class TestBMonthBegin(Base):
_offset = BMonthBegin
def test_offsets_compare_equal(self):
# root cause of #456
offset1 = BMonthBegin()
offset2 = BMonthBegin()
assert not offset1 != offset2
offset_cases = []
offset_cases.append((BMonthBegin(), {
datetime(2008, 1, 1): datetime(2008, 2, 1),
datetime(2008, 1, 31): datetime(2008, 2, 1),
datetime(2006, 12, 29): datetime(2007, 1, 1),
datetime(2006, 12, 31): datetime(2007, 1, 1),
datetime(2006, 9, 1): datetime(2006, 10, 2),
datetime(2007, 1, 1): datetime(2007, 2, 1),
datetime(2006, 12, 1): datetime(2007, 1, 1)}))
offset_cases.append((BMonthBegin(0), {
datetime(2008, 1, 1): datetime(2008, 1, 1),
datetime(2006, 10, 2): datetime(2006, 10, 2),
datetime(2008, 1, 31): datetime(2008, 2, 1),
datetime(2006, 12, 29): datetime(2007, 1, 1),
datetime(2006, 12, 31): datetime(2007, 1, 1),
datetime(2006, 9, 15): datetime(2006, 10, 2)}))
offset_cases.append((BMonthBegin(2), {
datetime(2008, 1, 1): datetime(2008, 3, 3),
datetime(2008, 1, 15): datetime(2008, 3, 3),
datetime(2006, 12, 29): datetime(2007, 2, 1),
datetime(2006, 12, 31): datetime(2007, 2, 1),
datetime(2007, 1, 1): datetime(2007, 3, 1),
datetime(2006, 11, 1): datetime(2007, 1, 1)}))
offset_cases.append((BMonthBegin(-1), {
datetime(2007, 1, 1): datetime(2006, 12, 1),
datetime(2008, 6, 30): datetime(2008, 6, 2),
datetime(2008, 6, 1): datetime(2008, 5, 1),
datetime(2008, 3, 10): datetime(2008, 3, 3),
datetime(2008, 12, 31): datetime(2008, 12, 1),
datetime(2006, 12, 29): datetime(2006, 12, 1),
datetime(2006, 12, 30): datetime(2006, 12, 1),
datetime(2007, 1, 1): datetime(2006, 12, 1)}))
@pytest.mark.parametrize('case', offset_cases)
def test_offset(self, case):
offset, cases = case
for base, expected in compat.iteritems(cases):
assert_offset_equal(offset, base, expected)
on_offset_cases = [(BMonthBegin(), datetime(2007, 12, 31), False),
(BMonthBegin(), datetime(2008, 1, 1), True),
(BMonthBegin(), datetime(2001, 4, 2), True),
(BMonthBegin(), datetime(2008, 3, 3), True)]
@pytest.mark.parametrize('case', on_offset_cases)
def test_onOffset(self, case):
offset, dt, expected = case
assert_onOffset(offset, dt, expected)
def BackTest_1_Beta(self):
d = self.Dates[0]
for i in xrange(
int(
self.Periods_Diff(self.Dates[0], self.Dates[1]) /
self.Frequency)):
# Compute the date for historical data extract
d_f = d - BMonthBegin() * self.Histo_Length
# Extract the Benchmark index returns
bnch = self.Benchmark.Extract_Index_Returns([d_f, d])
returns = self.Benchmark.Extract_Returns([d_f, d], d - MonthEnd())
assets = self.Benchmark.Extract_Compo(d - MonthEnd())
std_Index = np.std(bnch)
beta = np.empty(len(assets))
for i in xrange(len(beta)):
# Delete missing value in the estimation
beta[i] = bnch.iloc[:, 0].cov(returns.iloc[:, i]) / std_Index
self.RB_Weighted_1_risk(beta)
bckData = self.Benchmark.Extract_Returns(
[d + BDay(), (d + BDay()) + BMonthEnd() * self.Frequency],
d - MonthEnd()).loc[:, assets]
self.Compute_Performance(bckData)
self.Wght_Histo[d] = pd.DataFrame(self.Weights, index=assets)
d = (d + BMonthEnd() * self.Frequency) + BDay()
def build_prediction_model():
boe = load_boe_history()
shb = load_shb_history()
# filter BOE data to match SHB date range
shb_min = shb['valid_from'].min()
