Skip to content

Latest commit

 

History

History
121 lines (86 loc) · 3.8 KB

README.md

File metadata and controls

121 lines (86 loc) · 3.8 KB
Raw

btplotting

Library to add extended plotting capabilities to backtrader (https://www.backtrader.com/) using bokeh.

btplotting is based on the awesome backtrader_plotting (https://github.com/verybadsoldier/backtrader_plotting)

btplotting is a complete rework of backtrader_plotting with the live client in focus. Besides this, a lot of issues are fixed and new functionality is added. See the list below for differences.

What is different:

Basic:

  • No need for custom backtrader
  • Different naming / structure
  • Data alignment which allows to generate data for different data sources. This is useful when replaying or resampling data, for example to remove gaps.
  • Support for replay data
  • Different filtering of plot objects
  • Every figure has its own ColumnDataSource, so the live client can patch without having issues with nan values, every figure is updated individually
  • Plotting looks very similar to backtraders own plotting (order, heights, etc.)
  • Allows to generate custom columns, which don't have to be hardcoded.
  • This is being used to generate color for candles, varea values, etc.
  • Save images of strategy or a single data (for example save an image of data when a trade is happening)

Plotting:

  • Datas, Indicators, Observer and Volume have own aspect ratios, which can be configured in live client or scheme
  • Only one axis for volume will be added when using multiple data sources on one figure
  • Volume axis position is configureable in scheme, by default it is being plotted on the right side
  • Linked Crosshair across all figures
  • _skipnan, fill_gt, fill_lt, fill support
  • Plot objects can be filtered by one or more datanames or by plot group
  • Custom plot group, which can be configured in app or in live client by providing all plotids in a comma-separated list or by selecting the parts of the plot to display

Tabs:

  • Default tabs can be completely removed
  • New log panel to also include logging information
  • Can be extended with custom tabs (for example order execution with live client, custom analysis, etc.)

Live plotting:

(Live plotting is broken at the moment)

  • Navigation in live client (Pause, Backward, Forward)
  • Live plotting is done using an analyzer, so there is no need to use custom backtrader
  • Live plotting data update works in a single thread and is done by a DataHandler
  • Data update is being done every n seconds, which is configureable

Features

  • Interactive plots
  • Interactive backtrader optimization result browser (only supported for single-strategy runs)
  • Highly configurable
  • Different skinnable themes
  • Easy to use

Python >= 3.6 is required.

How to use

  • Add to cerebro as an analyzer:
from btplotting import BacktraderPlottingLive
  ...
  ...

cerebro = bt.Cerebro()
cerebro.addstrategy(MyStrategy)
cerebro.adddata(LiveDataStream())
cerebro.addanalyzer(BacktraderPlottingLive)
cerebro.run()
cerebro.plot()
  • If you need to change the default port or share the plotting to public:
cerebro.addanalyzer(BacktraderPlottingLive, address="*", port=8889)

Jupyter

In Jupyter you can plot to a single browser tab with iplot=False:

plot = btplotting.BacktraderPlotting()
cerebro.plot(plot, iplot=False)

You may encounters TypeError: <class '__main__.YourStrategyClass'> is a built-in class error.

To remove the source code tab use:

plot = btplotting.BacktraderPlotting()
plot.tabs.remove(btplotting.tabs.SourceTab)
cerebro.plot(plot, iplot=False)

Demos

https://happydasch.github.io/btplotting/

Installation

pip install git+https://github.com/happydasch/btplotting

Sponsoring

If you want to support the development of btplotting, consider to support this project.