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 asks how to backtest a stock-direction model that is retrained on historical price windows and used to generate trades. The author reports prediction accuracy and wants to test a strategy over multiple years, including risk controls such as stop…
The document offers career-development advice to a mathematics PhD with derivatives valuation experience who wants demonstrable C++ work implementing financial models. The answer suggests exploring public repositories focused on topics such as option pricing…
The note examines a MATLAB implementation of the log-likelihood for a GARCH model whose standardized innovations follow Hansen’s skew-t distribution. It asks whether a normalization parameter should be square-rooted and why the likelihood includes a term…
The document compares R with commercial tools such as Matlab and Mathematica for quantitative research. It notes that R has packages for statistical analysis, simulation, optimization, and numerical integration, then gathers user reports about areas where…
The document addresses why a basic GARCH model may be outperformed by realized GARCH or other realized-volatility models, especially when modeling intraday volatility. Its explanation focuses on information inputs rather than a strict difference between…
The document describes a forecasting setup for daily commodity prices using an LSTM network and predictor variables drawn from current and lagged observations. The author asks whether forecast accuracy can determine if the model is suitable for managing…
The document explains how least-squares Monte Carlo (LSM) estimates the value of callable derivatives and exposure measures without repeatedly running costly nested simulations. It simulates model states and their discounted future cash flows, then fits a…
The document raises a portfolio-construction question about coordinating an alpha model with a statistical risk model. The risk model decomposes returns into factor exposures and factor returns, leaving residual returns. The proposed alpha model uses…
The document discusses simulating returns without assuming a fixed unconditional distribution. It warns that financial returns have time-varying conditional moments, especially persistent volatility clustering, so resampling the unconditional return…
The document considers how machine learning could extend cross-sectional factor models used to explain security returns and estimate portfolio variance. It suggests modeling factor exposures as functions of observable security characteristics, such as size,…
The document addresses whether replacing a statistically significant linear regression with a Kalman filter will necessarily improve predictions. Its central answer is conditional: performance depends on the data and on how the filter is calibrated. A Kalman…
The note examines why Hidden Markov Models are used for financial prices despite their first-order Markov assumption: future states depend on the present state, rather than directly on the full path of prior observations. It also asks whether technical…
The note explains why a low R-squared can still be meaningful in return prediction. In regression, R-squared measures improvement in accounting for the target’s variation relative to a baseline that always predicts its mean. A positive value may indicate…
The discussion weighs whether one person can learn quantitative finance and build predictive trading models, especially with machine learning. It distinguishes learning the subject from earning a living through algorithmic trading: study partners can help…
The document considers how to estimate a commercial bank’s core deposits from ten years of monthly account balances. Its starting proposal is to take each account’s minimum balance in each month and sum those minima across accounts. The response says this…
The discussion compares principal component analysis (PCA) and clustering as ways to build risk models for portfolio optimization. PCA is presented as a method for reducing a set of risk variables to a smaller number of uncorrelated factors. Clustering…
The answer explains why published deep learning results in trading are relatively scarce and why performance depends on the market and available data. It argues that rich equity data, including news and company information, may support more complex models,…
The discussion explains that choosing the process-noise covariance Q and observation-noise covariance R is an estimation problem, with no universally exact setting for financial data. One suggested route is to fit covariances through least-squares…
The discussion distinguishes estimating implied volatility from observed option prices from using machine learning to approximate prices for complex derivatives. Inverting a standard option pricing model to find implied volatility is described as a simple,…
The discussion surveys ways to combine financial series observed at different frequencies and use them in forecasts. MIDAS models incorporate higher-frequency predictors into a lower-frequency target, such as daily market data in a quarterly economic…
This discussion asks whether input attributes need to be scaled before gradient tree boosting. One answer says scaling is generally unnecessary for tree-based boosting, since the distance-based concerns that motivate normalization for methods such as support…
The document describes a CPI forecasting regression whose author sees a moving average of the residuals moving in the same direction as CPI. The author asks whether this suggests omitted-variable bias, biased coefficients, or another modeling problem. The…
The document asks whether stock exposure to market, value, or momentum factors should be estimated using returns during factor drawdowns, since those losses may be linked to factor risk premiums. It also raises the estimation challenge: crash periods offer…
The discussion clarifies that “linearity” in classical statistics is broader than whether a formula contains only first-degree terms. It refers to properties of relationships and estimators, including linear combinations of random variables and linearity…