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.
5,018 documents
The document identifies the fitness-function setting used by a genetic programming module and lists its available objectives. These include information coefficient to information coefficient ratio, mutual information, long-only Sharpe, volatility and return…
The document describes a script that exports values from built-in MQL5 indicators to a CSV dataset for analysis. It identifies several indicator buffers, including the center and outer Bollinger Bands, ADX lines, MACD and signal lines, RVI, Stochastic, tick…
This research summary describes combining order-book data, which may reflect traders’ unexecuted intentions, with transaction-level data showing completed trades. The authors use both intuitive signal construction and machine learning to turn different types…
This page catalogs trading bot examples built around AI agents, ranging from copying reported investor or insider holdings to sentiment signals, agent debates, options strategies, intraday rules, and macro or sector portfolio discussions. It outlines…
This quick-start example outlines a FinRL workflow using a stock-trading environment and a Dow 30 ticker universe. A command-line mode selects among training, testing, and trading paths. The example configures daily Yahoo Finance data, technical indicators,…
This brief support exchange addresses whether StockRanker’s tree model can use L1 or L2 regularization and how to improve model stability. The response says that StockRanker does not currently expose regularization as a way to prevent overfitting. It…
This overview surveys the use of large language models and other AI methods across quantitative investing. It describes extracting sentiment and information from financial text, combining text with time-series data for market and risk forecasts, supporting…
The document introduces cryptocurrency trading bots as software that places trades according to preset rules. It describes three possible inputs: technical indicators such as moving averages and RSI, predictions based on historical data, and external trading…
This brief forum exchange asks how to interpret a random forest model’s predicted label after returns have been divided into twenty classes. The question is whether the output represents a class identifier or a return value, and whether regression metrics…
This case study describes how blockchain ratings company Metafide used a third-party data platform to support an AI and machine-learning product. Its required inputs spanned blockchain networks, crypto market data, decentralized finance, and several traded…
This reference lists configuration options for FreqAI, Freqtrade's machine-learning feature. General settings cover rolling training and inference windows, model identification and persistence, retraining frequency, model expiration, and prediction…
The article explains support vector machines (SVMs) as classifiers that seek a separating boundary between classes. It introduces the geometric intuition of mapping observations into a higher-dimensional feature space, where a nonlinear boundary in the…
This brief forum exchange addresses a quant workflow problem: a ZScoreNorm standardization step produces only missing values. The response explains that the operation performs cross-sectional normalization and suggests inspecting the input data for an…
The indicator estimates current and upcoming daily volume by averaging historical observations from the same weekday. Users choose the number of weeks to include and whether to display tick volume or real volume. The plotted average can be compared with…
This notebook outlines a machine learning workflow for daily CSI 300 constituent stocks. It loads historical bars and changing index membership filters, constructs an Alpha101 dataset, and divides the sample into training, validation, and test periods. The…
This article argues that LSTM models trained on historical prices can appear to forecast stocks accurately because predicted daily prices often remain close to the previous day’s price. When plotted against actual prices, this persistence can look…
The article summarizes a method for extracting priced common factors from stock returns. First, a Fama-MacBeth cross-sectional regression uses several return-predictive characteristics to estimate stocks’ expected returns. Stocks are then sorted into…
This article outlines a reproduction of a research approach that predicts whether a stock’s closing price will rise or fall over a future horizon. It applies exponential smoothing to Apple daily price data, then builds a binary target from the price change…
This tutorial introduces bias, variance, and irreducible error as sources of prediction error. It relates model complexity to underfitting and overfitting: overly simple models tend to perform poorly on both training and test data, while overly complex…
This discussion asks whether the maturation of China’s equity market will make quantitative investing harder. The concern is that institutional participation and competition may reduce behavioral inefficiencies and compress excess returns for both systematic…
This introductory guide outlines a workflow for training and backtesting a cryptocurrency trading agent using a scripting framework. It explains reinforcement learning as an approach in which an agent takes actions in a market environment and receives…
This configuration specifies a reinforcement learning setup for order execution using Proximal Policy Optimization. It defines a categorical action interpreter, a recurrent network, and a full-history state representation built from intraday and prior-day…
This brief Chinese-language forum post reports a data-quality issue discovered while debugging backtests: some runs returned no results, and inspection showed that certain factors were empty in many cases. The author notes that noticing the missing values…
The note addresses a stock forecasting workflow that displayed an unexpected prediction date and failed when publishing a simulated trade. Its proposed fix is to make sure the training period and prediction period are not set to the same dates. This is a…