def create_trade_sign(self, stock_price: pd.DataFrame) -> pd.DataFrame:
stock_price = stock_price.sort_values("date")
stock_price.index = range(len(stock_price))
InstitutionalInvestorsBuySell = Load.FinData(
dataset="InstitutionalInvestorsBuySell",
select=self.stock_id,
date=self.start_date,
end_date=self.end_date,
)
InstitutionalInvestorsBuySell = InstitutionalInvestorsBuySell.groupby(
["date", "stock_id"], as_index=False).agg({
"buy": np.sum,
"sell": np.sum
})
InstitutionalInvestorsBuySell["diff"] = (
InstitutionalInvestorsBuySell["buy"] -
InstitutionalInvestorsBuySell["sell"])
stock_price = pd.merge(
stock_price,
InstitutionalInvestorsBuySell[["stock_id", "date", "diff"]],
on=["stock_id", "date"],
how="left",
).fillna(0)
stock_price["signal_info"] = self.detect_Abnormal_Peak(
y=stock_price["diff"].values,
lag=10,
threshold=3,
influence=0.35,
)["signals"]
stock_price["signal"] = 0
stock_price.loc[stock_price["signal_info"] == -1, "signal"] = 1
stock_price.loc[stock_price["signal_info"] == 1, "signal"] = -1
return stock_price
def load_data(self, dataset, transpose=False):
#dataset = 'Shareholding'
data = Load.FinData(dataset=dataset,
select=self.stock_id,
date=self.date)
if len(data) == 0:
return data
if transpose:
data = Load.transpose(data)
data['stock_id'] = data['stock_id'].astype(str)
data['date'] = data['date'].astype(str)
return data
def load_taiwan_stock_margin_purchase_short_sale(self):
self.TaiwanStockMarginPurchaseShortSale = Load.FinData(
dataset="TaiwanStockMarginPurchaseShortSale",
select=self.stock_id,
date=self.start_date,
end_date=self.end_date,
)
self.TaiwanStockMarginPurchaseShortSale[[
"ShortSaleTodayBalance", "MarginPurchaseTodayBalance"
]] = self.TaiwanStockMarginPurchaseShortSale[[
"ShortSaleTodayBalance", "MarginPurchaseTodayBalance"
]].astype(int)
self.TaiwanStockMarginPurchaseShortSale[
"ShortSaleMarginPurchaseTodayRatio"] = (
self.
TaiwanStockMarginPurchaseShortSale["ShortSaleTodayBalance"] /
self.TaiwanStockMarginPurchaseShortSale[
"MarginPurchaseTodayBalance"])
def init_stock_info(self):
stock_info = Load.FinData(dataset='{}StockInfo'.format(self.country))
if 'industry_category' in stock_info.columns:
_bool = [
False if x in ['ETF', '不動產投資信託證券'] else True
for x in stock_info['industry_category']
]
stock_info = stock_info[_bool]
stock_info.index = range(len(stock_info))
stock_info['country'] = self.country
colname = ['stock_id', 'stock_name', 'country']
stockinfo = stock_info[colname]
_bool = [
True if x in self.stock_id else False
for x in stockinfo['stock_id']
]
self.stockinfo = stockinfo[_bool]
def init(self, base_data):
base_data = base_data.sort_values("date")
base_data.index = range(len(base_data))
stock_id = base_data["stock_id"].unique()
start_date = base_data["date"].min()
end_date = base_data["date"].max()
InstitutionalInvestorsBuySell = Load.FinData(
dataset="InstitutionalInvestorsBuySell",
select=stock_id,
date=start_date,
end_date=end_date,
)
InstitutionalInvestorsBuySell = InstitutionalInvestorsBuySell.groupby(
["date", "stock_id"], as_index=False).agg({
"buy": np.sum,
"sell": np.sum
})
InstitutionalInvestorsBuySell["diff"] = (
InstitutionalInvestorsBuySell["buy"] -
InstitutionalInvestorsBuySell["sell"])
base_data = pd.merge(
base_data,
InstitutionalInvestorsBuySell[["stock_id", "date", "diff"]],
