Changelog

This page summarizes the major changes in each release of backtesting.py. For the full history, see the CHANGELOG.md on GitHub.

​

0.x.x (upcoming)

In development. See the master branch for unreleased changes.

---

​

0.6.5

Released: 2025-07-30

• Added a Commission column to stats._trades DataFrame.
• Commission is now accounted for in Trade.pl and Trade.pl_pct.
• Plotting: Long/short entry/exit triangles are back in the P&L section.
• Plotting: Indicators declared with plot=False, overlay=True are now rendered in a muted style rather than hidden.

Bug fixes:

• Fixed commission calculation when specified as a relative amount.
• Fixed a case where the SL value in stats._trades was occasionally cleared.
• Ensured order size is always an integer to prevent rounding errors.
• Fixed functools.partial objects that lack a __module__ attribute in Python 3.9.

---

​

0.6.4

Released: 2025-03-30 Bug fixes:

• Fixed optimization hanging on Windows when a if __name__ == '__main__' guard was missing.
• Restored original price scale in FractionalBacktest plots.
• Fixed "'CAGR [%]' must match a key in pd.Series result of bt.run()" error.
• Fixed grid optimization on data with timezone-aware DatetimeIndex.

---

​

0.6.3

Released: 2025-03-11 Enhancements:

• backtesting.lib.TrailingStrategy now supports setting the trailing stop-loss by percentage.
• New backtesting.lib.MultiBacktest wrapper for multi-dataset backtesting.
• Backtest.run() is now wrapped in tqdm() for progress reporting.
• lib.FractionalBacktest(fractional_unit=) parameter renamed for clarity.
• Added market alpha and market beta statistics.

Plot improvements:

• Trade duration lines are now shown in the P&L section.
• Simplified P&L section using circular markers.
• Trades are only plotted when at least one trade is present.
• Indicator subplots use desired_num_ticks=3 for cleaner axes.
• Single legend item for indicators with a singular/default name.
• OHLC candlestick series is now a togglable legend item.
• Added xwheel_pan tool (activation-conditioned).
• Reduced default height of indicator charts; overridable via backtesting._plotting._INDICATOR_HEIGHT.

Bug fixes:

• Fixed Position.pl and Position.pl_pct occasionally disagreeing in sign.
• SL now always executes before TP when both are hit in the same bar.
• Fixed functools.partial missing __module__ attribute in Python 3.9.
• Fixed stop-market and TP hit in the same bar.

---

​

0.6.2

Released: 2025-02-19 Enhancements:

• Grid optimization now uses mp.Pool and mp.shm.SharedMemory for faster parallel execution.
• New backtesting.lib.FractionalBacktest class supporting fractional share/unit trading.
• backtesting.__all__ defined for cleaner from backtesting import * usage and IDE suggestions.

Bug fixes:

• Fixed remaining issues with trade_on_close=True.
• Fixed trades reported in reverse chronological order when finalize_trades=True.
• Fixed crosshair not linked across subplots.
• Fixed a Windows-specific error with datetime_arr.astype(np.int64).

---

​

0.6.1

Released: 2025-02-04

• Switched to joblib.Parallel for optimization. This significantly improves performance on Windows and has minimal impact on other platforms.

---

​

0.6.0

Released: 2025-02-04 Enhancements:

• New Backtest(spread=) parameter for spread/slippage applied at entry only.
• Commissions are now shown as "Commissions [$]" in trade stats.
• Multiple names can be assigned to vector indicators.
• stats['trades'] now includes SL and TP columns.
• stats['trades'] now includes entry and exit indicator values.
• New finalize_trades= option in Backtest.run() to close open trades at backtest end.

Bug fixes:

• Fixed Sharpe ratio calculation with a non-zero risk-free rate.
• Price comparisons now use <= / >= to align with TradingView behavior.
• Reduced memory footprint during optimization.
• Fixed annualized statistics with weekly or monthly data.
• Fixed AssertionError when cancelling orders in a loop.
• Fixed plot not rendering in VSCode Jupyter.
• Buy & Hold duration now matches trading duration.
• Fixed bt.plot(resample=True) with categorical indicators.

