iniself/noba

Noba not only backtrader as a quantitative investment platform, but also visualized using bokeh, which can get richer plot effects. Additionally, Noba also provide 'Ioc Container', 'Event System', 'Database Abstraction Layer', 'Pipeline System' and more.

What is Noba

Noba means not only backtrader :)

You can visit noba documentation for more information: EN(coming soon) | 中文

The core of Noba is an ioc container, through which you can create BB service, which based on Backtrader(one quantitative backtest system) and Bokeh (use bokeh as the backend, Backtrader can get richer plot effects) * BB service would like to thank backtrader_plotting and btplotting for providing the main code for using bokeh as the backend for the backtrader

from noba import core
bt =  core.make('bb')

Of course, you can also create your own services based container. Combined with Pipeline System and Event System (which can be created directly through containers), noba can enable your quantitative projects to work in a more engineering methods

from noba import core
pipline =  core.make('pipeline')
cleaned_data = pipline.via("handle").send(raw_data).through([ChangeDataType, RepeatRowData, ExceptionData, MissingData]).then(lambda raw_data:raw_data)

from noba import core
event = core.make('event')
db_event = event.hub(['read_database_complete'])
db_event.watch('read_database_complete', lambda data:..., always=True)
...
db_event.fire('read_database_complete')

More importantly, Noba can create database service objects (dber) through containers. This is a Database Abstraction Layer. Through configuration files(one json file) and unified one set of APIs, you can operate the most common databases

from noba import core
dber =  core.make('db')
stocks = db.table('daily').where('Open==3578.73').or_where('High==3652.46').set_index('Date').get_except('OpenInterest')

Getting Started

• Python >= 3.6 is required.
• Suggest using conda to manage virtual environments

Installation

pip install noba
# or
pip install git+https://github.com/iniself/noba

Init noba project

mkdir your_noba_project
cd your_noba_project
noba init

Preparation

Here you can do some project configuration and write your own service provider, and so on. Please refer to the NOBA documentation for details

Write strategy

vim main.py

Only give Live Mode example, about Normal Mode and Optstrategy Mode pls refer to NOBA documentation

• Add to cerebro as an analyzer (Live Mode):

  from noba import core
    ...
    ...
  bt = core.make('bb')
  cerebro = bt.Cerebro()
  cerebro.addstrategy(MyStrategy)
  cerebro.adddata(LiveDataStream()) # Note! Data is must Live Data
  cerebro.addanalyzer(bt.analyzers.Live, force_plot_legend=True, autostart=True)
  cerebro.run()
  # cerebro.plot() # do not run this line unless your data is not real-time

• If you need to change the default port or share the plotting to public:

  cerebro.addanalyzer(bt.analyzers.Live, address="localhost", port=8889)

• Note! In Jupyter you can plut to a single browser tab with iplot=False:

  cerebro.plot(plot, iplot=False)

Demos

https://iniself.github.io/noba/

Contact us

Telegram Channel: Aui_Say Discord Server: Aui and Friends

Sponsoring

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

• ETH: 0x0275779f70179748C6fCe1Fe5D7638DfA7e3F986

noba's Issues

python10无法安装noba，在安装依赖matplotlib3.2.2时出错。

是不是依赖的matplotlib版本太低了，不支持python10。

ERROR:bokeh.server.views.ws:Refusing websocket connection.

Hello! I got the following error in notebook. I tried both VS Code and Jupyter Notebook - same error.

Appreciate any solution!

ERROR:bokeh.server.views.ws:Refusing websocket connection from Origin 'vscode-webview://0lj9hh483va927cadklatiu5affjj5cn3b349he4cl71d36qc4p6/'; use --allow-websocket-origin=0lj9hh483va927cadklatiu5affjj5cn3b349he4cl71d36qc4p6 or set BOKEH_ALLOW_WS_ORIGIN=0lj9hh483va927cadklatiu5affjj5cn3b349he4cl71d36qc4p6 to permit this; currently we allow origins {'localhost:8889'} WARNING:tornado.access:403 GET /ws?id=348b26a3-e939-4e81-b383-da0df5147ff8&origin=bce4bbf1-101d-44db-883b-e3db4f31e0a2&swVersion=4&extensionId=&platform=electron&vscode-resource-base-authority=vscode-resource.vscode-cdn.net&parentOrigin=vscode-file%3A%2F%2Fvscode-app&purpose=notebookRenderer (::1) 18.06ms

'WindowsPath' object is not subscriptable

Throw an exception “'WindowsPath' object is not subscriptable” when “from noba import core”

in windows python 3.9

回测在没有界面的centos下面进行，有什么办法能把回测结果用一个网址放出来展示分析吗？

rt，谢谢指导！

Empty output (no graph) in Jupyter Lab

When running the sample:

from backtrader_bokeh import bt as bt_bokeh
plot = bt_bokeh.Bokeh()
cerebro.plot(plot, iplot=False)

there is nothing displayed in the Jupyter Lab cell output except: " BokehJS 2.4.3 successfully loaded."

Versions of packages:

• backtrader-bokeh==0.8.9
• bokeh==2.4.3
• jupyter-bokeh==3.0.7
• jupyter-lab: 3.4.8

Is there something else I may be missing?

加载多只股票的数据时，cerebro.plot渲染会报错