def _test_get_historical_data(self):
# product_code = 'ICE/TFM'
product_code = 'ICE/BPB'
start = datetime.date(2016, 1, 1)
end = datetime.date(2017, 12, 1)
date = start
step = relativedelta(months=1)
while date < end:
year = date.year
month = date.month
date = date + step
print("Date", date)
# continue
month_code = quandl_month_codes[month]
symbol = '{}{}{}'.format(product_code, month_code, year)
quotes = get_historical_data(service='quandl',
sym=symbol,
start=datetime.datetime(2010, 1, 1),
end=datetime.datetime.now())
quotes_settle = quotes['Settle']
num_quotes = len(quotes_settle)
last_price = pick_last_price(quotes_settle)
vol = calc_historical_volatility(quotes_settle)
print(symbol, year, month, num_quotes, last_price, vol)
def _test_get_quandl_data_ttf(self):
quotes = get_historical_data(service='quandl',
sym='ICE/TFMF2014',
start=datetime.datetime(2013, 1, 1),
col='Settle')
index = quotes.index
self.assertIsInstance(index[0], Timestamp)
self.assertEqual(len(quotes), 60)
def _test_get_quandl_data_goog(self):
quotes = get_historical_data('quandl',
'GOOG',
col='Close',
end=datetime.date(2017, 10, 26))
index = quotes.index
self.assertIsInstance(index[0], Timestamp)
self.assertTrue(len(quotes), 23)
def _test_get_quandl_data_wti(self):
quotes = get_historical_data(
service='quandl',
sym='ICE/TX2009',
start=datetime.datetime(2007, 1, 1),
end=datetime.datetime(2007, 2, 1),
col='Settle',
)
index = quotes.index
self.assertIsInstance(index[0], Timestamp)
self.assertEqual(len(quotes), 23)
def generate_calibration_params(start,
end,
markets,
get_historical_data=get_historical_data):
name = 'quantdsl.priceprocess.blackscholes.BlackScholesPriceProcess'
all_quotes = []
sigmas = []
date = start
all_market_names = []
# Iterate over all "markets" (e.g. 'GAS').
all_curves = {}
for market_name, market_spec in markets.items():
all_market_names.append(market_name)
forward_vols = []
forward_curve = []
while date <= end:
forward_year = date.year
forward_month = date.month
kwargs = market_spec.copy()
kwargs['sym'] += quandl_month_codes[forward_month] + str(
forward_year)
# Get the data.
quotes = get_historical_data(**kwargs)
all_quotes.append(quotes)
vol = calc_historical_volatility(quotes)
forward_vols.append(vol)
last = pick_last_price(quotes)
forward_curve.append((date, last))
# Next month.
date = date + relativedelta(months=1)
sigma = np.median(forward_vols)
sigmas.append(sigma)
all_curves[market_name] = forward_curve
# Todo: Align dates and drop rows that aren't full.
rho = calc_correlation(all_quotes)
return {
'name': name,
'market': all_market_names,
'sigma': sigmas,
'rho': rho,
'curve': all_curves,
}
def generate_calibration_params(start, end, markets, get_historical_data=get_historical_data):
name = 'quantdsl.priceprocess.blackscholes.BlackScholesPriceProcess'
all_quotes = []
sigmas = []
date = start
all_market_names = []
# Iterate over all "markets" (e.g. 'GAS').
all_curves = {}
for market_name, market_spec in markets.items():
all_market_names.append(market_name)
forward_vols = []
forward_curve = []
while date <= end:
forward_year = date.year
forward_month = date.month
kwargs = market_spec.copy()
kwargs['sym'] += quandl_month_codes[forward_month] + str(forward_year)
# Get the data.
quotes = get_historical_data(**kwargs)
all_quotes.append(quotes)
vol = calc_historical_volatility(quotes)
forward_vols.append(vol)
last = pick_last_price(quotes)
forward_curve.append((date, last))
# Next month.
date = date + relativedelta(months=1)
sigma = np.median(forward_vols)
sigmas.append(sigma)
all_curves[market_name] = forward_curve
# Todo: Align dates and drop rows that aren't full.
rho = calc_correlation(all_quotes)
return {
'name': name,
'market': all_market_names,
'sigma': sigmas,
'rho': rho,
'curve': all_curves,
}
def _test_get_google_data_goog(self):
# NB start and end doesn't seem to be effective with the 'google' service.
quotes = get_historical_data('google', 'GOOG', col='Close', limit=30)
index = quotes.index
self.assertIsInstance(index[0], Timestamp)
self.assertEqual(len(quotes), 30, str(quotes))