold threshold. Two threshold levels are associated with separate order identifiers and…
This stock-screening proposal looks for shares with an amplitude measure above 1, three technical indicators said to have bullish crosses, and three consecutive declining closes. Its sample Python logic approximates the multiple-cross condition with the 5-,…
The paper combines return prediction and portfolio construction in a single neural network, aiming to reduce the decision errors that can arise when forecasts are optimized separately. It compares a model-free network, which learns allocations directly, with…
This document describes a stock-selection rule that combines a daily high-low range threshold, a recent reversal pattern, and a negative MACD reading from two sessions earlier. It frames range as a volatility filter, the reversal as a possible change in…
This post proposes a technical stock screen combining price amplitude above a threshold, progressively rising lows, and a volume ratio between stated bounds. It describes the rising floor as a sign of improving price structure and uses relative volume to…
This community exchange explains how to use data downloaded with Supermind’s data ingestion command. A reply says the downloaded data can be accessed through the platform’s get_price function, without specifying the on-disk file format or a separate…
This note describes a histogram version of an adaptive moving average based on Perry Kaufman’s KAMA and an update by Vitali Apirine. The update accounts for where the close sits within the high–low range when adapting the average. The indicator combines the…
GodStra is a Freqtrade long strategy that builds a broad set of technical indicators and evaluates configurable entry and exit rules. Each rule compares a selected indicator with another indicator or a numeric threshold; supported operators include…
The document proposes screening Chinese stocks for amplitude above 1%, an aligned set of three moving averages, and a 2021-to-2018 revenue ratio above 1.1. It presents the moving-average condition as a technical signal and the revenue comparison as a basic…
This Chinese-language research summary explains how to estimate consensus net profit from analyst forecasts and turn that estimate into equity-selection signals. It proposes weighting forecasts by report recency and by each analyst’s prior adjusted forecast…
This short forum exchange describes a backtesting issue in VeighNa: several built-in strategies reportedly generated many trades that closed at a price of zero when using data from the Wind Python API configured through VN Station. A respondent suggests that…
This Chinese research summary examines whether intraday data can support sector rotation signals, focusing on realized skewness and the share of volatility attributable to downside moves. It describes constructing industry level factors inspired by high…
A trader reports that a stock-selection strategy using TA-Lib indicators produces different results each time it is run. The response points to an SQL query in one module that lacks an explicit sort order as the likely cause. When rows arrive in an…
The document presents a Chinese equity screening rule that selects stocks with turnover between 3% and 12%, displayed first-level bid volume greater than ask volume, and reported holder concentration below 20%. The candidates are then ranked by price change,…
The document describes a configurable engine that combines up to 11 common indicators into long or short signals. Users can require every enabled indicator to agree or allow any one of them to trigger. The indicators include moving average crossovers, RSI,…
The document presents a method for moderating drawdowns by reducing the capital base used to size trades as portfolio value falls. It first defines maximum drawdown as the largest peak-to-later-trough decline relative to the earlier value. Its adjustment…
The document outlines a mean-reversion strategy for two assets selected for cointegration. It uses the Engle–Granger two-step approach: regress one price series on the other, test whether the residuals are stationary, then fit an error-correction model and…
This user post describes rolling model training for stock forecasts using two approaches. The StockRanker example updates the model annually, fitting on one year of data and predicting over the following year. The XGBoost example instead constructs monthly…
This configuration sets up a Qlib experiment that uses a gated recurrent unit model with Alpha360 features to rank CSI 300 constituents. Feature values are robustly normalized with outlier clipping and missing-value filling; labels use cross-sectional rank…
This document outlines a Chinese equity screening rule that combines daily price movement with a short moving average and a lower low. The stated final criteria are amplitude above 1, the close crossing above a five-day moving average, and today’s low…
This article examines how training-window length affects an AI stock-selection model. It compares long windows, expanded year by year from 2005 to 2021, with shorter windows ranging from a month to a few years. The author recommends checking each window…
This note proposes screening Chinese stocks using several short-term conditions: daily amplitude above 1, a three-session limit-up sequence on the previous day, exclusion of ST-designated stocks, and selection before 10 a.m. It also invokes a five-session…
This note describes a Chinese equity screen that ranks stocks by turnover ratio and keeps the top 100, then applies an opening-price condition: the stock was at its price limit at 9:15 and its gain by 9:25 is below 6%. It presents the conditions as a way to…