shb_max = shb['valid_from'].max()
mask = (boe['Date'] >= shb_min) & (boe['Date'] <= shb_max)
boe = boe.loc[mask]
# reduce BOE data to start-of-month values
boe['fom'] = boe['Date'].apply(lambda x: BMonthBegin().rollback(x))
boe = boe[boe['Date'] == boe['fom']]
# add in monthly changes
for term in TERMS:
boe[f'{term}ydiff'] = boe[f'{term}y'].diff()
shb[f'{term}diff'] = shb[term].diff()
df = boe.merge(shb, on='period', how='inner')
# drop the first row - which will contain NaN
df = df.iloc[1:-1]
for term in TERMS:
X = df[[f'{term}ydiff']]
y = df[[f'{term}diff']]
# using default 5-split
tss = TimeSeriesSplit(n_splits=5).split(X)
model = LinearRegression().fit(X, y)
scores = cross_val_score(model, X, y, cv=tss)
r2 = scores.mean()
store(term, model, 'sklearn_LinearRegression_CV', r2)
def _initial_fx_load(self, start_date = '2011-06-01', end_date = '2016-09-01'):
self.db_method = 'replace'
date = pd.to_datetime(start_date)
stop_date = pd.to_datetime(end_date) - BMonthBegin() + BMonthBegin()
date_list = [date.strftime('%Y-%m-%d')]
while date != stop_date:
date = date + BMonthBegin()
date_list.append(date.strftime('%Y-%m-%d'))
date_list.append('2012-01-03')
date_list.append('2013-01-02')
date_list.append('2014-01-02')
date_list.append('2015-01-02')
date_list.append('2016-01-04')
self._get_open_fx(date_list)
fx_dic = _get_open_fx(date_list)
_load_fx_to_db(fx_dic,db_method='replace')
return True
def BackTest_MinVar(self):
d = self.Dates[0]
for i in xrange(
int(
self.Periods_Diff(self.Dates[0], self.Dates[1]) /
self.Frequency)):
d_f = d - BMonthBegin() * self.Histo_Length
vcov, assets = self.Benchmark.Extract_VCov_Matrix([d_f, d],
d - MonthEnd())
self.MinVariance(vcov, 0.001)
bckData = self.Benchmark.Extract_Returns(
[d + BDay(), (d + BDay()) + BMonthEnd() * self.Frequency],
d - MonthEnd()).loc[:, assets]
self.Compute_Performance(bckData)
self.Wght_Histo[d] = pd.DataFrame(self.Weights, index=assets)
d = (d + BMonthEnd() * self.Frequency) + BDay()
def BackTest_1_Sigma(self):
d = self.Dates[0]
for i in xrange(
int(
self.Periods_Diff(self.Dates[0], self.Dates[1]) /
self.Frequency)):
d_f = d - BMonthBegin() * self.Histo_Length
Std, assets = self.Benchmark.Extract_Sigma([d_f, d],
d - MonthEnd())
self.RB_Weighted_1_risk(Std)
bckData = self.Benchmark.Extract_Returns(
[d + BDay(), (d + BDay()) + BMonthEnd() * self.Frequency],
d - MonthEnd()).loc[:, assets]
self.Compute_Performance(bckData)
self.Wght_Histo[d] = pd.DataFrame(self.Weights, index=assets)
d = (d + BMonthEnd() * self.Frequency) + BDay()
def chart_rate_moves():
boe = load_boe_history()
shb = load_shb_history()
# filter BOE data to match SHB date range
shb_min = shb['valid_from'].min()
shb_max = shb['valid_from'].max()
mask = (boe['Date'] >= shb_min) & (boe['Date'] <= shb_max)
boe = boe.loc[mask]
# reduce BOE data to start-of-month values
boe['fom'] = boe['Date'].apply(lambda x: BMonthBegin().rollback(x))
boe = boe[boe['Date'] == boe['fom']]
# add in monthly changes
for rate in TERMS:
boe[f'{rate}ydiff'] = boe[f'{rate}y'].diff()
shb[f'{rate}diff'] = shb[rate].diff()
df = boe.merge(shb, on='period', how='inner')
rows = 1
cols = 5
fig, axs = plt.subplots(rows,
cols,
figsize=(15, 4),
squeeze=False,
sharey=True)
for idx, rate in enumerate(TERMS):
