https://github.com/sparrowapps/systemt
2016년 7월 14일 골든 크로스 결과만 보자면
sqlite> select * from MAGC;
0|A004560|현대비앤지스틸
1|A175330|JB금융지주
2|A004960|한신공영
3|A900110|이스트아시아홀딩스
4|A085310|엔케이
5|A000155|두산우
6|A016880|웅진
7|A059090|미코
8|A200780|비씨월드제약
9|A082740|두산엔진
10|A110570|넥솔론
11|A100030|모바일리더
12|A160550|NEW
13|A027050|코리아나
14|A083450|GST
15|A089600|나스미디어
16|A041440|에버다임
17|A200230|텔콘
18|A090990|코리아03호
19|A065440|이루온
20|A079650|서산
21|A008700|아남전자
22|A024120|KB오토시스
23|A050760|에스폴리텍
24|A008970|동양철관
sqlite>
네이버님의 골든크로스 결과보다 몇몇 종목이 더 나 왔다.
#-*- coding: utf-8 -*-
import pandas as pd
from pandas import Series, DataFrame
import sqlite3
import numpy as np
from logger import get_logger
from stockcode import get_code_list
from settings import START_DATE
from cp_constant import *
import win32com.client
def checkData( cursor, table_name ):
row = cursor.execute("SELECT 1 FROM sqlite_master WHERE type='table' AND name='{}'".format(table_name)).fetchone()
if row is None: return False
row = cursor.execute("SELECT COUNT(*) FROM '{}'".format(table_name)).fetchone()
if row is None : return False
return True
def makeDataFrame( code ):
with sqlite3.connect("price.db") as con:
cursor = con.cursor()
table_name = code
if checkData( cursor, table_name ) == False: return False
df = pd.read_sql("SELECT * FROM '{}'".format(table_name), con, index_col=None)
return df
def isMAGoldCross( df, MA1 = 20, MA2 = 60 ):
df['short_ma'] = pd.rolling_mean(df['CLOSE'],MA1)
df['long_ma'] = pd.rolling_mean(df['CLOSE'],MA2)
df['signal'] = 0.0
if len(df) < MA2 * 4 : return False
df['signal'][MA1:] = np.where(df['short_ma'][MA1:] > df['long_ma'][MA1:], 1.0, 0.0)
df['positions'] = 0.0
df['positions'] = df['signal'].diff()
if df['signal'][len(df)-1] == 1 and df['positions'][len(df)-1] == 1:
return True
else:
return False
def isBBandSignal( df, period = 20):
df['Bol_upper'] = pd.rolling_mean(df['CLOSE'], window=period) + 2* pd.rolling_std(df['CLOSE'], period, min_periods=period)
df['Bol_lower'] = pd.rolling_mean(df['CLOSE'], window=period) - 2* pd.rolling_std(df['CLOSE'], period, min_periods=period)
df['signal'] = 0.0
df['signal'] = np.where(df['Bol_lower'] > df['CLOSE'], 1.0, 0.0)
df['positions'] = 0.0
df['positions'] = df['signal'].diff()
if len(df) < 2: return False
if df['signal'][len(df)-2] == 1 and df['positions'][len(df)-2] == 1:
if df['signal'][len(df)-1] == 0 and df['positions'][len(df)-1] == -1.0:
return True
else:
return False
else:
return False
def isMACDSignal( df, n1 = 12, n2= 26, c= 9):
df['MACD'] = pd.ewma(df['CLOSE'], span=n1) - pd.ewma(df['CLOSE'], span=n2)
df['MACD_Signal'] = pd.ewma(df['MACD'], span=c)
df['signal'] = 0.0
df['signal'] = np.where(df['MACD'] < df['MACD_Signal'], 1.0, 0.0)
df['positions'] = 0.0
df['positions'] = df['signal'].diff()
if df['signal'][len(df)-1] == 1 and df['positions'][len(df)-1] == 1:
return True
else:
return False
def run():
with sqlite3.connect("analyze.db") as con:
cursor = con.cursor()
code_magc = {'CODE':[],
'NAME':[]}
code_bb = {'CODE':[],
'NAME':[]}
code_macd = {'CODE':[],
'NAME':[]}
for code, name in get_code_list():
df = makeDataFrame( code )
res = isMAGoldCross( df, 20, 60 )
if res == True:
code_magc['CODE'].append(code)
code_magc['NAME'].append(name)
get_logger().debug("MA20, MA60 Golden Cross {}{}".format(code,name))
res = isBBandSignal( df, 20 )
if res == True:
code_bb['CODE'].append(code)
code_bb['NAME'].append(name)
get_logger().debug("BBnad lower after up {}{}".format(code,name))
res = isMACDSignal( df, 12, 26, 9)
if res == True:
code_macd['CODE'].append(code)
code_macd['NAME'].append(name)
get_logger().debug("MACD sig {}{}".format(code,name))
magc = DataFrame(code_magc)
bb = DataFrame(code_bb)
macd = DataFrame(code_magc)
magc.to_sql("MAGC", con, if_exists='replace', chunksize=1000)
get_logger().debug("MAGC {} saved.".format(len(magc)))
bb.to_sql("BB", con, if_exists='replace', chunksize=1000)
get_logger().debug("BB {} saved.".format(len(bb)))
macd.to_sql("MACD", con, if_exists='replace', chunksize=1000)
get_logger().debug("MACD {} saved.".format(len(macd)))
if __name__ == '__main__':
run()
