Stock market analysis library written in Python.
- Copyright (C) 2016-2017 by Chris Pappalardo <cpappala@gmail.com>
- License: Creative Commons (CC) BY-NC-ND 4.0
- Documentation: https://stockbot-lib.readthedocs.org/
- Live demo: https://stockbot.tech
- Market data sourcing from Yahoo!, CNBC, and zipline bundles
- S&P500 stock listing scraper
- ADX, DI, and Stochastic technical indicators implemented using TA-lib
- Average Directional Movement Index (ADX) ranking for portfolios
- Trending and oscillating instrument trading algorithms for zipline
Install the latest package with:
$ pip install stockbotThe dependencies are not trivial and may not install properly on your system through pip. We recommend developing and deploying your projects that use Stockbot in Docker containers with the necessary packages pre-installed.
To do this, first build the quantopian/zipline Docker image with the following command:
$ docker build -t quantopian/zipline https://github.com/quantopian/zipline.gitThen use a docker-compose configuration similar to Stockbot's production cluster here. Build your containers with:
$ docker-compose buildTo run Stockbot's trading algorithms in zipline and to use certain functions, you will need to download a quantopian-quandl bundle:
$ docker-compose run --rm stockbot zipline ingest -b quantopian-quandlStockbot can provide you with a list of S&P500 stocks from Wikipedia:
>>> from stockbot.core import get_sp500_list
>>> get_sp500_list()
[u'MMM', u'ABT', u'ABBV', u'ACN', u'ATVI', u'AYI', u'ADBE', ... u'ZTS']To get a delayed quote from Yahoo! use `get_yahoo_quote`:
>>> from stockbot.sources import get_yahoo_quote
>>> get_yahoo_quote('YHOO')
{'volume': 3405057, 'last': 41.0, 'symbol': 'YHOO', 'datetime': datetime.datetime(2016, 11, 22, 18, 0, tzinfo=<UTC>), 'high': 41.4, 'low': 40.83, 'open': 41.2, 'change': -0.11}Or a real-time quote from CNBC using `get_cnbc_quote`:
>>> from stockbot.sources import get_cnbc_quote
>>> next(get_cnbc_quote('YHOO'))
{'volume': 3528566, 'last': 41.04, 'symbol': u'YHOO', 'datetime': datetime.datetime(2016, 11, 22, 21, 0, tzinfo=<UTC>), 'high': 41.395, 'low': 40.83, 'open': 41.2, 'change': -0.07}Note:: get_cnbc_quote returns a generator.
Stockbot returns quote data using a dict like object stockbot.marketdata.MarketData that performs certain data and datetime processing.
Historical data can be obtained from Yahoo! using `get_yahoo_hist`:
>>> from stockbot.sources import get_yahoo_hist
>>> get_yahoo_hist('YHOO')
{'high': 41.48, 'last': 41.110001, 'datetime': datetime.datetime(2016, 11, 21, 21, 0, tzinfo=<UTC>), 'volume': 11338000, 'low': 40.939999, 'close': 41.110001, 'open': 41.439999}Historical data can also be obtained from zipline bundles using the get_zipline_hist function:
>>> from stockbot.sources import get_zipline_hist
>>> get_zipline_hist('YHOO', 'close', datetime.datetime(...))
2016-01-04 00:00:00+00:00 31.41
Freq: C, Name: Equity(3177 [YHOO]), dtype: float64Look up symbols with stockbot.sources.get_symbol which searches Yahoo! finance for the passed term.
Zipline trading algorithms that utilize the Directional Movement technical indicator system are provided in stockbot.algo. For example, the following zipline trading algorithm would use ADX and DI to trade the top trending stocks and Stochastic Oscillators to trade the top oscillating stocks in the S&P 500 index:
from logbook import Logger
from stockbot.algo.core import (
adx_init,
trade_di,
trade_so,
)
from stockbot.core import get_sp500_list
def initialize(context):
return adx_init(
context,
name='adx_di_so',
top_rank=5,
bot_rank=5,
di_window=14,
symbols=get_sp500_list(),
log=Logger('Stockbot'),
)
def handle_data(context, data):
# increment counter and log datetime
context.i += 1
context.adx['log'].info('processing %s' % context.get_datetime())
# trade trending S&P500 stocks using the DI system
trade_di(
context,
data,
window=context.adx['di_window'],
portfolio=[i for (i, adx) in context.adx['top']],
capital_ppi=1.0/(len(context.adx['top'])+len(context.adx['bot'])),
log=context.adx['log'],
)
# trade oscillating S&P500 stocks using the SO system
trade_so(
context,
data,
window=context.adx['di_window'],
portfolio=[i for (i, adx) in context.adx['bot']],
capital_ppi=1.0/(len(context.adx['top'])+len(context.adx['bot'])),
log=context.adx['log'],
)To run this algorithm in a docker container, copy the code above into a file and issue the following:
$ docker-compose run --rm stockbot zipline run -f <file> --start <date> --end <date>Use the the <YYYY-M-D> format for dates. Use -o /path/file.pickle to capture pickled results that can be used in python.