def df_hv(code, date):
try:
now = datetime.now()
start_day = now - timedelta(days=480)
start_day = start_day.strftime('%Y-%m-%d')
if code[:2] == 'IO':
df = get_inx('000300', start_day)
else:
fn = get_dss() + 'fut/bar/day_' + code + '.csv'
df = pd.read_csv(fn)
df = df[(df.date >= start_day) & (df.date <= date)]
df = df.set_index('date')
df = df.sort_index()
df['ln'] = np.log(df.close)
df['rt'] = df['ln'].diff(1)
df['hv'] = df['rt'].rolling(20).std()
df['hv'] *= np.sqrt(242) * 100
df = df.iloc[-242:, ]
df['value'] = df['hv']
df = df[['value']]
df.index.name = code + '_hv'
return df
except Exception as e:
s = traceback.format_exc()
to_log(s)
return None
def get_trade_preday(date):
from nature import get_inx
df = get_inx('000300')
date_list = list(df.date)
try:
i = date_list.index(date)
except:
i = -1
return date_list[i + 1]
def calc_hv(date):
df = get_inx('000300', '2020-01-01', date)
df = df.set_index('date')
df = df.sort_index()
df['ln'] = np.log(df.close)
df['rt'] = df['ln'].diff(1)
df['hv'] = df['rt'].rolling(60).std()
df['hv'] *= np.sqrt(242)
cur = df.iloc[-1, :]
return float(cur.hv), float(cur.close)
def hv_show(code):
# hv()
now = datetime.now()
start_day = now - timedelta(days=480)
start_day = start_day.strftime('%Y-%m-%d')
if code == '000300':
df = get_inx('000300', start_day)
else:
fn = get_dss() + 'fut/bar/day_' + code + '.csv'
df = pd.read_csv(fn)
df = df[df.date >= start_day]
df = df.set_index('date')
df = df.sort_index()
df['ln'] = np.log(df.close)
df['rt'] = df['ln'].diff(1)
df['hv'] = df['rt'].rolling(20).std()
df['hv'] *= np.sqrt(242)
df = df.iloc[-242:, :]
cur = df.iloc[-1, :]
hv = round(cur['hv'], 2)
hv_rank = round((hv - df['hv'].min()) / (df['hv'].max() - df['hv'].min()),
2)
hv_percentile = round(len(df[df.hv < hv]) / 242, 2)
plt.figure(figsize=(13, 7))
plt.title(code + ' hv: ' + str(hv) + ' hv Rank: ' +
str(hv_rank) + ' hv Percentile: ' + str(hv_percentile))
plt.xticks(rotation=90)
plt.plot(df['hv'], '-*')
plt.grid(True, axis='y')
ax = plt.gca()
for label in ax.get_xticklabels():
label.set_visible(False)
for label in ax.get_xticklabels()[1::30]:
label.set_visible(True)
for label in ax.get_xticklabels()[-1:]:
label.set_visible(True)
fn = 'static/hv_show.jpg'
plt.savefig(fn)
plt.cla()
r = ''
fn = 'hv_show.jpg'
now = str(int(time.time()))
r = '<img src=\"static/' + fn + '?rand=' + now + '\" />'
return r
def hs300_spread_show(start_day):
df_300 = get_inx('000300', start_day)
df_300 = df_300.set_index('date')
df_300 = df_300.sort_index()
# print(df_300.tail())
fn = get_dss() + 'fut/cfg/opt_mature.csv'
df_opt = pd.read_csv(fn)
df = df_opt[(df_opt.pz == 'IO') &
(df_opt.flag == df_opt.flag)] # 筛选出不为空的记录
df = df.sort_values('symbol')
symbol_list = list(df.symbol)
# print(symbol_list)
plt.figure(figsize=(12, 8))
plt.title('hs300_spread')
for symbol in symbol_list:
code = 'IF' + symbol[2:]
fn = get_dss() + 'fut/bar/day_' + code + '.csv'
if os.path.exists(fn) == False:
continue
df = pd.read_csv(fn)
df = df[df.date >= start_day]
df = df.set_index('date')
df = df.sort_index()
df['value'] = df.close - df_300.close
# print(df.tail())
plt.plot(df.value, label=code)
plt.xticks(rotation=45)
plt.grid(True, axis='y')
ax = plt.gca()
# for label in ax.get_xticklabels():
# label.set_visible(False)
# for label in ax.get_xticklabels()[1::25]:
# label.set_visible(True)
# for label in ax.get_xticklabels()[-1:]:
# label.set_visible(True)
plt.legend()
