def __init__(self,
df_metrics,
df_baseindex=None,
cd_strategy='naked_put',
id_baseindex=c.Util.STR_INDEX_300SH,
cd_marutity_days=3):
self.start_date = df_metrics[c.Util.DT_DATE].values[0]
self.end_date = df_metrics[c.Util.DT_DATE].values[-1]
if df_baseindex is None:
df_baseindex = get_data.get_index_mktdata(self.start_date,
self.end_date,
c.Util.STR_INDEX_300SH)
self.min_holding = 20
self.init_fund = c.Util.BILLION
self.slippage = 0
self.m = 1 # 期权notional倍数
if cd_strategy == 'short_straddle':
self.m = 0.5
self.moneyness_rank = -4
self.nbr_maturity = 0
self.cd_trade_price = c.CdTradePrice.VOLUME_WEIGHTED
self.account = BaseAccount(init_fund=c.Util.BILLION,
leverage=1.0,
rf=0.03)
self.optionset = BaseOptionSet(df_metrics)
self.optionset.init()
self.index = BaseInstrument(df_baseindex)
self.index.init()
self.cd_strategy = cd_strategy
self.cd_maturity_days = cd_marutity_days
self.init_index = self.index.mktprice_close()
def __init__(self, start_date, end_date):
self.ttm = 30
self.buywrite = BuyWrite.BUY
self.fund = Util.BILLION
self.invest_underlying_ratio = 0.7
# self.invest_underlying_ratio = 1
self.slippage = 2
self.start_date = start_date
self.end_date = end_date
hist_date = start_date - datetime.timedelta(days=40)
df_future_c1 = get_dzqh_cf_c1_minute(start_date, end_date, 'if')
df_future_c1_daily = get_dzqh_cf_c1_daily(hist_date, end_date, 'if')
df_futures_all_daily = get_dzqh_cf_daily(
start_date, end_date, 'if') # daily data of all future contracts
df_index = get_index_mktdata(
start_date, end_date,
'index_300sh') # daily data of underlying index
df_index = df_index[df_index[Util.DT_DATE].isin(
Util.DZQH_CF_DATA_MISSING_DATES) == False].reset_index(drop=True)
# df_index.to_csv('df_index.csv')
self.trade_dates = sorted(df_future_c1_daily[Util.DT_DATE].unique())
self.df_vol_1m = Histvol.hist_vol(df_future_c1_daily)
# df_parkinson_1m = Histvol.parkinson_number(df_future_c1_daily)
self.df_garman_klass = Histvol.garman_klass(df_future_c1_daily)
# df_hist_vol = self.df_vol_1m.join(self.df_garman_klass, how='left')
# df_hist_vol.to_csv('../../data/df_hist_vol.csv')
self.underlying = BaseInstrument(df_data=df_index)
self.underlying.init()
self.synthetic_option = SytheticOption(
df_c1_data=df_future_c1,
# df_c1_daily=df_future_c1_daily,
df_futures_all_daily=df_futures_all_daily,
df_index_daily=df_index)
self.synthetic_option.init()
self.account = BaseAccount(self.fund, leverage=20.0, rf=0.0)
self.trading_desk = Trade()
self.init_spot = self.synthetic_option.underlying_state_daily[
Util.AMT_CLOSE]
self.df_analysis = pd.DataFrame()
# end_date = datetime.date(2016, 12, 31)
# start_date = datetime.date(2017, 1, 1)
# end_date = datetime.date(2017, 12, 31)
start_date = datetime.date(2016, 2, 1)
end_date = datetime.date(2017, 1, 31)
# start_date = datetime.date(2015, 3, 1)
# end_date = datetime.date(2018, 8, 20)
close_out_date = end_date
min_holding = 15
nbr_maturity = 1
# nbr_maturity = 0
slippage = 0
pct_underlying_invest = 0.7
res = {}
df_metrics = get_data.get_50option_mktdata(start_date, end_date)
df_underlying = get_data.get_index_mktdata(start_date, end_date,
c.Util.STR_INDEX_50ETF)
# df_underlying.to_csv('../accounts_data/df_underlying.csv')
def get_option_unit(option_put: BaseOption, underlying_value: float,
target_delta):
if target_delta is None:
unit = np.floor(underlying_value / option_put.strike() /
option_put.multiplier()) # 期权名义本金等于标的市值
else:
delta = option_put.get_delta(option_put.get_implied_vol())
unit = np.floor(target_delta * underlying_value / delta /
option_put.multiplier()) # 根据目标delta值
return unit