on=["stock_id", "date"],
how="left",
).fillna(0)
base_data["signal_info"] = self.detect_Abnormal_Peak(
y=base_data["diff"].values,
lag=10,
threshold=3,
influence=0.35,
)["signals"]
base_data["signal"] = 0
base_data.loc[base_data["signal_info"] == -1, "signal"] = 1
base_data.loc[base_data["signal_info"] == 1, "signal"] = -1
return base_data
def load_institutional_investors_buy_sell(self):
self.InstitutionalInvestorsBuySell = Load.FinData(
dataset="InstitutionalInvestorsBuySell",
select=self.stock_id,
date=self.start_date,
end_date=self.end_date,
)
self.InstitutionalInvestorsBuySell[[
"sell", "buy"
]] = (self.InstitutionalInvestorsBuySell[["sell", "buy"
]].fillna(0).astype(int))
self.InstitutionalInvestorsBuySell = (
self.InstitutionalInvestorsBuySell.groupby(["date", "stock_id"],
as_index=False).agg({
"buy":
np.sum,
"sell":
np.sum
}))
self.InstitutionalInvestorsBuySell["diff"] = (
self.InstitutionalInvestorsBuySell["buy"] -
self.InstitutionalInvestorsBuySell["sell"])
from FinMind.Data import Load
import datetime
date = str( datetime.datetime.now().date() - datetime.timedelta(30) )
date2 = str( datetime.datetime.now().date() - datetime.timedelta(200) )
date3 = str( datetime.datetime.now().date() - datetime.timedelta(400) )
#---------------------------------------------------------------
print('load TaiwanStockInfo')
TaiwanStockInfo = Load.FinData(dataset = 'TaiwanStockInfo')
print( TaiwanStockInfo[:5] )
print('load TaiwanStockPrice {} '.format(TaiwanStockInfo.loc[2,'stock_id']))
TaiwanStockPrice = Load.FinData(
dataset = 'TaiwanStockPrice',
select = TaiwanStockInfo.loc[2,'stock_id'],
date = date)
print( TaiwanStockPrice[:5] )
print('load 財報 FinancialStatements {} '.format(TaiwanStockInfo.loc[200,'stock_id']))
TaiwanStockFinancialStatements = Load.FinData(
dataset = 'FinancialStatements',
select = TaiwanStockInfo.loc[200,'stock_id'],
date = date3)
print( TaiwanStockFinancialStatements[:5] )
print('load 台股配息 TaiwanStockStockDividend {} '.format(TaiwanStockInfo.loc[200,'stock_id']))
TaiwanStockStockDividend = Load.FinData(
dataset = 'TaiwanStockStockDividend',
select = TaiwanStockInfo.loc[200,'stock_id'],
from FinMind.Data import Load
import datetime
date = str(datetime.datetime.now().date() - datetime.timedelta(30))
date2 = str(datetime.datetime.now().date() - datetime.timedelta(200))
date3 = str(datetime.datetime.now().date() - datetime.timedelta(400))
_index = 0
# ---------------------------------------------------------------
print("load TaiwanStockInfo")
TaiwanStockInfo = Load.FinData(dataset="TaiwanStockInfo")
print(TaiwanStockInfo.head())
print("load TotalMarginPurchaseShortSale")
TotalMarginPurchaseShortSale = Load.FinData(
dataset="TotalMarginPurchaseShortSale", date=date)
print(TotalMarginPurchaseShortSale[:5])
print("load TaiwanStockNews")
TaiwanStockNews = Load.FinData(dataset="TaiwanStockNews",
date=date,
select="2317")
print(TaiwanStockNews[:5])
print("load TaiwanStockPrice {} ".format(TaiwanStockInfo.loc[_index,
"stock_id"]))
TaiwanStockPrice = Load.FinData(dataset="TaiwanStockPrice",
select="2330",
date=date2)
print(TaiwanStockPrice[:5])
print("load StockDividendResult {} ".format(TaiwanStockInfo.loc[_index,
from FinMind.Data import Load
import datetime
date = str(datetime.datetime.now().date() - datetime.timedelta(30))
