LoginSign Up
← Back to all sparks
P

portfoliobacktest

INFRA · APIS
Velocity0.0

R package portfoliobacktest by dppalomar — release notes from GitHub.

github.com/dppalomar/portfoliobacktest ↗

Four dormant years end with a test fix, not a feature

quantitative-financebacktestingportfolio-optimizationr-packagemaintenance
◆Current state
portfolioBacktest runs portfolio strategies across multiple datasets and resampled windows, producing performance tables, leaderboards and charts. After a run of substantive releases through 2021 and early 2022 it went quiet for four years. The May 2026 release does one thing: fix unit tests against a breaking change in PerformanceAnalytics::Return.portfolio(), which now returns results for the full input series rather than only from the first weight date.
◆Where it's heading
The development work is behind it. The 2019 to 2022 releases built the substance — parallel execution, resampling that works beyond stock data and beyond daily bars, transaction-cost-aware designs through w_current, and a switch from compounded to uncompounded returns in the Sharpe and annualized return calculations. Since then the package has moved only when a dependency forced it, and even then only far enough to keep the tests green.
◆Prediction
Nothing in the entries suggests planned work; the next release is most likely triggered by another dependency change, on the same four-year-gap pattern.

◆Recent moves

  1. 3mo ago

    Tests fixed for a PerformanceAnalytics breaking change

    Unit tests adjusted for a breaking change in PerformanceAnalytics::Return.portfolio(), which now returns the full input return series rather than starting at the first weight date. Test-side only, and the only release in four years.

    View source ↗
  2. 4y ago

    Sharpe ratio charts and reworked weight outputs

    summaryTable() gains kable output, backtestChartSharpeRatio() is added, backtestBoxPlot() accepts a reference portfolio, and portfolioBacktest() returns w_optimized and w_rebalanced in place of w_designed plus X_lin for debugging. The uniform benchmark is renamed 1/N.

    View source ↗
  3. 4y ago

    Uncompounded return metrics and transaction-cost-aware designs

    Sharpe ratio and annualized return switch from compounded to uncompounded returns, changing every number the package reports for existing backtests. The portfolio function also receives w_current, opening the door to transaction-cost-aware and smart-rebalancing designs.

    View source ↗
  4. 5y ago

    Resampling generalised beyond daily stock data

    Parallel mode is reimplemented on pbapply, stockDataResample() becomes financialDataResample() to handle non-stock data such as crypto, MDP and MSRP join the benchmarks, and non-daily data is supported through a bars_per_year argument.

    View source ↗
  5. 6y ago

    CRAN example fixes and vignette references

    CRAN example fixes and a vignette revision adding references on the dangers of backtesting. The add_performance() function is listed here and again in 0.3.0.

    View source ↗
  6. 6y ago

    Drawdown, cumulative return and stacked bar charts

    Three chart functions arrive — cumulative returns, drawdown and stacked bar — alongside fixes for global variables under parallel execution and for leaderboards where a portfolio fails on every dataset.

    View source ↗