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 describes a practitioner’s attempt to extend univariate GARCH risk analysis to a three-series DCC-GARCH model. The workflow fits separate conditional volatility models, combines them through a dynamic conditional correlation specification, and…
The document describes a one-minute foreign-exchange backtest that chooses a trade direction using the next bar’s close. Because that information is unavailable at entry time, the apparent advantage is a look-ahead bias rather than a usable signal. It also…
The document discusses ways to backtest a daily signal that indicates buying, selling, or holding an equity position. It names R packages for strategy testing and portfolio analysis, while noting that a simple strategy can also be modeled with return and…
The discussion considers a trading research process that clusters historical intraday observations, such as daily vectors of hourly exchange rates, and then measures the next day's average return for each cluster. A researcher might use the return associated…
The document discusses timestamp reliability in free historical data from Gain Capital, especially for combining non-FX instruments or multiple feeds in a backtest. One respondent reports that observations were correctly ordered within individual CSV files…
The post considers how to decide whether an implied volatility smile has unusually large skew. It defines local skew as the slope of implied volatility against log moneyness, while noting that the raw slope depends on the volatility level and may need…
The discussion addresses building a daily indicator for which Eurozone government bond is cheapest to deliver against a bond futures contract, for use in a repo-rate panel regression. It gives a Bloomberg Excel history query for the futures CTD identifier…
The post considers lag-one sample autocorrelation in daily stock log returns and asks when its scaled estimator is asymptotically standard normal under a model with normally distributed returns. It seeks a reference for the assumptions that justify this…
The document discusses implementing the nine Piotroski F-score signals with company financial data. The response treats the questions as accounting and data-mapping issues, and recommends consulting the original research for precise definitions. It suggests…
The document explains how to forecast a lower-tail, two-day Value at Risk (VaR) from a Gaussian GARCH(1,1) model and how to evaluate the forecasts. Rather than using the fitted, in-sample quantiles as the main backtest, the accepted answer recommends…
The note derives a Monte Carlo estimator for a European log-contract whose terminal payoff is the logarithm of the asset price relative to its initial price. Under risk-neutral Black–Scholes dynamics, Itô’s lemma gives a lognormal terminal asset price, which…
The document raises a measurement concern for historical tests of covered interest rate parity (CIP). It notes that interbank forward quotes are commonly expressed against the US dollar, with a stated exception, and asks whether inconsistencies between…
The document explains how to interpret a table reporting the risk of a dynamically delta-hedged option position. It describes simulating stock-price paths at the stated weekly frequency, computing the option delta at each step, and adjusting the share…
The document considers whether historical VaR for a single instrument can be extended to a one-year horizon. The main answer warns that long-horizon estimates are highly sensitive to the assumed or estimated mean return. Since drift is difficult to estimate…
The document seeks practical examples of spreadsheet and database applications used on trading desks or in bank middle offices, particularly to prepare for interviews. Responses point to several kinds of learning material: QuantLib as a source of tested…
The document explains why dates supplied to QuantLib’s futures rate helpers may not appear as nodes in a bootstrapped yield curve. A futures contract provides a forward rate for an accrual period, so its start date is not the tenor date inferred by the…
The document examines why a single simulated geometric Brownian motion path with zero drift can produce a maximum likelihood drift estimate far from zero. It defines log returns and gives the likelihood setup used to estimate drift and volatility, then…
The document explains why realized volatility used to evaluate forecasts is not directly observable. Researchers must choose a proxy, such as one derived from returns, and recognize that the resulting measure depends on assumptions about prices, sampling,…
The document asks how to represent bracket orders in a Python backtester. A bracket consists of an entry order linked to child exit orders, typically a stop loss and a take profit, with one exit canceling the other after execution. The response recommends…
The document considers estimating hedge ratios for spreads between Treasury futures when historical DV01 data are unavailable. It describes empirical DV01 as an estimate from a regression of futures price changes against changes in the yield of the…
The document discusses how to organize historical equity data so backtests can handle delistings, acquisitions, ticker changes, dividends, and splits. One proposed design assigns stable internal identifiers to companies and separately tracks stock series and…
The document addresses a discrepancy between historical stock prices shown in a finance website interface and values in its downloadable data. The example shows an apparent inconsistency between daily high and low values and the displayed close, alongside…
The document describes a fixed-risk position-sizing approach used in spread betting. The trader chooses a fraction of account equity to risk, measures the distance between entry and stop in points or pips, and divides the cash risk allowance by that distance…
The document considers whether antithetic variates must produce a positive Monte Carlo estimate for a European call when every payoff from the original simulated draws is zero. The answer is no: pairing each draw with its antithetic counterpart does not…