The SharpeR package supports statistical inference on the Sharpe ratio: computing confidence intervals on the Sharpe, as well as the 'maximal Sharpe' of the Markowitz portfolio, estimating power, and so on.
The MarkowitzR package supports statistical inference on the elements of the Markowitz portfolio, including the linear conditional expectation case.
The PDQutils package allows one to approximately compute the distribution, density, and quantile functions (the so-called 'p, d and q' functions in R) of probability distributions when only the moments or cumulants are known.
The sadists package is a showcase for the
PDQutilspackage. It provides density, distribution, quantile, and random generation for several classes of probability distributions.
The madness package accumulates derivatives of multivariate-valued functions of multivariate input by forward differentiation. The primary use case for this is in the delta method: one can write code in R to transform simple means into more elaborate estimates, while
madnesscomputes the derivatives automatically.
The fromo package provides fast numerically robust computation of moments and cumulants. Both summarization and rolling computations are supported. Mostly this package was an exercise in learning Rcpp.
The BWStest package provides the unpaired two-sample test of the hypothesis of equal probability distributions, as described by Baumgartner Weiss and Schindler. The test is reasonably powerful in comparison to the classical tests (Kolmogorov-Smirnov, Wilcoxon, and Cramer-Von Mises). Another fun exercise in Rcpp.