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 document proposes estimating common return factors and asset loadings with a rolling singular value decomposition of volatility-normalized close-to-close returns. It then applies the estimated loadings to other return intervals, such as overnight or…
The document discusses generating synthetic foreign-exchange data for algorithm testing, including return paths, OHLC bars, tick timing, and bid-ask spreads. Its main caution is that Gaussian noise around a trend is a poor representation of high-frequency FX…
The note explains how numbered Bloomberg crude oil indices represent exposure to futures contracts at different points along the Brent or WTI futures curve. The number indicates the contract’s forward position relative to the front month. The example…
The document outlines a proposed search process for automated intraday strategies in FX and equity index futures. The plan uses short forecast horizons, many candidate features, genetic feature selection, and SVM models ranked by cross-validation error. The…
The document responds to a request for rigorous resources that connect portfolio-optimization theory with practical implementation. It points readers to several books using R, Matlab, or S, and briefly compares their audiences and coverage. Topics mentioned…
The document surveys data sources that systematic traders can use beyond historical prices, including traded volume, company fundamentals, and alternative data such as satellite imagery, news text, supply-chain information, and card transactions. Its main…
The document asks how to calculate a ten-day, 99% Value at Risk for a portfolio that combines shares with short European calls on the same company. It gives the share value, annual volatility, and option delta, but the response cautions that delta alone does…
The document asks how to validate a performing-loan Loss Given Default model by comparing its estimates with realized workout LGD. It distinguishes a model developed using a through-the-cycle sample of defaulted loans from estimates produced for a…
The document considers a lookback call in a two-period Cox–Ross–Rubinstein binomial model. Its payoff is the terminal asset price minus the lowest price reached across the life of the option, so the payoff depends on the path as well as the terminal price.…
The document discusses how to choose the number of hidden states in a hidden Markov model, using financial return data as an example. It compares selecting states by out-of-sample log likelihood with choosing the best in-sample fit using an information…
The document addresses backtesting crypto strategies that calculate indicators on daily candles while processing finer-grained data for entries, exits, and risk controls. Its accepted answer recommends using Backtrader’s data replay feature with minute data…
The document explains how to choose observations for an exponential moving average (EMA) on bars longer than one day. It reports that closing prices are a common input: the final trading-day close for a weekly bar and the final trading-day close for a…
The document asks how to turn forecasts from an ARIMA model of daily power futures returns into prediction intervals for prices. It describes a simple approach: reverse the data transformations, compound the forecasted returns from the latest observed price,…
The document asks whether correlations should be calculated from simple returns, log returns, or log prices, and whether the choice is merely conventional. The response ties the answer to the intended use: when correlations will drive a simulation, estimate…
The document proposes estimating portfolio profit distributions by simulating monthly generation for each energy asset and electricity forward prices, multiplying generation by price to estimate asset revenues, summing across assets, and taking lower-tail…
The document addresses how to normalize real-time observations for a support vector machine trained with automatic scaling. Its central guidance is to apply the mean and standard deviation estimated from the training data to out-of-sample observations.…
The document clarifies how to calculate a one-step-ahead density forecast from an estimated ARMA-GARCH model with Gaussian errors. At each forecast origin, estimate or update the model using information available so far, then obtain the forecast conditional…
The document considers quantitative modeling of interest-only adjustable-rate mortgages. It proposes using index forward curves to project how mortgage rates may change after an initial fixed-rate period, then examining effects on borrower costs and payment…
The document describes an analysis of daily returns for eleven MSCI World sectors and the MSCI ACWI index. The analyst groups daily observations into weekly and monthly periods, calculates a correlation for each group, and applies the Augmented Dickey-Fuller…
The document considers how to assess bid, ask, and volume data from two vendors when observations do not share exact timestamps. The questioner samples prices at regular time intervals, finds that plots look similar, but sees weak cross-correlation after…
The document examines how to annualize a cumulative return when capital was invested during separated intervals. It contrasts annualizing over the 450 days actually invested with annualizing over the full three-year span. The responses distinguish a…
The document explains why historical stock research can be biased when datasets omit firms that have gone bankrupt or otherwise left the market. It notes that some widely used finance sites focus on surviving companies, so researchers seeking bankrupt or…
The document considers whether official central bank exchange rate fixings can stand in for end-of-day spot forex data in lower-frequency analysis. The response points to the European Central Bank’s daily exchange rate series, noting that its rates are fixed…
The discussion examines how adjusted and unadjusted stock prices affect backtests, especially cointegrated pairs trading. Adjusted prices can remove artificial jumps from splits and dividends, which helps when measuring returns or momentum. But later…