def stat_feat_15(security_symbol, start_date, end_date, thrd=-0.001, interval=None, conn=None):
conn = connect_to_db() if conn is None else conn
df = download_data_daily(conn, security_symbol, start_date, end_date)
features = feature_extraction_fallraise(df)
pos_cnt = 0
neg_cnt = 0
pos = 0
neg = 0
expp_list = list()
for index, row in features.iterrows():
if row['lr02'] < row['lr01'] < thrd:
if row['lr0'] > 0.000:
pos_cnt += 1
pos += row['lr0']
elif row['lr0'] < -0.000:
neg_cnt += 1
neg += row['lr0']
if interval is not None and pos_cnt + neg_cnt >= interval:
exp_p = (pos + neg) / (pos_cnt + neg_cnt)
expp_list.append(round(100*exp_p, 2))
pos_cnt = 0
neg_cnt = 0
pos = 0
neg = 0
if interval is None:
exp_p = (pos + neg)/(pos_cnt + neg_cnt)
print(f"corr: {pos_cnt} with expected return {round(100*pos/pos_cnt, 2)}%, wrong: {neg_cnt} with expected return "
f"{round(100*neg/neg_cnt, 2)}%, Strategy Total Expected Return {round(100*exp_p, 2)}%")
else:
print(expp_list)
def candle_stick(security_symbol, date_key_int, conn=None, df=None):
"""
plot candle stick chart
:param conn: db engine obj
:param security_symbol: string
:param date_key_int: int
:return:
"""
if df is None:
conn = connect_to_db() if conn is None else conn
df = download_data_intraday(conn, security_symbol, date_key_int, "1m")
df = estimate_vwap_intraday(df)
print(f"plotting chart for {security_symbol} on {date_key_int}...")
fig = go.Figure(data=[
go.Candlestick(x=df['time_stamp'],
open=df['open_price'],
high=df['high_price'],
low=df['low_price'],
close=df['close_price'],
name="Candle Stick"),
go.Line(x=df['time_stamp'], y=df['vwap'], mode="lines", name="VWAP")
])
fig.update_layout(title=f'Daily Price Visualization For {date_key_int}',
yaxis_title=f'{security_symbol} Stock')
fig.show()
def dist_feat_15(security_symbol, start_date, end_date, thrd=-0.001, conn=None):
conn = connect_to_db() if conn is None else conn
df = download_data_daily(conn, security_symbol, start_date, end_date)
features = feature_extraction_fallraise(df)
features_ss = features.loc[(features['lr01'] < thrd) & (features['lr02'] < features['lr01'])]
print(f"Mean: {features_ss['lr0'].mean()}, Sigma: {features_ss['lr0'].std()}")
# print(features_ss)
features_ss.hist(column='lr0', bins=500)
plt.show()
def stat_feat_14(security_symbol,
start_date,
end_date,
thrd=-0.001,
interval=None,
conn=None):
conn = connect_to_db() if conn is None else conn
df = download_data_daily(conn, security_symbol, start_date, end_date)
features = feature_extraction_fallraise(df)
pos_cnt = 0
neg_cnt = 0
pos = 0
neg = 0
expp_list = list()
triggers = list()
for index, row in features.iterrows():
if row['lr01'] < row['lr0'] < thrd and row['close_price'] < row[
'vwap_ma_3']:
if row['lr5-5'] > 0.000:
pos_cnt += 1
pos += row['lr5-5']
ann = {
'x': row['date_key_int'],
'y': row['high_price'],
'text': round(row['lr5-5'], 2)
}
triggers.append(ann)
elif row['lr5-5'] < -0.000:
neg_cnt += 1
neg += row['lr5-5']
ann = {
'x': row['date_key_int'],
'y': row['high_price'],
'text': round(row['lr5-5'], 2)
}
triggers.append(ann)
if interval is not None and pos_cnt + neg_cnt >= interval:
exp_p = (pos + neg) / (pos_cnt + neg_cnt)
expp_list.append(round(100 * exp_p, 2))
pos_cnt = 0
neg_cnt = 0
pos = 0
neg = 0
if interval is None and (pos_cnt + neg_cnt) > 0:
exp_p = (pos + neg) / (pos_cnt + neg_cnt)
pos_rate = round(100 * pos / pos_cnt, 2) if pos_cnt > 0 else "N/A"
neg_rate = round(100 * neg / neg_cnt, 2) if neg_cnt > 0 else "N/A"
print(
f"corr: {pos_cnt} with expected return {pos_rate}%, wrong: {neg_cnt} with expected return "
f"{neg_rate}%, Strategy Total Expected Return {round(100 * exp_p, 2)}%"
)
return exp_p, triggers
else:
print(expp_list)
print(features.iloc[-1]['date_key_int'])
return expp_list, triggers
def candle_stick_daily(security_symbol,
start_date,
end_date,
conn=None,
df=None,
annotations=None):
"""
:param security_symbol:
:param start_date:
:param end_date:
:param conn:
:param df:
:param annotations: [{'x': date_key_int, 'y': high_price, 'text': 'falling?'}]
:return:
"""
def __update_annotations(fig, anns):
if anns is not None:
for ann in anns:
fig.add_annotation(x=pd.to_datetime(str(ann['x']),
format='%Y%m%d'),
y=ann['y'],
text=ann['text'])
fig.update_annotations(
dict(xref="x",
yref="y",
showarrow=True,
arrowhead=7,
ax=0,
ay=-40))
return True
if df is None:
conn = connect_to_db() if conn is None else conn
df = download_data_daily(conn, security_symbol, start_date, end_date)
df['date_key'] = df['date_key_int'].apply(
lambda x: pd.to_datetime(str(x), format='%Y%m%d'))
print(f"plotting chart for {security_symbol} ...")