row = idx % rows
col = idx // rows
ax = axs[row][col]
residual = df[f'{rate}diff'] - df[f'{rate}ydiff']
# ax.plot(df['Date'], residual, 'r+', label='residual')
# ax.plot(boe['Date'], boe[f'{rate}ydiff'], 'g', label='boe diff')
# ax.plot(shb['valid_from'], shb[f'{rate}diff'], 'c.', label='shb diff')
ax.plot(df[f'{rate}ydiff'], df[f'{rate}diff'], 'b+')
ax.set_title(f'{rate} year rate')
ax.legend()
plt.show()
'D' : Day(),
'B' : BDay(),
'H' : Hour(),
'T' : Minute(),
'S' : Second(),
'L' : Milli(),
'U' : Micro(),
None : None,
# Monthly - Calendar
'M' : MonthEnd(),
'MS' : MonthBegin(),
# Monthly - Business
'BM' : BMonthEnd(),
'BMS' : BMonthBegin(),
# Annual - Calendar
'A-JAN' : YearEnd(month=1),
'A-FEB' : YearEnd(month=2),
'A-MAR' : YearEnd(month=3),
'A-APR' : YearEnd(month=4),
'A-MAY' : YearEnd(month=5),
'A-JUN' : YearEnd(month=6),
'A-JUL' : YearEnd(month=7),
'A-AUG' : YearEnd(month=8),
'A-SEP' : YearEnd(month=9),
'A-OCT' : YearEnd(month=10),
'A-NOV' : YearEnd(month=11),
'A-DEC' : YearEnd(month=12),
'A' : YearEnd(month=12),
def test_offsets_compare_equal(self):
# root cause of #456
offset1 = BMonthBegin()
offset2 = BMonthBegin()
assert not offset1 != offset2
def test_quarterly_dont_normalize():
date = datetime(2012, 3, 31, 5, 30)
offsets = (QuarterBegin, QuarterEnd, BQuarterEnd, BQuarterBegin)
for klass in offsets:
result = date + klass()
assert (result.time() == date.time())
@pytest.mark.parametrize(
'offset',
[MonthBegin(), MonthEnd(),
BMonthBegin(), BMonthEnd()])
def test_apply_index(offset):
rng = pd.date_range(start='1/1/2000', periods=100000, freq='T')
ser = pd.Series(rng)
res = rng + offset
res_v2 = offset.apply_index(rng)
assert (res == res_v2).all()
assert res[0] == rng[0] + offset
assert res[-1] == rng[-1] + offset
res2 = ser + offset
# apply_index is only for indexes, not series, so no res2_v2
assert res2.iloc[0] == ser.iloc[0] + offset
assert res2.iloc[-1] == ser.iloc[-1] + offset
import pandas as pd
from pandas.tseries.offsets import BMonthBegin
from datetime import date
from datetime import datetime
#d = date.today()
#offset = BMonthBegin()
#offset.rollforward(d)
#offset.rollback(d)
offset = BMonthBegin()
def getFirstDay(d):
day = datetime.strptime(str(d), '%Y%m%d').date()
fd = offset.rollforward(day)
return fd
df = pd.read_csv("newticker.csv")
df['new'] = df['Month'].apply(lambda x: getFirstDay(x))
df.to_csv("NewFile.csv")
full_df = full_df.dropna()
print(full_df)
"""# Create list of tickers from volume screened df"""
volScreenList = list(full_df.columns)
print(volScreenList)
print(len(volScreenList))
"""# Create Prices Data frame from volume screened list and remove tickers with avg price <1"""
#Get dates
todayD = date.today()
offset = BMonthEnd(-1)
offsetFirst = BMonthBegin(-13)
# Get first mo. first day, last mo. last day
lastDayLastMonth = todayD+offset
firstDayFirstMonth = todayD+offsetFirst
# Append prices to tickers
prices_df = pd.DataFrame()
for ticker in volScreenList:
try:
prices_df[ticker] = historical_datas[ticker]['close'].loc[f"{firstDayFirstMonth.year}-{firstDayFirstMonth.month}-{firstDayFirstMonth.day}" : f"{lastDayLastMonth.year}-{lastDayLastMonth.month}-{lastDayLastMonth.day}"]
print(f"completed {ticker}")
except:
print(f"error with {ticker}")