---

​

0.5.0

Released: 2025-01-21 Enhancements:

• New Backtest.optimize(method="sambo") using SAMBO, replacing the previous method="skopt".
• New 'CAGR [%]' (compound annual growth rate) statistic in run results.

Bug fixes:

• Fixed “stop-loss executed at a higher than market price” error.
• Fixed a bug with buy()/sell() when size=0.
• Fixed Order.__repr__ when Order.tag is non-numeric.

---

​

0.4.0

Released: 2025-01-21 Enhancements:

• New 'Kelly Criterion' statistic.
• New Backtest.plot(plot_trades=) parameter to toggle trade markers.
• New Order.tag for tracking individual orders and trades.

---

​

0.3.3

Released: 2021-12-13

• Fixed random data generation with recent NumPy versions.
• Fixed Pandas deprecation warnings.
• Replaced Bokeh 3.0 deprecated APIs.

---

​

0.3.2

Released: 2021-08-03

• New backtesting.lib.compute_stats() method for computing strategy performance outside of Backtest.run().
• Improved plotting speed and optimization performance on large datasets.
• Commission constraints now support negative values (market-maker rebates).
• Backtest.plot() now returns the Bokeh figure object for further customization.

---

​

0.3.1

Released: 2021-01-25

• Avoided pandas.Index deprecation warnings.
• Fixed Backtest.plot(show_legend=False) compatibility with recent Bokeh.

---

​

0.3.0

Released: 2020-11-24

• Faster model-based optimization using scikit-optimize.
• Optional randomized grid search for faster optimization.
• Annualized Return, Volatility, Sharpe, Sortino, and Calmar statistics.
• Open trades are now automatically closed at the end of a backtest run.
• New Backtest.plot(plot_return=) parameter, analogous to plot_equity=.
• Updated Expectancy formula.

---

​

0.2.4

Released: 2020-10-27

• New lib.random_ohlc_data() OHLC data generator for testing.
• Equity is now aggregated on 'last' during plot resampling.
• Updated Buy & Hold stats calculation to be long-only.

---

​

0.2.3

Released: 2020-09-10

• Linked hover crosshairs across all subplots.
• Clicking a legend glyph now toggles indicator visibility.
• Fixed Bokeh tooltip showing literal  .

---

​

0.2.2

Released: 2020-08-21 Internal maintenance release.

---

​

0.2.1

Released: 2020-08-03

• Added Trade.entry_time and Trade.exit_time attributes.
• Fixed SL/TP hit on the same day a position was opened.

---

​

0.2.0

Released: 2020-07-15

• New Order, Trade, and Position API.
• New pandas accessors .df and .s on Strategy.data.
• New Backtest(exclusive_orders=) parameter: closes previous trades on new orders.
• New Backtest(hedging=) parameter: makes FIFO trade closing optional.
• New bt.plot(reverse_indicators=) parameter.
• New bt.plot(resample=) with automatic downsampling for large datasets.
• Geometric mean return used in Sharpe/Sortino computation.

---

​

0.1.8

Released: 2020-07-14

• Added Profit Factor statistic.

---

​

0.1.7

Released: 2020-03-23

• Fixed support for 2-D indicators.
• Fixed tooltip Date field formatting with Bokeh 2.0.0.

---

​

0.1.6

Released: 2020-03-09 Internal maintenance release.

---

​

0.1.5

Released: 2020-03-02 Internal maintenance release.

---

​

0.1.4

Released: 2020-02-25 Internal maintenance release.

---

​

0.1.3

Released: 2020-02-24

• Show number of trades on OHLC plot legend.
• Added agg= parameter to lib.resample_apply().
• Reset position price and related state after closing a position.
• Fixed pandas insertion error on Windows.

---

​

0.1.2

Released: 2019-09-23

• Plot now spans 100% of browser width.

---

​

0.1.1

Released: 2019-09-23

• Avoided multiprocessing issues on Windows.
• Added scatter plot indicator support.

---

​

0.1.0

Released: 2019-01-15 Initial public release of backtesting.py.