0.0.77

0.0.77

• Fixed issue #467 - CVaR calculation returning NaN for DataFrame inputs:

  • The conditional_value_at_risk() function now properly handles DataFrame inputs
  • When filtering DataFrames, NaN values are now correctly removed before calculating the mean
  • CVaR calculations are now consistent between Series and DataFrame inputs
  • This fix ensures accurate risk metrics in HTML reports when using benchmarks

• Confirmed issue #468 is already resolved:

  • The "mode.use_inf_as_null" pandas option error reported in v0.0.64 no longer occurs
  • This issue was resolved in a previous version through updates to pandas compatibility

What's Changed

• Fix CVaR calculation for DataFrames and confirm resolution of mode.use_inf_as_null error by @ranaroussi in #471

Full Changelog: 0.0.76...0.0.77

0.0.76

0.0.76

• Fixed issue #457 - Inconsistent benchmark EOY returns in reports:

  • Benchmark yearly returns now remain consistent regardless of strategy's trading calendar
  • HTML reports preserve original benchmark data for accurate EOY calculations
  • Previously, benchmark returns would change when aligned to different strategies' trading days
  • Added comprehensive tests to verify benchmark consistency across different comparisons

• Improved timezone handling for cross-market comparisons:

  • All resampling operations now normalize timezones to prevent comparison errors
  • Mixed timezone-aware and timezone-naive data can now be compared without errors
  • Data is converted to UTC then made timezone-naive for consistent comparisons
  • Fixes "Cannot compare dtypes datetime64[ns] and datetime64[ns, UTC]" errors

• Fixed FutureWarning for pandas pct_change():

  • Updated all pct_change() calls to use fill_method=None parameter
  • Prevents "fill_method='pad' is deprecated" warnings in pandas 2.x
  • Ensures compatibility with future pandas versions

0.0.75

0.0.75

• Fixed FutureWarning for deprecated pandas frequency aliases:
  • Updated make_index default rebalance parameter from "1M" to "1ME"
  • Ensures compatibility with pandas 2.2.0+ without warnings
  • The _compat module already handles conversion for older pandas versions

Full Changelog: 0.0.73...0.0.75

0.0.74

Updates:

• Completed fix for issue #463 - DataFrame handling in qs.reports functions:
  • kelly_criterion: Fixed improper use of 'or' operator with Series values
    • Now properly detects Series vs scalar inputs
    • Handles zero and NaN values correctly for DataFrames
  • recovery_factor: Added proper DataFrame input handling
    • Detects when max_dd is a Series and handles accordingly
    • Prevents "truth value of Series is ambiguous" errors
  • All functions now tested with qs.reports.html() and qs.reports.metrics() with benchmarks
  • Verified working with exact code examples from issue reporters

Full Changelog: 0.0.73...0.0.74

0.0.73

Addressed issue #463
Full Changelog: 0.0.72...0.0.73

0.0.72

• Fixed ValueError "truth value of Series is ambiguous" for DataFrame inputs in multiple stats functions:
• sortino: Properly handles Series downside deviation from DataFrame inputs
• outlier_win_ratio: Handles Series positive_mean calculations correctly
• outlier_loss_ratio: Handles Series negative_mean calculations correctly
• risk_return_ratio: Handles Series standard deviation properly
• ulcer_performance_index: Handles Series ulcer index values
• serenity_index: Handles Series std and denominator calculations
• gain_to_pain_ratio: Handles Series downside calculations
• All functions now properly return Series for DataFrame inputs and scalars for Series inputs

0.0.71

Added comprehensive divide by zero protection across multiple stats functions

Full Changelog: 0.0.70...0.0.71

0.0.70

What's Changed

• Fixed chart naming inconsistency: renamed "Daily Active Returns" to "Daily Active Returns (Cumulative Sum)" and "Daily Returns" to "Daily Returns (Cumulative Sum)" to accurately reflect that charts show cumulative values (fixes issue #454)
• Fixed CAGR calculation bug where years were incorrectly calculated using calendar days instead of trading periods, causing drastically reduced CAGR values (fixes issue #458)
• Fixed inconsistent EOY returns for benchmarks by preserving original benchmark data for aggregation while aligning to strategy index for other calculations (fixes issue #457)

Full Changelog: 0.0.69...0.0.70

Version 0.0.69

What's Changed

• Added periods parameter to calmar() function to support custom annualization periods (fixes #455)
• Updated reports.py to pass periods parameter to Calmar ratio calculation for consistency with other metrics

Bug Fixes

• Fixed issue #455: Calmar ratio periods parameter is missing

Full Changelog: 0.0.68...0.0.69

0.0.68

What's Changed

• Fix ValueError when comparing Series with scalar (Issue #448) by @ranaroussi in #450

Full Changelog: 0.0.67...0.0.68