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
Mean-variance optimization can produce unstable allocations when expected returns and covariance inputs are estimated with error. The discussion surveys practical ways to reduce that sensitivity: shrink historical estimates toward common targets, use robust…
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 poses a calibration question for a stock modeled as a monthly random walk with independent, identically distributed changes. Each monthly change is assumed to equal either a positive or negative fixed amount with equal probability, giving the…
The document raises a modeling concern about applying historical-return bootstrapping to a government bond product under the PRIIPs Category 3 market-risk measure. A bootstrap based on past returns may project prices that do not reflect the bond’s…
The document raises practical questions about using block bootstrap to generate synthetic histories for a stock portfolio and a market index. Its central concern is preserving pairwise dependence across assets while resampling time-series data. It asks…
The document outlines a way to estimate day-over-day changes in historical simulation VaR without fully revaluing the portfolio. Begin with the historical risk vector, adjust it for new and removed positions using their sensitivities, and update portfolio…
The document outlines data management issues that can affect backtests using intraday US equity quotes and trades. Beyond missing records, timestamp alignment, and cleaning unusual trades, it highlights changing symbol histories and trading sessions,…
The document discusses stopping a Monte Carlo valuation of a European option when the estimate is sufficiently precise. The proposed criterion compares a confidence-interval half-width, estimated as a normal critical value times the sample standard error,…
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 clarifies a common backtesting problem in a model that uses a completed candle’s closing price to generate a signal for the next period. The training setup is not automatically biased merely because it predicts a future value from past…
The document presents code intended to simulate a GARCH(1,1) process. It updates conditional variance using a constant plus a coefficient on the previous squared return and another on the previous variance, then multiplies the resulting volatility by a…
The discussion explains a common source of discontinuities when building return series for standardized contracts such as IMM swaps or credit default swaps. A fixed-maturity CDS quote can remain stable across a roll date, while its label as an on-the-run…
The answer proposes simulating a price process with a fixed number of hourly steps and bounded daily returns. It sets the size of each step from the upper and lower daily limits and the number of steps, then uses Bernoulli draws to choose the direction of…
The document asks why historical return quantiles can be used for Value-at-Risk when returns are dependent, as in ARCH-type models. The response clarifies that dependence among returns does not prevent modeling their marginal distribution: volatility models…
The document asks how to judge whether an estimated covariance matrix predicts future portfolio risk. It considers comparing a training-period estimate with a later sample covariance using the Frobenius norm, then suggests evaluating whether a mean-variance…
The document asks whether mean-variance optimization research remains worthwhile when expected-return estimates are noisy. It cites a discussion of out-of-sample results in which more accurate volatility forecasts alone did not materially improve the Sharpe…
The document raises an alternative to the common factor backtest that buys the highest-ranked group and shorts the lowest-ranked group. It quotes a research proposal to construct a long-only portfolio with exposure to a chosen factor while targeting a…
The document presents a QuantLib setup for building a USD OIS discount curve and a SOFR projection curve from deposit, futures, and swap quotes, then pricing swaps at matching quoted tenors. The author reports small differences between the calculated fair…
The document examines whether a short run of profitable manual trades in equity-index futures can support expectations of unusually high annual returns. The responses caution that a small number of recent trades is weak evidence for persistent skill, and…
The document asks whether a basic Black–Scholes spreadsheet for European options can be adapted to price American options. It suggests Whaley’s approximation as a possible spreadsheet method when limited accuracy is acceptable. The approach requires…
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 document compares row-oriented and column-oriented storage for financial time series, focusing on tick trades and OHLCV bars. One answer favors conventional relational databases for a single-symbol-per-table workload with frequent writes, citing the…
The document distinguishes a historical quantile estimate from a parametric Value-at-Risk forecast based on Student’s t returns. It describes fitting a t-distribution to log returns, then estimating a one-day-ahead VaR by taking a lower-tail quantile from…
The document presents ways to assess how the number of Monte Carlo scenarios affects an estimated value at risk. One approach repeats the VaR calculation many times at a fixed scenario count, then uses the sample standard deviation of those estimates as an…