end_date = datetime.date(2018, 8, 8) dt_histvol = start_date - datetime.timedelta(days=40) min_holding = 18 """ commodity option """ # name_code = name_code_option = c.Util.STR_M # df_metrics = get_data.get_comoption_mktdata(start_date, end_date,name_code) # df_future_c1_daily = get_data.get_future_c1_by_option_daily(dt_histvol, end_date, name_code, min_holding) """ 50ETF option """ name_code = c.Util.STR_IH name_code_option = c.Util.STR_50ETF df_metrics = get_data.get_50option_mktdata(start_date, end_date) df_future_c1_daily = get_data.get_dzqh_cf_c1_daily(dt_histvol, end_date, name_code) name_code_index = c.Util.STR_INDEX_50SH df_index = get_data.get_index_mktdata(dt_histvol,end_date,name_code_index) """ 历史波动率 """ # df_vol_1m = Histvol.hist_vol(df_index) # df_parkinson_1m = Histvol.parkinson_number(df_future_c1_daily) # df_garman_klass = Histvol.garman_klass(df_future_c1_daily) # df_data = df_future_c1_daily.join(df_vol_1m,on=c.Util.DT_DATE,how='left') # df_data = df_data.join(df_garman_klass,on=c.Util.DT_DATE,how='left') # df_data = df_data.join(df_parkinson_1m,on=c.Util.DT_DATE,how='left') # df_data = df_data.dropna() # df_data = df_future_c1_daily """ iv premium """ # df_iv = get_data.get_iv_by_moneyness(start_date,end_date,name_code_option) # # df_iv['selected_date'] = df_iv[c.Util.DT_MATURITY].apply(lambda x:x-datetime.timedelta(days=30)) # df_iv = df_iv[df_iv['selected_date']==df_iv[c.Util.DT_DATE]]
res = am.fit(disp='off')
print(res.summary())
forecasts = res.forecast(horizon=5)
print(forecasts.variance.iloc[-3:]**0.5 * math.sqrt(252) / 100)
print(self.df_data.iloc[-3:])
def recursive_forecast_generation(self):
am = arch_model(self.returns, vol='Garch', p=1, q=1, dist='t')
end_loc = 100 # 基于一百个数据预测
max_loc = len(self.returns.index) - 1
forecasts = {}
for i in range(max_loc - end_loc):
# sys.stdout.write('.')
sys.stdout.flush()
res = am.fit(last_obs=i + end_loc, disp='off')
temp = res.forecast(horizon=3).variance**0.5 * math.sqrt(252) / 100
fcast = temp.iloc[i + end_loc - 1]
fcast['hv_20'] = self.df_data.iloc[i + end_loc]['hv_20']
fcast['iv'] = self.df_data.iloc[i + end_loc][Util.PCT_IMPLIED_VOL]
forecasts[fcast.name] = fcast
print()
print(pd.DataFrame(forecasts).T)
df_c1 = get_index_mktdata(datetime.date(2018, 1,
1), datetime.date(2018, 12, 31),
Util.STR_INDEX_300SH)[[Util.DT_DATE, Util.AMT_CLOSE]]
arch = Arch(df_c1)
arch.recursive_forecast_generation()
"""按当日价格调整保证金，计算投资组合盯市价值"""
bkt_account.mkm_update_portfolio(evalDate,self.portfolio)
# print(evalDate, bkt_optionset.eval_date, ' , ', bkt_account.npv, bkt_account.mtm_long_positions)
"""Back Test Settings"""
# start_date = datetime.date(2015, 3, 1)
start_date = datetime.date(2017, 10, 1)
end_date = datetime.date(2018, 1, 21)
"""Collect Mkt Date"""
df_option_metrics = get_50option_mktdata(start_date, end_date)
df_index_metrics = get_index_mktdata(start_date,end_date,'index_50etf')
"""Run Backtest"""
bkt_strategy = BktStrategyCollar(df_option_metrics, df_index_metrics)
bkt_strategy.set_min_holding_days(8)
df_index_ma = bkt_strategy.get_moving_average_signal(get_index_ma(start_date,end_date,'index_50etf'))
df_vix_boll = bkt_strategy.get_bollinger_signal(get_index_ma(start_date,end_date,'index_cvix'),cd_long='60')
df_index_boll = bkt_strategy.get_bollinger_signal(get_index_ma(start_date,end_date,'index_50etf'),cd_long='60')
df_iv = bkt_strategy.get_bollinger_signal_calculate(get_pciv_ratio(end_date),start_date)
# df_iv = bkt_strategy.get_percentile_signal(get_put_call_iv(end_date),start_date)
bkt_strategy.set_index_ma(df_index_ma)
bkt_strategy.set_index_boll(df_index_boll)
from back_test.model.base_account import BaseAccount
from back_test.model.base_instrument import BaseInstrument
