def TransformToBinary(dataset , enable_indicator = True , pred_days = 7):
# data: 1 ~ -31 days, every day will compare to previous 1 day (5) and 30 days (5) and next 7 day (1).
# Hence, binary string will be (N-30-7)x(5+5+1) array
size = dataset.shape[0]-30 - pred_days
binary_string = np.zeros((size , 11) , dtype=np.int)
for i in range(size):
index = i + pred_days # get the index of dataset
thisday = dataset[index]
nextday = dataset[index-pred_days] #next day
prvday = dataset[index+1] #previous day
monthdata = dataset[index+1:index+31] #previous 30 days
averagedata = np.sum(monthdata ,axis= 0)/30 #average of previous 30 days
binary_string[i][:5] = thisday[1:] > prvday[1:] #represent the up and downs compared to previous day
binary_string[i][5:10] = thisday[1:] > averagedata[1:] #represent the up and downs compared to prvious 30 days
binary_string[i][-1] = nextday[1] > thisday[1] # up and down
if enable_indicator:
arr1 = indicator.probability(dataset, 30 , pred_days).reshape(-1,1)
arr2_1 = indicator.comAvg(dataset, 10, 5 , pred_days).reshape(-1,1)
arr2_2 = indicator.comAvg(dataset, 30, 10 ,pred_days).reshape(-1,1)
arr2_3 = indicator.comAvg(dataset, 30, 7 , pred_days).reshape(-1,1)
arr2 = np.hstack((arr2_1, arr2_2, arr2_3))
arr3 = indicator.RSI(dataset , pred_days)
input_string = np.hstack((arr1, arr2, arr3, binary_string))
else:
input_string = binary_string
return input_string
def indicatorPage():
print('indicatorPage', file=sys.stderr)
# POST
if request.method == 'GET':
try:
stock_name = request.args['stock_name']
indicatorSelect = request.args['indicatorSelect']
#lists = None
if indicatorSelect == "1":
lists = indicator.MTM(stock_name)
return render_template('indicatorPage.html', lists=json.dumps(lists), indicatorSelect="MTM")
elif indicatorSelect == "2":
lists = indicator.RSI(stock_name)
return render_template('indicatorPage.html', lists=json.dumps(lists), indicatorSelect="RSI")
else:
lists = indicator.MACD(stock_name)
return render_template('macdIndicator.html', lists=json.dumps(lists), indicatorSelect="MACD")
#except Exception as e: print(e)
except:
return 'Fail'
# if not POST, then it is GET
return render_template('dashboardUser.html')
def Xindicator(self, syms, dates):
sma = ind.SMA(syms, 20, dates)
bb = ind.BB(syms, 20, 2, dates)
_, macd = ind.MACD(syms, 12, 26, 9, dates)
rsi = ind.RSI(syms, 14, dates)
X = pd.DataFrame({'SMA': sma, 'BB': bb, "MACD": macd, "RSI": rsi})
return X
def calculate_thres():
symbol = 'JPM'
sd=dt.datetime(2008,1,1)
ed=dt.datetime(2009,12,31)
date = pd.date_range(sd - dt.timedelta(days=50), ed)
stockvals = get_data([symbol], date)
stockvals = stockvals.loc[:, [symbol]]
sma = ind.SMA(symbol, 20, date, gen_plot = False)
sma = sma[sma.index >= sd]
bb = ind.BB(symbol, 20, 2, date, gen_plot = False)
bb = bb[bb.index >= sd]
_, macd = ind.MACD(symbol, 12, 26, 9, date, gen_plot = False)
macd = macd[macd.index >= sd]
rsi = ind.RSI(symbol, 14, date, gen_plot = False)
rsi = rsi[rsi.index >= sd]
bounds = [(0, 1), (60, 85), (15,40)]
optimal_thres = differential_evolution(optimalize, bounds, args=(sma, bb, macd, rsi, ))
return(optimal_thres)
def testPolicy(symbol = 'JPM', sd=dt.datetime(2008,1,1), ed=dt.datetime(2009,12,31), sv = 100000):
date = pd.date_range(sd - dt.timedelta(days=50), ed)
stockvals = get_data([symbol], date)
stockvals = stockvals.loc[:, [symbol]]
sma = ind.SMA(symbol, 20, date, gen_plot = False)
sma = sma[sma.index >= sd]
