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
The discussion considers whether high-dimensional Sobol sequences are practical for large market simulations and Monte Carlo pricing, where the required dimension may exceed the limits of common implementations. One answer points to established Sobol…
The document considers statistical tools for testing whether historical changes in a market variable appear random, with the aim of identifying unusual patterns and consistency over time. It points to the NIST test suite, a collection originally used to…
The document discusses narrowing a large universe of candidate equity pairs after an Augmented Dickey–Fuller test and half-life calculation leave many candidates. Suggested filters include liquidity, trading costs, backtesting with cross-validation, and…
The document argues that risk management can sit beside an ML or AI alpha model rather than trying to interpret its internal features. A separate portfolio controller can translate forecasts or directional signals into feasible positions, accounting for…
The document addresses how to obtain approximate critical values for the Johansen cointegration trace test when the system has more variables than commonly tabulated references cover. It proposes an approximation based on a correction factor that depends on…
The document considers whether a backtest with 50 trades and a 60% win rate reflects skill or chance. It argues that repeatedly comparing the observed result with randomly generated strategies does not, by itself, establish a false-positive probability. A…
The document asks how to search for predictive relationships between a target series and many candidate variables. Its example is an inverted, three-month-delayed variable that appears to lead stock prices, and the proposed application is predicting future…
The discussion compares accounts of quote data and order handling in the QuantConnect backtesting environment, while also mentioning Quantopian and other tools in the question. One answer says the platform lacked bid and ask data at the time described and…
The document surveys research and practical ideas for using news, blogs, social media, and company reports to forecast equity returns or other financial variables. It describes vendor sentiment scores and studies that apply nonlinear models, including trees,…
The document describes a Monte Carlo setup for studying the profit and loss of a long out-of-the-money call that is delta hedged over time. The option is purchased using the volatility assumed at inception, while simulated underlying price paths reflect the…
The discussion explains why a trading desk’s non-stationary daily P&L does not by itself prove that next year’s P&L is unpredictable. A desk’s P&L may reflect strategies and exposures that differ from the return stream of an investible asset, so…
The document discusses how to answer a broad interview question about forecasting a stock’s future behavior. It distinguishes estimating returns from measuring risk, mentioning historical mean returns, volatility, market sensitivity, and risk-adjusted…
The document introduces the Campbell–Shiller regression as a way to test the expectations hypothesis of interest rates. It describes comparing a yield spread with a subsequent change in yields, while noting that empirical estimates may differ from the…
The document responds to a broad request for a practical starting point for forecasting European government or other bond prices from historical data. Its central modeling recommendation is to forecast bond returns rather than prices directly. It then…
The document raises a replication question about Ledoit and Wolf’s covariance matrix shrinkage method. The author is trying to reproduce a plot relating Percentage Relative Improvement in Average Loss (PRIAL) to the ratio of the number of variables to the…
The document compares finite difference methods with Monte Carlo for pricing options on baskets of assets. It explains the dimensionality problem: a finite difference grid grows rapidly with the number of underlyings, making multidimensional PDE methods…
The document discusses plotting the historical minimum and maximum returns over holding periods of increasing length, alongside a central band, to visualize how observed outcomes vary by horizon. The accepted replies suggest variance ratios as a related…
The discussion addresses how to assess an intraday equity strategy whose backtest shows about one cent of gross profit per trade before transaction costs. It emphasizes that Sharpe ratio and profit per share are useful, but neither establishes viability…
This document examines fitting ARIMA models across rolling windows to track coefficient changes. Its central difficulty is that automatic model selection may choose a different order in each window, so the first reported coefficient may represent an…
The document discusses adding candlestick-pattern features to a machine-learning model that predicts whether a stock’s next daily close will rise or fall. The researcher scores patterns by the share of observed cases in which price moved in the pattern’s…
The discussion surveys software and methods for testing options strategies, including multi-leg spreads. It explains why this work is harder than testing linear instruments: historical option chains are large, implied volatility histories may be unavailable,…
The discussion distinguishes elicitation from backtesting. Expected Shortfall (ES) lacks a scoring function that uniquely elicits it as a point forecast, but that limitation does not by itself make ES impossible to backtest. The document cites work proposing…
The question concerns whether a short-horizon foreign exchange forecasting model should use bid prices, ask prices, or a blended series, and how to evaluate trades in a backtest. The response distinguishes measuring market movement from simulating executable…
The document discusses how to design a cross-asset pricing library that supports multiple models and numerical methods while keeping products and payoffs immutable for serialization. Its central design principle is to keep a payoff focused on contract terms…