date2 = str(datetime.datetime.now().date() - datetime.timedelta(200))
date3 = str(datetime.datetime.now().date() - datetime.timedelta(400))
#---------------------------------------------------------------
print('load TaiwanStockInfo')
TaiwanStockInfo = Load.FinData(dataset='TaiwanStockInfo')
print(TaiwanStockInfo[:5])
_index = 477
print('load TaiwanStockPrice {} '.format(TaiwanStockInfo.loc[_index,
'stock_id']))
TaiwanStockPrice = Load.FinData(dataset='TaiwanStockPrice',
select=['2330', '2317'],
date=date2)
print(TaiwanStockPrice[:5])
print('load 財報 FinancialStatements {} '.format(
TaiwanStockInfo.loc[_index, 'stock_id']))
TaiwanStockFinancialStatements = Load.FinData(
dataset='FinancialStatements',
select=list(TaiwanStockInfo.loc[1530:1555, 'stock_id']),
date=date3)
print(TaiwanStockFinancialStatements[:5])
# transpose
data = Load.transpose(TaiwanStockFinancialStatements)
print('load 台股配息 TaiwanStockStockDividend {} '.format(
TaiwanStockInfo.loc[_index, 'stock_id']))
c2.test_split_join()
c2.test_str_long()
c2.test_align()
c2.test_replace()
c3.introduction()
'''
c8.introduction()
c8.test_write()
print(c.stock_data_folder_path)
# code = 2867
# holly.test_load_historical_data(code, 2018, 10, 2019, 4)
# holly.test_load_historical_data_gradually(2031, c.first_year, c.second_year)
# holly.test_load_historical_data_gradually(2867, 2000, 2018)
# holly.test_load_historical_data_gradually(2867, 2011, 2018)
'''
for item in c.securities_20MA:
holly.test_load_historical_data_gradually(item, c.first_year, c.second_year)
'''
from FinMind.Data import Load
TaiwanStockInfo = Load.FinData(dataset='TaiwanStockInfo')
data = Load.FinData(dataset='TaiwanStockPrice',
select=['2425'],
date='2017-10-10')
# data = Load.FinData(dataset = 'TaiwanStockStockDividend',select = ['2425'],date = '2017-10-10')
print(data)
if isinstance(date, str):
date = datetime.datetime.strptime(date, '%Y-%m-%d %H:%M:%S')
elif isinstance(date, datetime.date):
date = datetime.datetime.combine(date,
datetime.datetime.min.time())
date = date + datetime.timedelta(minutes=1)
second = date - datetime.datetime(1970, 1, 1, 0, 0, 0)
second = (second.days * 24 * 60 * 60 + second.seconds) * 1000
return second
chart = Highstock()
data = Load.FinData(dataset='TaiwanStockPrice',
select='2330',
date='2018-10-10')
print(' change type of dataframe to highcharts ')
data['date'] = date2millisecond(data['date'])
list_data = []
for i in range(len(data)):
tem = [
int(data.loc[i, 'date']),
float(data.loc[i, 'max']),
float(data.loc[i, 'min'])
]
list_data.append(tem)
chart.add_data_set(list_data, 'arearange', 'Temperatures')
from FinMind.Data import Load
import datetime
date = str(datetime.datetime.now().date() - datetime.timedelta(30))
date2 = str(datetime.datetime.now().date() - datetime.timedelta(200))
date3 = str(datetime.datetime.now().date() - datetime.timedelta(400))
#---------------------------------------------------------------
print('load TaiwanStockInfo')
TaiwanStockInfo = Load.FinData(dataset='TaiwanStockInfo')
print(TaiwanStockInfo[:5])
_index = 9010
print('load TotalMarginPurchaseShortSale')
TotalMarginPurchaseShortSale = Load.FinData(
dataset='TotalMarginPurchaseShortSale', date=date)
print(TotalMarginPurchaseShortSale[:5])
print('load TaiwanStockNews')
TaiwanStockNews = Load.FinData(dataset='TaiwanStockNews',
date=date,