bb = ind.BB(symbol, 20, 2, date, gen_plot = False)
bb = bb[bb.index >= sd]
_, macd = ind.MACD(symbol, 12, 26, 9, date, gen_plot = False)
macd = macd[macd.index >= sd]
rsi = ind.RSI(symbol, 14, date, gen_plot = False)
rsi = rsi[rsi.index >= sd]
thres = [0.99243522, 84.70103144, 25.01519814]
"""
from calculate_thres
"""
order_signal = signal(sma, bb, macd, rsi, thres)
order = Order(order_signal)
return order
def createFactorsTable(instrument=None, dbpath=None, con=None, **factors):
'''
Only have to input on of these 3:
:param instrument:
:param dbpath:
:param con:
:param factors: No use temporarily
:return:
How to use:
createFactorsTable(instrument='EUR_USD')
-------------------------------------------------------------
createFactorsTable(dbpath='E:/FinanceData/Oanda/EUR_USD.db')
-------------------------------------------------------------
con=sqlite3.connect('E:/FinanceData/Oanda/EUR_USD.db')
createFactorsTable(con=con)
con.close()
'''
close = False
if len(factors) == 0:
factors = {
'D': ['time', 'closeBid', 'highBid', 'lowBid'],
'COT': ['publish', 's-l', 's-l_diff'],
'HPR': ['time', 'position', 'position_diff']
}
if dbpath is None:
dbpath = '%s/%s.db' % (savePath, instrument)
print(dbpath)
if con is None:
con = sqlite3.connect(dbpath)
close = True
data = {}
start = 0
for k in factors.keys():
f = factors[k]
data[k] = read_sql(k, con=con).get(f)
startTime = data[k].get_value(0, f[0])
if start < startTime:
start = startTime
data['COT'].columns = ['time', 's-l', 's-l_diff']
if close:
con.close()
price = data['D']
momentum = indicator.momentum(price['time'], price['closeBid'], period=60)
data['momentumfast'] = momentum
momentum = indicator.momentum(price['time'], price['closeBid'], period=130)
data['momentumslow'] = momentum
data['atr'] = indicator.ATR(price['time'],
price['highBid'],
price['lowBid'],
price['closeBid'],
period=10)
data['mafast'] = indicator.MA(price['time'],
price['closeBid'],
period=60,
compare=True)
data['maslow'] = indicator.MA(price['time'],
price['closeBid'],
period=130,
compare=True)
adx = indicator.ADX(price['time'],
price['highBid'],
price['lowBid'],
price['closeBid'],
period=10)
data['ADX'] = adx
data['RSI'] = indicator.RSI(price['time'], price['closeBid'], period=10)
data['MACD'] = indicator.MACD(price['time'],
price['closeBid'],
out=['hist'])
histdiff = [0]
for h in data['MACD'].index.tolist()[1:]:
histdiff.append(data['MACD'].get_value(h, 'hist') -
data['MACD'].get_value(h - 1, 'hist'))
data['MACD'].insert(2, 'hist_diff', histdiff)
data['ADX-mom'] = indicator.momentum(adx['time'],
adx['ADX%s' % 10],
period=5)
data['ADX-mom'].columns = ['time', 'ADX-mom']
data.pop('D')
out = None
for k in sorted(data.keys()):
# if k == 'D':
# continue
v = data[k]
select = v[v.time >= start]
for i in select.index:
t = select.get_value(i, 'time')
select.set_value(i, 'time', datetime.date.fromtimestamp(t))
# print(select)
if out is None:
out = select.drop_duplicates('time').set_index('time')
else:
out = out.join(select.drop_duplicates('time').set_index('time'),
how='outer')
for c in out.columns:
former = out.get_value(out.index.tolist()[0], c)
for t in out.index.tolist()[1:]:
v = out.get_value(t, c)
# print(t,c,v,type(v))
if math.isnan(v):
out.set_value(t, c, former)
former = out.get_value(t, c)
return out.dropna()
def rsi(self, data, column='Velocity', period=5):
RSI = ind.RSI(data['time'], data[column], period)
return data.merge(RSI, 'outer', 'time')