from back_test.model.constant import Util, OptionType, LongShort, ExecuteType
from back_test.model.trade import Trade
from data_access.get_data import get_dzqh_cf_daily, get_dzqh_cf_c1_daily, \
get_dzqh_cf_c1_minute, get_index_mktdata
from OptionStrategyLib.VolatilityModel.historical_volatility import historical_volatility_model as Histvol
start_date = datetime.date(2018, 4, 1)
end_date = datetime.date(2018, 6, 1)
hist_date = start_date - datetime.timedelta(days=40)
df_future_c1 = get_dzqh_cf_c1_minute(start_date, end_date, 'if')
df_future_c1_daily = get_dzqh_cf_c1_daily(hist_date, end_date, 'if')
df_futures_all_daily = get_dzqh_cf_daily(
start_date, end_date, 'if') # daily data of all future contracts
df_index = get_index_mktdata(start_date, end_date,
'index_300sh') # daily data of underlying index
df_index = df_index[df_index[Util.DT_DATE].isin(
Util.DZQH_CF_DATA_MISSING_DATES) == False].reset_index(drop=True)
df_vol_1m = Histvol.hist_vol(df_future_c1_daily)
# df_parkinson_1m = Histvol.parkinson_number(df_future_c1_daily)
# df_garman_klass = Histvol.garman_klass(df_future_c1_daily)
# df_hist_vol = df_vol_1m.join(df_parkinson_1m, how='left')
# df_hist_vol = df_hist_vol.join(df_garman_klass, how='left')
df_future_c1_daily = df_future_c1_daily[
df_future_c1_daily[Util.DT_DATE] >= start_date].reset_index(drop=True)
synthetic_option = SytheticOption(df_c1_minute=df_future_c1,
df_c1_daily=df_future_c1_daily,
df_futures_all_daily=df_futures_all_daily,
df_index_daily=df_index)
synthetic_option.init()
import numpy as np
from OptionStrategyLib.VolatilityModel.kernel_density import kde_sklearn
from Utilities.PlotUtil import PlotUtil
from scipy import stats
pu = PlotUtil()
start_date = datetime.date(2010, 1, 1)
end_date = datetime.date.today()
dt_histvol = start_date
""" 50ETF option """
name_code = c.Util.STR_CU
name_code_index = c.Util.STR_INDEX_50SH
# df_future_c1_daily = get_data.get_dzqh_cf_c1_daily(dt_histvol, end_date, name_code)
df_future_c1_daily = get_data.get_mktdata_future_c1_daily(
dt_histvol, end_date, name_code)
df_index = get_data.get_index_mktdata(start_date, end_date, name_code_index)
df_future_c1_daily[c.Util.AMT_YIELD] = np.log(
df_future_c1_daily[c.Util.AMT_CLOSE]).diff(periods=20)
df_index[c.Util.AMT_YIELD] = np.log(
df_index[c.Util.AMT_CLOSE]).diff(periods=20)
df_future_c1_daily = df_future_c1_daily.dropna()
df_index = df_index.dropna()
r_future = np.array(df_future_c1_daily[c.Util.AMT_YIELD])
r_index = np.array(df_index[c.Util.AMT_YIELD])
s = np.random.normal(0, 0.15, 1000)
m1 = np.mean(r_future)
sk1 = stats.skew(r_future)
std1 = np.std(r_future)
from OptionStrategyLib.OptionStrategy.protective_put.HedgeIndexByOptions import HedgeIndexByOptions
import back_test.model.constant as c
import datetime
import pandas as pd
from data_access import get_data
from Utilities.PlotUtil import PlotUtil
import matplotlib.pyplot as plt
start_date = datetime.date(2015, 4, 16)
end_date = datetime.date(2018, 11, 1)
dt_histvol = start_date - datetime.timedelta(days=500)
df_metrics = get_data.get_50option_mktdata(start_date, end_date)
df_index = get_data.get_index_mktdata(dt_histvol, end_date,
c.Util.STR_INDEX_50ETF)
df_1 = get_data.get_index_mktdata(dt_histvol, end_date, c.Util.STR_INDEX_50SH)
df_c1 = get_data.get_mktdata_future_c1_daily(dt_histvol, end_date,
c.Util.STR_IH)
df_all = get_data.get_mktdata_future_daily(dt_histvol, end_date, c.Util.STR_IH)
df_1.to_csv('50sh.csv')
df_c1.to_csv('ih.csv')
cd_direction_timing = 'ma'
cd_strategy = 'bull_spread'
cd_volatility = 'close_std'
cd_short_ma = 'ma_3'
cd_long_ma = 'ma_20'
cd_std = 'std_10'