select='2317')
print(TaiwanStockNews[:5])
print('load TaiwanStockPrice {} '.format(TaiwanStockInfo.loc[_index,
'stock_id']))
TaiwanStockPrice = Load.FinData(dataset='TaiwanStockPrice',
select='2330',
date=date2)
print(TaiwanStockPrice[:5])
print('load 財報 FinancialStatements {} '.format(
from FinMind.Data import Load
import numpy as np
from keras.layers.core import Dense, Dropout
from keras.layers import LSTM , GRU
from keras.models import Sequential
from sklearn.preprocessing import MinMaxScaler
import matplotlib.pyplot as plt
import math
from keras.optimizers import Adam
print('input data')
data = Load.FinData(
dataset = 'TaiwanStockPrice',
select = '2330.TW',
date = '2010-01-01')
#colname = ['date', 'open', 'high', 'low', 'close', 'volume']
#data = data[colname]
print('select close price')
stock_price = data['Close'].values.astype('float32')
stock_price = stock_price.reshape(len(stock_price), 1)
print(' 畫圖 ')
plt.plot(stock_price)
plt.show()
print(' 歸一化 ')
scaler = MinMaxScaler(feature_range=(0, 1))
stock_price = scaler.fit_transform(stock_price)
print(' 取 80% data 當作 training data, 20% data 當作 testing data 做模型驗證 ')
train_size = int(len(stock_price) * 0.8)
def init(self, base_data):
base_data = base_data.sort_values("date")
stock_id = base_data["stock_id"].unique()
start_date = base_data["date"].min()
end_date = base_data["date"].max()
TaiwanStockMarginPurchaseShortSale = Load.FinData(
dataset="TaiwanStockMarginPurchaseShortSale",
select=stock_id,
date=start_date,
end_date=end_date,
)
TaiwanStockMarginPurchaseShortSale[[
"ShortSaleTodayBalance", "MarginPurchaseTodayBalance"
]] = TaiwanStockMarginPurchaseShortSale[[
"ShortSaleTodayBalance", "MarginPurchaseTodayBalance"
]].astype(int)
TaiwanStockMarginPurchaseShortSale[
"ShortSaleMarginPurchaseTodayRatio"] = (
TaiwanStockMarginPurchaseShortSale["ShortSaleTodayBalance"] /
TaiwanStockMarginPurchaseShortSale[
"MarginPurchaseTodayBalance"])
InstitutionalInvestorsBuySell = Load.FinData(
dataset="InstitutionalInvestorsBuySell",
select=stock_id,
date=start_date,
end_date=end_date,
)
InstitutionalInvestorsBuySell[[
"sell", "buy"
]] = (InstitutionalInvestorsBuySell[["sell",
"buy"]].fillna(0).astype(int))
InstitutionalInvestorsBuySell = InstitutionalInvestorsBuySell.groupby(
["date", "stock_id"], as_index=False).agg({
"buy": np.sum,
"sell": np.sum
})
InstitutionalInvestorsBuySell["diff"] = (
InstitutionalInvestorsBuySell["buy"] -
InstitutionalInvestorsBuySell["sell"])
base_data = pd.merge(
base_data,
InstitutionalInvestorsBuySell[["stock_id", "date", "diff"]],
on=["stock_id", "date"],
how="left",
).fillna(0)
base_data = pd.merge(
base_data,
TaiwanStockMarginPurchaseShortSale[[
"stock_id", "date", "ShortSaleMarginPurchaseTodayRatio"
]],
on=["stock_id", "date"],
how="left",
).fillna(0)
base_data.index = range(len(base_data))
base_data["signal"] = 0
sell_mask = (base_data["ShortSaleMarginPurchaseTodayRatio"] >=
self.ShortSaleMarginPurchaseTodayRatioThreshold) & (
base_data["diff"] > 0)
base_data.loc[sell_mask, "signal"] = -1
buy_mask = (base_data["ShortSaleMarginPurchaseTodayRatio"] <
self.ShortSaleMarginPurchaseTodayRatioThreshold) & (
base_data["diff"] < 0)
base_data.loc[buy_mask, "signal"] = 1
return base_data
import math
import datetime
from tabulate import tabulate
from wave_info.deal_with_wave import *
