Simple strategy backtesting using Zipline

Zipline is a backtesting engine for Python, if you’re a Quantopian member you should be familiar with it since it’s the one they’re using. It provides metrics about the strategy such as returns, standard deviations, Sharpe ratios etc. basically everything you need to know in order to validate or not a strategy before going live.

Zipline can be install using pip:

pip install zipline

If you’re on Windows I suggest using Conda:

conda install -c Quantopian zipline

Here is the basic structure of a strategy in Zipline:

from zipline.api import order, record, symbol

def initialize(context): 
    pass

def handle_data(context, data): 
    order(symbol('AAPL'), 10)
    record(AAPL=data.current(symbol('AAPL'), 'price'))

In initialize you can set some global variables used for the strategy such as a list of stocks, certain parameters, the maximum percentage of portfolio invested.
Then handle_data is entered at every tick, that’s where your strategy logic should be. You can check previous articles and incorporate strategies into your code.

Let’s breakdown the handle_data() code.

The order() function let you create an order, here we specify the AAPL ticker (Apple stock) with a quantity of 10. A positive value means you’re buying 10 stocks, a negative value would mean you’re selling the stock.

Then, the record() function allows you to save the value of a variable at each iteration. Here, you’re saving the current stock price under the variable named AAPL, you’ll then be able to retrieve that information in the backtest result, this way you can compare your strategy performance versus the stock price.

Now you want to finally backtest the strategy and see if it’s profitable. To do that, run the following command:

zipline run -f your_strategy.py --start 2015-1-1 --end 2020-1-1 -o your_strategy.pickle

This command is going to run the backtest between 2015-01-01 and 2020-01-01 and output the result into a pickle file for later analysis. The pickle is simply a Pandas DataFrame with a line per day and (a lot of) columns regarding your strategy, such as the return, the number of orders, the portofolio size and so on.