# This concept was inspired by QSTK:

# http://wiki.quantsoftware.org/index.php?title=QuantSoftware_ToolKit.

# However, no source code from that toolkit was used or consulted directly for

# this code. (The concept only)

from pylab import errorbar, xlabel, ylabel, show, legend

from numpy import array, arange

def event_profiler(asset, truth, periods=5):

cut = []

for i in range(periods, len(asset) - periods):

if truth[i] == 1 and asset[i] > 0:

cut.append(asset[i - periods:i + periods] / asset[i])

return array(cut)

def plot_event_profile(events, name=''):

mn = events.mean(axis=0) - 1.0

st = events.std(axis=0)

errorbar(arange(len(mn)), mn, st, label=name)

xlabel('Periods', fontsize=16)

ylabel('Price Change %', fontsize=16)

if len(name) > 1:

legend(loc='best', shadow=True, fancybox=True)

show()