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 examines why Hill estimates from simulated skewed Student-t samples can differ between the left and right tails even when the distribution is parameterized with a common degrees-of-freedom value. It attributes the observed discrepancy to…
The document examines how to make rolling one-step-ahead volatility forecasts from a stochastic volatility model fitted to DAX returns using the R package stochvol. The example squares demeaned log returns, fits the model to a training sample, and then…
The document describes how to generate paired observations with specified marginal distributions and dependence structure. Given a bivariate copula, it forms a conditional distribution by differentiating the copula with respect to its first argument.…
The document explains how parametric and nonparametric statistical models differ in the structure they impose. A parametric model specifies a relationship between variables and a probability distribution for its random component, with quantities such as…
The document considers clustering methods such as k-means and k-medoids in quantitative finance. Its concrete proposed use is to identify market regimes in which relationships between assets may differ. For example, a strategy based on gold hedging equity…
The document asks whether a diffusion equation can be made deterministic by replacing its random Brownian increment with the square root of a time increment and dropping the random normal variable. The accepted explanation emphasizes that this substitution…
This discussion weighs categorizing continuous predictors, such as age or credit history, against leaving them continuous in logistic regression and related credit models. Supporters argue that carefully chosen bins can represent nonlinear or stepwise…
The document asks how to optimize a trading strategy using standard deviation bands on detrended prices, potentially with a neural network or genetic algorithm. The response says optimization requires explicit choices of parameters, such as band count and…
The document explains confusion around the parameterization of fractionally integrated GARCH models, especially the polynomial used in the FIGARCH representation and its relation to the ARCH and GARCH polynomials. It presents FIGARCH as replacing the unit…
The document compares a simple market-making setup with more detailed order-placement models for reinforcement learning. In the simple setup, a maker posts bid and ask quotes, holds them fixed for a time interval, then updates them; executions during the…
The document proposes a convolutional neural network for market-making decisions and asks whether related research exists. Its suggested inputs combine current market depth, the trader’s previous position price and size, and a sequence of recent order-book…
The document asks how to evaluate a binary support vector machine that predicts NASDAQ stock-price direction using V-fold cross-validation. The described procedure trains on all but one subset, predicts the held-out subset, and repeats until every subset has…
The document proposes a function that estimates the probability of a specified VIX increase over a chosen future horizon from a volatility time series. It suggests a Markov-switching regime model or another suitable forecasting approach, and gives an example…
The document points to references on fractional Brownian motion, including material on its numerical simulation and its potential use in financial modeling. Its central modeling feature is long-range dependence, which can make it relevant to financial time…
The question asks how to represent market inputs whose sizes differ across stocks, such as dividend schedules and volatility surfaces, when training a neural network that expects fixed-size inputs. It notes that principal-component methods can compress yield…
The document identifies the R package midasr as a tool for econometric work with mixed-frequency data and points to examples for reverse MIDAS regression. It names two approaches: reverse unrestricted MIDAS, associated with Xu and coauthors, and reverse…
The note considers estimating unknown drift and volatility parameters in a stochastic differential equation from discrete historical observations. It begins with an Euler-style increment approximation: the observed change is modeled as drift over the time…
The document considers whether retail banking models for deposits, cards, loans, and underwriting can use ideas beyond generic econometric or machine learning methods. It distinguishes risk-neutral valuation used within financial intermediation from the real…
The document poses a model-misspecification question: when returns are generated by a stochastic volatility process, can a fitted GARCH model recover or closely track the underlying latent volatility? It describes observations as a volatility scale…
The document frames closing an existing share position as a choice of liquidation time before a fixed deadline. Under its baseline assumptions—zero-drift geometric Brownian motion, known volatility, immediate full execution, and no market impact—it compares…
The document asks why a technology process is written as an exponential trend multiplied by a stochastic component, with the log component following an autoregressive process. It questions whether the trend could instead be included directly in the state…
The document discusses a high-frequency machine-learning target designed to distinguish price changes large enough to overcome transaction costs from smaller mid-price movements. A rise in the mid-price alone may still lose money if a trader buys at the ask…
A price array without timestamps cannot be classified with certainty as forward or reversed: either ordering is a possible series of stock prices. The document describes clues that may make reversal more or less plausible, especially in longer records. For…
The document asks how to approximate a GARCH(1,1) variance recursion from a discrete mean-reverting stochastic variance process associated with the Heston model. Returns are modeled as the square root of variance times a unit-variance shock, while variance…