from wave_logic.wave_logic_function import *
from matplotlib import pyplot as plt
from FinMind.Data import Load
import numpy as np
import json
import pandas as pd
TaiwanStockInfo = Load.FinData(dataset = 'TaiwanStockInfo')
rs=test1(8088)
print(tabulate(rs[0], headers='keys', tablefmt='psql'))
rs[0].to_csv("earn.csv")
print(tabulate(rs[1], headers='keys', tablefmt='psql'))
rs[1].to_csv("lose.csv")
def test1(stock_id,
some_days_ago=700,
rolling_num=20,
nfactor=1,
stop_loss_factor=10):
try:
#資料準備區
data = Load.FinData(
dataset='TaiwanStockPrice',
select=[str(stock_id)],
date=str(datetime.date.today() -
datetime.timedelta(days=some_days_ago))) #資料取得
critical_point_data = get_critical_point(data, rolling_num) #簡化波型(粗糙)
simple_wave = get_simple_wave(
critical_point_data) #處理簡化過的波型 把兩個下跌的波整併成一個波
wave_info = get_wave_info(simple_wave,
critical_point_data) #計算使用波浪理論需要的參數
lose_list = []
earn_list = []
#邏輯運算區
for i in wave_info[wave_info['type1_buy']].iterrows():
#必要資訊
start_index = i[1]['next_index'] #起始日期的INDEX
list_obj = i[1] #上面那個表(essential_data)的一列
now_index = start_index #開始迴圈時當前日期的index
#邏輯資訊
start_point = list_obj['close'] #第一波的最高點 當超過最高點時就啟動了
wave_control_condition = list_obj[
'last_mid_point'] #第一波的中間點 當低於這個點波型破壞
first_wave_gap = abs(start_point - wave_control_condition) * 2
purchase = False
purchase_price = 0
stop_loss_factor = stop_loss_factor #跌超過20%就停損
nfactor = nfactor #漲超過第一波的n倍就停利
#logic compute
while (now_index <= (len(data) - 1)):
now_price = data['close'][now_index]
now_index += 1
if (purchase):
stoploss = stoploss_or_not(purchase_price, now_price,
stop_loss_factor,
wave_control_condition)
if stoploss:
purchase = False
lose_list.append([
-now_price + purchase_price, now_price,
start_index, now_index, first_wave_gap
])
break
book_profit = book_profit_or_not(now_price, purchase_price,
first_wave_gap, nfactor)
if book_profit:
purchase = False
earn_list.append([
now_price - purchase_price, now_price, start_index,
now_index, first_wave_gap
])
break
elif (not purchase):
buy = buy_or_not(now_price, wave_control_condition,
start_point)
if buy:
purchase = True
purchase_price = now_price
else:
pass
earn_list = pd.DataFrame(earn_list,
columns=['賺了', '當前價格', '買入日', '賣出日', '波一漲幅'])
earn_list['號碼'] = stock_id
lose_list = pd.DataFrame(lose_list,
columns=['賠了', '當前價格', '買入日', '賣出日', '波一漲幅'])
lose_list['號碼'] = stock_id
return [earn_list, lose_list]
except KeyError:
print('沒抓到資料')
from FinMind.Data import Load
import numpy as np
#import pandas as pd
from keras.layers.core import Dense, Dropout
from keras.layers import LSTM, GRU
from keras.models import Sequential
from sklearn.preprocessing import MinMaxScaler
import matplotlib.pyplot as plt
import math
from keras.optimizers import Adam
TaiwanStockInfo = Load.FinData(dataset='TaiwanStockInfo')
print('input data')
data = Load.FinData(dataset='TaiwanStockPrice',
select='2330',
date='2010-01-01')
#colname = ['date', 'open', 'high', 'low', 'close', 'volume']
#data = data[colname]
print('select close price')
date = [str(d) for d in data['date']]
stock_price = data['close'].values.astype('float32')
stock_price = stock_price.reshape(len(stock_price), 1)
print(' 畫圖 ')
plt.plot(stock_price)
plt.show()
print(' 取 80% data 當作 training data, 20% data 當作 testing data 做模型驗證 ')
train_size = int(len(stock_price) * 0.8)