A module containing useful packages for testing and writing trading algorithms. The idea of this package is to simulate a stock exchange allowing an algorithm to enter or leave positions. The user will be able to evaluate the performance of the algorithm using various tools available in this package.
To test a trading algorithm, we must first define an instance of the ‘Portfolio’ class:
from portfolio import *
pf = Portfolio(balance=50000)
This will define a new instance of the ‘Portfolio’ class. Normally we don’t need to specify the starting balance as it defaults to 50k on declaration.
The Portfolio class has a number of attributes associated with it, but the ones that we are most interested in are the buy, sell, and sell_all methods.
We can buy and sell stocks from the ‘Stock’ class. For now this is a bit messy, but will be taken care of for the user later in development.
New instances of the ‘Stock’ class require a Pandas data frame for their data. We’ll use the Apple Inc. stock data for this example:
import pandas as pd
from stock import *
aapl_data = pd.read_csv(‘AAPL_data.csv’, index_col=0)
aapl_stock = Stock(‘AAPL’, aapl_data)
Now with the Stock class defined, we can enter and exit different positions. The stock class uses instances of the ‘Point’ class to store each data point, so any time we want to enter a position we will need to convert one of them to an instance of either the Share or Option classes.
entry_datetime = aapl_data.history.keys()[0]
pf.buy(aapl_stock, 'long', entry_datetime, 10)
Now that we’ve entered a long position, we can check the portfolio balance, as well as its open positions, exposure, and its percentage holdings in each stock (which is accessed using get_holdings()).