MLFinLab 2.1.0: Bar Indexing and Volatility Estimator Changes

Summary

This release announcement describes changes to MLFinLab, a toolkit for developing machine learning based trading systems. Bar generation now returns timestamps as a DataFrame index, aligning its output with downstream functions and avoiding manual index conversion. The volatility estimators now select OHLC inputs by lowercase column names instead of relying on column order, which the release says could previously produce incorrect results without warning. Users may need to adjust column names and remove code that resets the timestamp index.

Other changes include Python 3.9 support and a required probability-scoring flag for cross-validation, intended to prevent confusion when a scoring function needs predicted probabilities. The announcement gives examples of the revised data format and estimator inputs, but it does not provide performance comparisons or empirical evidence about trading outcomes. These are software interface and reliability notes; they explain expected code migration rather than introduce or validate a trading method.

Key ideas

• Bar generation now places timestamps in the DataFrame index automatically.
• Volatility estimators identify OHLC data by lowercase column names rather than implicit column order.
• The change may require renaming input columns and removing manual timestamp indexing.
• The cross-validation scoring function now requires users to specify whether probability predictions are needed.
• The release describes compatibility and usability changes, not evidence of improved trading performance.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.