block-bootstrap

Fast, dependence-aware resampling and uncertainty for time series and ragged panels.

Open-source investment analytics platform bridging academic research and retail finance. Features include portfolio risk decomposition [Fama-French Five Factor Model], retirement sustainability modeling [Block Bootstrap Monte Carlo], max drawdown/CVaR dashboards, and risk-return optimisation [Markowitz, Ledoit-Wolf] via an intuitive user interface.

Estimate confidence intervals in means of correlated time series with a small number of effective samples (like molecular dynamics simulations).

Paper VII of Statistical Pharmacology via Kakutani Dichotomy: kakutani_pharma, a Python pipeline for Kakutani indices of MD conformational ensembles. Ledoit-Wolf regularized CKI with an exact three-way decomposition, split-trajectory null subtraction, within-half block bootstrap, and pocket-centred shell-scaling exponents. Validated on a synthetic

15. Višestruko / panel sečenje sliding window stability, expanding window, block bootstrap, stratified by era, cohort split, concept-drift detectors (ADWIN, DDM, Page-Hinkley)

Monte Carlo blackjack engine measuring conditional edge by true count, then Kelly bet sizing with risk of ruin from a block bootstrap

Pre-registered market-signal studies: sha256-frozen hypotheses, block-bootstrap p-values, Benjamini-Hochberg correction and a hold-out opened once. Both studies failed and were abandoned.

Probabilistic forecasting and walk-forward validation for Brazilian markets.

How often a nominal 95% bootstrap confidence interval actually covers, when the data are autocorrelated. One file, numpy only, reproducible.

Non-parametric portfolio risk simulator using circular block bootstrap (Politis-Romano). Simulates outcome distributions, VaR/CVaR, drawdown, DCA/SIP -- with walk-forward calibration and 52 + 58 QA invariant checks.

Paper VIII of Statistical Pharmacology via Kakutani Dichotomy: validation on real PDB ensembles (ubiquitin NMR, adenylate kinase), Isserlis bound and influence-function block selection, bootstrap coverage calibration. Code, data, paper.

Monte Carlo study of dollar-cost-averaging strategies, 1980–2026. Ranks 56 portfolios (static, age-glide, momentum/signal) by median final wealth vs. terminal drawdown pain via block-bootstrap simulation in taxable and tax-free accounts. Full write-up in wealth_report.pdf.