df_res = pd.DataFrame()
hedging1 = HedgeIndexByOptions(df_index,
start_date = datetime.date(2015, 2, 1)
# start_date = datetime.date(2017, 2, 1)
end_date = datetime.date(2018, 8, 31)
d1 = start_date
min_holding = 15
nbr_maturity = 1
slippage = 0
pct_underlying_invest = 1.0
##############
alpha = 0.0
moneyness = -5
#################
df_metrics = get_data.get_50option_mktdata(start_date, end_date)
df_option_underlying = get_data.get_index_mktdata(start_date, end_date,
c.Util.STR_INDEX_50ETF)
df_underlying = get_data.get_index_mktdata(start_date, end_date,
c.Util.STR_INDEX_300SH_TOTAL_RETURN)
calendar = c.Calendar(sorted(df_underlying[c.Util.DT_DATE].unique()))
pu = PlotUtil()
df = pd.DataFrame()
#
d1 = calendar.firstBusinessDayNextMonth(d1)
d2 = d1 + datetime.timedelta(days=365)
while d2 <= end_date:
print(d1)
df_metrics_1 = df_metrics[(df_metrics[c.Util.DT_DATE] >= d1) & (
df_metrics[c.Util.DT_DATE] <= d2)].reset_index(drop=True)
df_underlying_1 = df_underlying[(df_underlying[c.Util.DT_DATE] >= d1) & (
# df_res = pd.DataFrame()
# for cd_strategy in ['short_call','short_put','short_straddle']:
# naked_short = NakedShort(df_metrics,df_baseindex,cd_strategy=cd_strategy,cd_marutity_days=5)
# naked_short.moneyness_rank=-2
# naked_short.back_test()
# naked_short.account.account.to_csv('../../accounts_data/naked_short_account_'+ str(cd_strategy) +'_' + str(naked_short.moneyness_rank) + '_2015.csv')
# naked_short.account.trade_records.to_csv('../../accounts_data/naked_short_records_' + str(cd_strategy) +'_' + str(naked_short.moneyness_rank) + '_2015.csv')
# res = naked_short.account.analysis()
# print(res)
# df_res[cd_strategy] = res
# df_res.to_csv('../../accounts_data/res_cd_strategy_' + str(naked_short.moneyness_rank) + '_2015.csv')
df_metrics = get_data.get_50option_mktdata(datetime.date(2015, 2, 9),
datetime.date(2018, 11, 20))
df_baseindex = get_data.get_index_mktdata(datetime.date(2015, 2, 9),
datetime.date(2018, 11, 20),
c.Util.STR_INDEX_50SH)
df_res = pd.DataFrame()
for cd_strategy in ['short_call', 'short_put', 'short_straddle']:
naked_short = NakedShort(df_metrics,
df_baseindex,
cd_strategy=cd_strategy,
cd_marutity_days=0)
naked_short.moneyness_rank = -4
naked_short.back_test()
naked_short.account.account.to_csv(
'../../accounts_data/naked_short_account_' + str(cd_strategy) + '_' +
str(naked_short.moneyness_rank) + '_2015.csv')
naked_short.account.trade_records.to_csv(
'../../accounts_data/naked_short_records_' + str(cd_strategy) + '_' +
df_res.to_excel(writer, name_code)
dt_end = df_metrics[c.Util.DT_DATE].unique()[-1]
dt_yesterday = df_metrics[c.Util.DT_DATE].unique()[-2]
print('Finished ', name_code, dt_end, dt_yesterday)
# df_holdings = trade_volume(dt_end, dt_yesterday, df_metrics, name_code, core_id,df_res)
# df_holdings.to_excel(writer, 'holdings_'+name_code)
name_codes = [
c.Util.STR_CF, c.Util.STR_C, c.Util.STR_RU, c.Util.STR_M, c.Util.STR_CU,
c.Util.STR_SR
]
dt_start = datetime.date(2000, 1, 1)
for (idx1, name_code) in enumerate(name_codes):
print(name_code)
df_res = pd.DataFrame()
df_future_c1_daily = get_data.get_gc_future_c1_daily(
dt_start, end_date, name_code)
df_future_c1_daily = hist_vol1(
df_future_c1_daily[[c.Util.DT_DATE, c.Util.AMT_CLOSE]])
df_future_c1_daily = df_future_c1_daily.sort_values(by=c.Util.DT_DATE,
ascending=False)
df_future_c1_daily.to_excel(writer, name_code + '_hv')
df_50etf = get_data.get_index_mktdata(dt_start, end_date, 'index_50etf')
df_50etf = hist_vol1(df_50etf[[c.Util.DT_DATE, c.Util.AMT_CLOSE]])
df_50etf = df_50etf.sort_values(by=c.Util.DT_DATE, ascending=False)
df_50etf.to_excel(writer, 'index_50etf' + '_hv')
writer.save()
