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 addresses frequent Value at Risk exceedances in backtests of portfolios whose VaR was estimated with historical simulation using rolling samples of 250 or 500 daily returns. It frames the issue as potential underestimation of risk and mentions…
Volatility forecast evaluation depends on what quantity is forecast and what observation serves as the benchmark. Possible targets include volatility, variance, or log volatility; because true volatility is latent, evaluation may use implied volatility or a…
The document raises practical questions about using XGBoost to predict whether the next candle’s closing price will move above or below a chosen threshold from the current open. The proposed feature set combines market sentiment, macroeconomic variables,…
The document explains the popularity of GARCH(1,1) as a volatility model. Its simple structure, connection to established time series methods, and ability to represent volatility clustering and heavy tailed returns help explain its use. The discussion cites…
The document asks how to annualize a Sharpe ratio for an intraday strategy evaluated with prices sampled every ten seconds. The trader describes computing each interval return, summing those returns to form a daily result, and then applying the usual…
The document explains why rolling multi-day returns are statistically dependent and how that dependence affects covariance. It expands the covariance of two overlapping two-day return series into same-day covariance terms and cross-time covariance terms.…
The document describes a study design for sorting S&P 600 stocks into three portfolios according to historical return volatility. The proposed process estimates each stock’s volatility over a formation year, assigns stocks to terciles at the year end, and…
The document asks how to generate many asset return histories that match summary statistics from capital market assumptions, including expected returns, quartiles, volatilities, and incomplete correlation information. One response suggests initially setting…
The document explains why ARMA coefficients estimated together with a GARCH variance model can differ from coefficients estimated in an ARMA model alone. The variance component accounts for changing volatility, which can affect the mean-equation estimates.…
The document asks how to find investment products whose systematic strategies and performance can be inspected and reproduced. It points to exchange-traded funds as a practical starting point, especially products built to track quantitative or…
The document offers research guidance for studying whether neural networks can predict liquidity or bid-ask spreads from market depth data. It recommends preparing the data carefully, considering zero or negative spreads, and planning how limited…
The document outlines how to approximate critical values for an augmented Dickey–Fuller test by simulating the test statistic under its null hypothesis. The test is based on a regression for the change in a series that includes its level, an intercept and…
The document describes a QuantLib issue encountered while bootstrapping a futures-based yield curve. A FuturesRateHelper constructed with explicit start and end dates, an ASX futures type, and numeric price and convexity inputs returns a null maturity date;…
The document frames look-ahead leakage checks as mutation testing. At a historical cutoff, a valid mutation preserves information available at that time while changing only future data; if the historical artifact changes, this suggests the pipeline used…
The document asks whether portfolios built from simple rules should be called efficient. In Markowitz portfolio theory, efficiency has a specific meaning: for a given expected return, a portfolio has the lowest possible risk, placing it on the efficient…
The document addresses how to measure more than the largest drawdown: it seeks the count, average depth, and average length of drawdown episodes that exceed a chosen duration threshold. It presents two approaches using price data in Pandas. One calculates…
The document describes a replication attempt involving empirical equity premium prediction. It calculates a dividend-inclusive return from consecutive index prices, adds dividends to the ending index value, converts the resulting return to logarithms, and…
The document explains why predictive accuracy alone does not establish that a trading strategy will earn money. Forecasts and backtests can omit execution costs, commissions, order-book depth, bid and ask prices, and the effect of the strategy’s own orders…
The document outlines a practical architecture for calculating value at risk when a trading system already generates simulated market scenarios, such as interest rate curves. In that setup, the main task is to reprice instruments under each scenario rather…
The document asks how to generate observations from a four-factor t-copula when the dependence matrix changes over time. Its setting is a series of 2,000 correlation matrices from a copula-DCC model, with one matrix intended for each time step. The accepted…
The document asks how to determine whether a strategy’s performance edge has weakened across time periods. It mentions comparing period means with a t-test as an approach encountered elsewhere, while expressing concern that this test may be unreliable for…
The document questions whether daily leveraged ETFs can serve as long-term investments and whether switching between bullish and bearish funds could improve returns. It describes a simple simulation that compounds daily leveraged returns generated from…
This note frames trading under an assumed known stochastic price model as a finite-horizon stochastic control problem. The trader’s state includes capital, current position, price, and model parameters; the model evolves price and parameters using noise. The…
The document considers a daily classification task: using index features to predict whether the next day's open-to-close move is up or down. The proposed workflow forms a direction label, selects and scales features, trains an SVM on a rolling window, and…