fn = 'static/hs300_spread_show.jpg'
plt.savefig(fn)
plt.cla()
r = ''
fn = 'hs300_spread_show.jpg'
now = str(int(time.time()))
r = '<img src=\"static/' + fn + '?rand=' + now + '\" />'
return r
def hv():
# df = get_inx('000300', '2019-06-01', '2020-07-15')
df = get_inx('000300', '2019-06-01')
# df = df.sort_values('date')
df = df.set_index('date')
df = df.sort_index()
df['ln'] = np.log(df.close)
df['rt'] = df['ln'].diff(1)
df['hv'] = df['rt'].rolling(20).std()
df['hv'] *= np.sqrt(242)
df = df.iloc[-242:, :]
# print(df.head())
# print(df.tail())
cur = df.iloc[-1, :]
hv = round(cur['hv'], 2)
hv_rank = round((hv - df['hv'].min()) / (df['hv'].max() - df['hv'].min()),
2)
# print('hv: ', hv)
# print('hv Rank: ', hv_rank)
# print(' 均值：',df['hv'].mean())
# print('0.2分位数:',np.percentile(df['hv'], 20))
# print('0.5分位数:',np.percentile(df['hv'], 50))
# print('0.8分位数:',np.percentile(df['hv'], 80))
hv_percentile = round(len(df[df.hv < hv]) / 242, 2)
# print('hv percentile: ', hv_percentile)
plt.figure(figsize=(13, 7))
plt.title('hv:' + str(hv) + ' hv Rank:' + str(hv_rank) +
' hv Percentile:' + str(hv_percentile))
plt.xticks(rotation=45)
plt.plot(df['hv'], '-*')
plt.grid(True, axis='y')
ax = plt.gca()
for label in ax.get_xticklabels():
label.set_visible(False)
for label in ax.get_xticklabels()[1::30]:
label.set_visible(True)
for label in ax.get_xticklabels()[-1:]:
label.set_visible(True)
# plt.show()
fn = 'static/vol_hv.jpg'
plt.savefig(fn)
plt.cla()
def calc_greeks(year):
c = input('请确认波指已维护最新值，press y to continue: ')
if c != 'y':
return
fn_sigma = get_dss() + 'backtest/IO/hs300波指.csv'
df_sigma = pd.read_csv(fn_sigma)
df_sigma = df_sigma.set_index('date')
mature_dict = {'IO2002':'2020-02-21','IO2003':'2020-03-20','IO2004':'2020-04-17','IO2005':'2020-05-15','IO2006':'2020-06-19',
'IO2007':'2020-07-17','IO2009':'2020-09-18','IO2012':'2020-12-18','IO2103':'2021-03-19'}
fn1 = get_dss() + 'backtest/IO/' + 'IO' + year + '.csv'
fn2 = get_dss() + 'backtest/IO/' + 'IO' + year + '_greeks.csv'
df1 = pd.read_csv(fn1)
df2 = pd.read_csv(fn2)
df1 = df1.loc[len(df2):,:]
# print(df1.head(3))
for i, row in df1.iterrows():
df_index = get_inx('000300', row.date, row.date)
if len(df_index) == 0:
print('缺少沪深300日线数据')
assert False
S0 = df_index.iat[0,3]
# S0 = 3912.57
call = True if row['symbol'][7] == 'C' else False
K = float( row['symbol'][-4:] )
r = 0.03
C0 = float( row['close'] )
sa = float( df_sigma.at[row.date, 'close'] ) / 100
# sa = 0.2178
# print(sa)
if row['symbol'][:6] not in mature_dict.keys():
assert False
date_mature = mature_dict[ row['symbol'][:6] ]
date_mature = datetime.datetime.strptime(date_mature, '%Y-%m-%d')
today = datetime.datetime.strptime(row.date, '%Y-%m-%d')
T = float((date_mature - today).days) / 365
if T > 0:
if call == True:
n = 1
iv = bsm_call_imp_vol(S0, K, T, r, C0)
else:
n = -1
iv = bsm_put_imp_vol(S0, K, T, r, C0)
row['hs300'] = S0
row['sigma'] = sa
row['delta9'] = delta(S0,K,r,T,sa,n)
row['gamma'] = gamma(S0,K,r,T,sa)
row['theta'] = theta(S0,K,r,T,sa,n)
row['vega'] = vega(S0,K,r,T,sa)
row['iv'] = iv
else:
row['hs300'] = S0
row['delta9'] = 0
row['gamma'] = 0
row['theta'] = 0
row['vega'] = 0
row['sigma'] = 0
df = pd.DataFrame([row])
df.to_csv(fn2, index=False, mode='a', header=None)