fig = go.Figure(data=[
go.Candlestick(x=df['date_key'],
open=df['open_price'],
high=df['high_price'],
low=df['low_price'],
close=df['close_price'],
name="Candle Stick")
])
__update_annotations(fig, annotations)
fig.update_layout(title=f'Daily Price Visualization',
yaxis_title=f'{security_symbol} Stock')
fig.show()
def show_all_fallraise(security_symbol,
start_date,
end_date,
f_thd=-0.01,
r_thd=0.02,
conn=None):
conn = connect_to_db() if conn is None else conn
df = download_data_daily(conn, security_symbol, start_date, end_date)
df = log_return(df)
signals = get_falls_raises(df, falls_thd=f_thd, raises_thd=r_thd)
signals['raises'].extend(signals['falls'])
print(len(signals['raises']), len(signals['falls']))
candle_stick_daily(security_symbol,
start_date,
end_date,
conn,
annotations=signals['raises'])
def rank_performances(file_name, start_date, end_date):
"""
:param file_name:
:param start_date:
:param end_date:
:return:
"""
# take second element for sort
def __take_second(elem):
return elem[1]
secs = read_sec_list(file_name)
conn = connect_to_db()
results = list()
for sec in secs:
df = download_data_daily(conn, sec, start_date, end_date)
perf = get_performance_log_return(df)
results.append([sec, perf['mean'], perf['std']])
results.sort(key=__take_second)
for each in results:
print(each)
email_notice.content_type = 'ETL Warning!'
email_notice.content = 'Error(s) occurred in ETL. Please see the log file for details.'
else:
logfile = None
msg = email_notice.email_msg(logfile=logfile)
email_notice.send_email(msg)
if __name__ == '__main__':
if os.path.exists('error.log'): os.remove('error.log')
# Read stock symbols in
symbol_file = 'stock_symbol.csv'
symbol_list = pd.read_csv(symbol_file)['symbol'].to_list()
# Create postgresql connection and cursor
conn = db_util.connect_to_db()
if conn is not None:
table_list = [{
'table_name': 'us_stock_daily',
'res': 'D'
}, {
'table_name': 'us_stock_1m',
'res': '1'
}]
for table in table_list:
now = int(datetime.datetime.now().timestamp())
init_time = time.time()
for i, symbol in enumerate(symbol_list):
# prevent requests from exceeding limit (60 queries per minute)
if not (i + 1) % 60:
sleeptime = init_time + 61 - time.time(
from utils.db_util import connect_to_db
from utils.helpers import read_sec_list
from utils.visualization import candle_stick
from analytics.oscillation_20 import find_oscs
if __name__ == '__main__':
conn = connect_to_db()
securities = read_sec_list("sec_list.csv")
negs = 0
poss = 0
osc_lst = dict()
for security in securities:
oscs, n, p = find_oscs(conn, security, 20200604, 20200605)
if len(oscs) != 0:
osc_lst[security] = oscs
negs += n
poss += p
conn = None
print(f"positive: {poss}; negative: {negs}")
print("ready to plot, press anykey to continue:")
for security in osc_lst:
for date_key_int in osc_lst[security]:
input()
candle_stick(security,
date_key_int,
df=osc_lst[security][date_key_int])
