def var_3d_surface(self):
# iteration for only working date in range of start and end date
holidays_PL = holidays.PL()
start_dt = pd.to_datetime(self.start_date).date()
end_dt = pd.to_datetime(self.end_date).date()
date_lst = []
var_3d_frame = pd.DataFrame(columns=[f'{n + 1}_day' for n in range(self.number_days_var)])
for i in range(int((end_dt - start_dt).days) + 1):
bd = start_dt + datetime.timedelta(i)
if bd in holidays_PL or bd.weekday() > 4:
continue
date_lst.append(bd)
# var_3d_frame['date'] = pd.to_datetime(date_lst)
# var_3d_frame.set_index('date', inplace=True)
# print(var_3d_frame)
capital = []
df_dict = {}
for x in date_lst:
reload = RiskAnalysis(start_date=self.start_date, end_date=self.end_date,
eval_date=str(x), number_days_var=self.number_days_var, show_plot=False)
df_dict[x] = reload.n_day_var()
capital.append(reload.df_pf.value_now.sum())
var_3d_frame = pd.DataFrame.from_dict(df_dict, orient='index', columns=var_3d_frame.columns)
fig = go.Figure(data=[go.Surface(z=var_3d_frame.values)])
fig.update_layout(title='VaR surface', autosize=False, height=800)
return [var_3d_frame, fig, capital]
def number_of_business_days_without_holidays(date_from: date,
date_to: date) -> int:
if date_from > date_to:
raise ValueError('date_from should be earlier than date_to')
result = 0
for day in pd.date_range(date_from, date_to):
if day not in holidays.PL() and not is_weekend(day):
result += 1
return result
def fit(self):
start = pd.Timestamp(self.start_day).normalize()
end = pd.Timestamp(self.end_day).normalize()
self.df = pd.DataFrame(index=pd.date_range(
start=start, end=end, freq='1H', normalize=True, closed='left'))
self.df['holiday'] = self.df.index.to_series().map(
lambda x: int(x in holidays.PL() or x.weekday() in [5, 6]))
self.df['peak7'] = self.df.index.to_series().map(
lambda x: int(x.hour in range(7, 22)))
self.df['peak5'] = ((self.df['holiday'] == 0) &
(self.df['peak7'] == 1)).map(int)
prediction_day = self.df.index + pd.Timedelta('1 days')
weekday = prediction_day.weekday
month = prediction_day.month
self.df = self.df.assign(weekday=weekday, month=month)
return self
def is_holiday_or_weekend(self):
return self.date in holidays.PL() or self.is_weekend()
def __init__(self) -> None:
self.holidays_calendar = holidays.PL()
country_holidays = {
"CA": holidays.CA(),
"CO": holidays.CO(),
"MX": holidays.MX(),
"US": holidays.US(),
"NZ": holidays.NZ(),
"AU": holidays.AU(),
"DE": holidays.DE(),
"AT": holidays.AT(),
"DK": holidays.DK(),
"UK": holidays.UK(),
"IE": holidays.IE(),
"ES": holidays.ES(),
"CZ": holidays.CZ(),
"SK": holidays.SK(),
"PL": holidays.PL(),
"PT": holidays.PT(),
"NL": holidays.NL(),
"NO": holidays.NO(),
"IT": holidays.IT(),
"SE": holidays.SE(),
"JP": holidays.JP(),
"BE": holidays.BE(),
"ZA": holidays.ZA(),
"SI": holidays.SI(),
"FI": holidays.FI(),
"CH": holidays.CH()
}
def get_holiday():
def setUp(self):
self.holidays = holidays.PL()
def pnl_analysis(trade_hist,
symbol_tik,
start="2020-04-24",
end="2020-07-02",
show_chart=False,
benchmark=False):
pl_holidays = holidays.PL()
st = pd.to_datetime(start).date()
ed = pd.to_datetime(end).date()
lst_bd = []
try:
file_name = 'pf_pnl.csv'
pkl_check = pd.read_csv(file_name)
pkl_check['date'] = pd.to_datetime(pkl_check['date'])
pkl_check['date'] = pkl_check.date.dt.date
pkl_last_date = pkl_check.date.max()
if pkl_last_date >= ed:
start_date_mask = pkl_check.date >= st
end_date_mask = pkl_check.date <= ed
df_pnl = pkl_check[start_date_mask & end_date_mask]
err_param = False
else:
err_param = True
except FileNotFoundError:
err_param = True
print(f'No such file, reloading script')
if err_param:
for i in range(int((ed - st).days) + 1):
bd = st + datetime.timedelta(i)
if bd in pl_holidays or bd.weekday() > 4:
continue
lst_bd.append(bd)
date_lst = []
pnl_lst = []
val_open_lst = []
val_now_lst = []
for x in lst_bd:
pf_data = portfolio_preparation(
data_preparation(trade_hist, symbol_tik, str(x)), symbol_tik,
str(x))
pnl_l = pf_data.pnl_live.sum()
pnl_c = pf_data.pnl_closed.sum()
val_open = pf_data.value_at_open.sum()
val_now = pf_data.value_now.sum()
# print(f'DATE:{x} | PNL TOTAL: {pnl_l + pnl_c:.2f}')
date_lst.append(pd.to_datetime(x).date())
pnl_lst.append(pnl_l + pnl_c)
val_open_lst.append(val_open)
val_now_lst.append(val_now)
df_pnl = pd.DataFrame(
list(zip(date_lst, pnl_lst, val_open_lst, val_now_lst)),
columns=['date', 'pnl_total', 'val_open_lst', 'val_now_lst'])
df_pnl.to_csv('pf_pnl.csv', index=False)
# condition below will modify df_pnl and return dataframe which compare daily return on portfolio and wig20
if benchmark:
df_pnl = df_pnl.copy()
df_pnl['%_change_cumulative'] = ((
(df_pnl.val_open_lst + df_pnl.pnl_total) / df_pnl.val_open_lst) -
1) * 100
df_pnl['%_1_day_shift'] = abs(
df_pnl['%_change_cumulative'].shift(1).fillna(0))
df_pnl['%_daily_change'] = df_pnl['%_change_cumulative'] - df_pnl[
'%_1_day_shift']
benchmark_symbol = 'WIG20'
benchmark_data = pd.read_csv(f'./mkt_data/{benchmark_symbol}.csv')
benchmark_data = benchmark_data[['Date', 'Close']]
# benchmark_data['Date'] = pd.to_datetime(benchmark_data['Date'])
benchmark_data['Date'] = pd.to_datetime(benchmark_data['Date'])
benchmark_data['Date'] = benchmark_data.Date.dt.date
last_date = pd.to_datetime(benchmark_data.iloc[-1, 0])
if ed > last_date:
hd.data_download(symbol_tik, end=str(ed))
benchmark_data = benchmark_data.loc[benchmark_data.Date >= (
st - datetime.timedelta(1))]
benchmark_data['1d_shift'] = benchmark_data['Close'].shift(1).fillna(0)
benchmark_data['%1d_change'] = (
(benchmark_data['Close'] / benchmark_data['1d_shift']) - 1) * 100
benchmark_data = benchmark_data.loc[(benchmark_data.Date >= st)
& (benchmark_data.Date <= ed)]
benchmark_data['%_change_cumulative'] = benchmark_data[
'%1d_change'].cumsum()
return [df_pnl, benchmark_data]
if show_chart:
plt.plot(df_pnl.date, df_pnl.pnl_total, color='#8f9805')
plt.show()
return df_pnl
def wrapper():
date = f()
pl_holidays = holidays.PL()
return pl_holidays[
date] if date in pl_holidays else f"Brak święta w dniu {date}"