Backtesting a Moving Average Crossover Strategy¶
In this example we use get_pricing to load 10 years' worth of historical financial data for Apple's stock (i.e. the ticker symbol AAPL).
We then define a Dual Moving Average Crossover algorithm with zipline, the open source backtesting library that powers Quantopian.
Finally, we backtest our strategy against our loaded trade data and create visualization of our entry- and exitpoints.
In [1]:
# Import Zipline, the open source backester, and a few other libraries that we will use
import zipline
from zipline import TradingAlgorithm
from zipline.api import order_target, record, symbol, history, add_history
import pytz
from datetime import datetime
import matplotlib.pyplot as pyplot
import numpy as np
In [16]:
# Load data from get_trades for AAPL
data = get_pricing(
# ['AAPL'],
[symbols(24)],
start_date='2002-01-01',
end_date = '2015-02-15',
frequency='daily'
)
data.price.plot(use_index=False)
Out[16]:
In [17]:
data
Out[17]:
In [18]:
# Define the algorithm - this should look familiar from the Quantopian IDE
# For more information on writing algorithms for Quantopian
# and these functions, see https://www.quantopian.com/help
def initialize(context):
# Register 2 histories that track daily prices,
# one with a 100 window and one with a 300 day window
add_history(100, '1d', 'price')
add_history(300, '1d', 'price')
context.i = 0
context.aapl = symbol('AAPL')
def handle_data(context, data):
# Skip first 300 days to get full windows
context.i += 1
if context.i < 300:
return
# Compute averages
# history() has to be called with the same params
# from above and returns a pandas dataframe.
short_mavg = history(100, '1d', 'price').mean()
long_mavg = history(300, '1d', 'price').mean()
# Trading logic
if short_mavg[context.aapl] > long_mavg[context.aapl]:
# order_target orders as many shares as needed to
# achieve the desired number of shares.
order_target(context.aapl, 100)
elif short_mavg[context.aapl] < long_mavg[context.aapl]:
order_target(context.aapl, 0)
# Save values for later inspection
record(AAPL=data[context.aapl].price,
short_mavg=short_mavg[context.aapl],
long_mavg=long_mavg[context.aapl])
In [19]:
# Analyze is a post-hoc analysis method available on Zipline.
# It accepts the context object and 'perf' which is the output
# of a Zipline backtest. This API is currently experimental,
# and will likely change before release.
def analyze(context, perf):
fig = pyplot.figure()
# Make a subplot for portfolio value.
ax1 = fig.add_subplot(211)
perf.portfolio_value.plot(ax=ax1, figsize=(16,12))
ax1.set_ylabel('portfolio value in $')
# Make another subplot showing our trades.
ax2 = fig.add_subplot(212)
perf['AAPL'].plot(ax=ax2, figsize=(16, 12))
perf[['short_mavg', 'long_mavg']].plot(ax=ax2)
perf_trans = perf.ix[[t != [] for t in perf.transactions]]
buys = perf_trans.ix[[t[0]['amount'] > 0 for t in perf_trans.transactions]]
sells = perf_trans.ix[
[t[0]['amount'] < 0 for t in perf_trans.transactions]]
# Add buy/sell markers to the second plot
ax2.plot(buys.index, perf.short_mavg.ix[buys.index],
'^', markersize=10, color='m')
ax2.plot(sells.index, perf.short_mavg.ix[sells.index],
'v', markersize=10, color='k')
# Set figure metadata
ax2.set_ylabel('price in $')
pyplot.legend(loc=0)
pyplot.show()
In [20]:
# NOTE: This cell will take a few minutes to run.
# Create algorithm object passing in initialize and
# handle_data functions
algo_obj = TradingAlgorithm(
initialize=initialize,
handle_data=handle_data
)
# HACK: Analyze isn't supported by the parameter-based API, so
# tack it directly onto the object.
algo_obj._analyze = analyze
# Run algorithm
perf_manual = algo_obj.run(data.transpose(2,1,0))
In [21]:
import pyfolio as pf
In [23]:
returns, positions, transactions, gross_lev = pf.utils.extract_rets_pos_txn_from_zipline(perf_manual)
In [29]:
pf.create_bayesian_tear_sheet(returns, live_start_date='2011-01-01')
In [27]:
pf.create_full_tear_sheet(returns, positions=positions, transactions=transactions,
gross_lev=gross_lev, live_start_date='2011-01-01', bayesian=True)
In [ ]: