def get_dh(ticker, nfcs=4, nss=4, pgDb=None, sector=None):
""" get eps and stock data and then create dg,dh,dfcs
where
dg: earnings data from temp_eps table
dh: corresponding eps, differnet in PE reverse ratio, and quarterly returns
dfcs: [hwX] forecasts of [ma] period
"""
#- get EPS data
sqx = "SELECT *,\"epsDif\" as ma,rptdate::int/100 as yyyymm FROM temp_eps WHERE ticker={!r} ORDER BY rptdate".format(
ticker)
dg = sqlQuery(sqx, pgDb)
#dg = dg.rename({"pbdate":"asof"},axis='columns')
#- get closing price data
sDu = pull_stock_data(ticker, days=1000, src='iex')[['close', 'pbdate']]
sDu = sDu.rename({"close": "price"}, axis='columns')
cur_prc = sDu['price'][-1]
dx = freq_d2m(sDu, method='last', fq='M')
dx['pbdate'] = dx['pbdate'].astype('int')
dx.loc[:, 'yyyymm'] = (dx['pbdate'][:] / 100).astype('int')
#dx['rtm'] = dx['price'].diff()
dx['rtm'] = dx['price'].pct_change(
) * 100.0 # monthly return and rescale to %
dx['pr1'] = dx['price'].shift()
"""
dx['rt1'] = dx['rt'].shift(-1)
dx['rt2'] = dx['rt'].shift(-2)
dx['pb1'] = dx['pbdate'].shift(-1)
dx['pb2'] = dx['pbdate'].shift(-2)
"""
#- merge data
#dh = pd.merge(dx[['rt1','pb1','rt2','pb2','rt','yyyymm','pbdate','price']],dg[['ma','yyyymm','asof','ticker']],on=['yyyymm'])
dh = pd.merge(dx[['rtm', 'yyyymm', 'pbdate', 'price', 'pr1']],
dg[['ma', 'yyyymm', 'asof', 'ticker']],
on=['yyyymm'])
dh.loc[:, 'ma'] = dh['ma'] / dh['pr1'] * 100.0 # use PE-ratio difference
dh['rt'] = dh['price'].pct_change() * 100.0 # quarterly return
ntr = 5 # nth powter of taylor series
vv = dh['ma'].dropna().astype(float)
if len(vv) < 1:
return [], [], []
for j in range(nfcs):
vv.append(taylor_appx(vv, ntr))
hwX = vv[-nfcs:]
hwN = len(hwX)
hwS = json.dumps(list(hwX))
dfcs = pd.DataFrame(
{
'hwX': [hwS],
'ticker': [ticker],
'prc_cur': [cur_prc],
'sector': [sector]
},
columns=['sector', 'ticker', 'hwX', 'prc_cur'])
return dfcs, dh, dg
def run_fcs(ticker, debugTF=False, funcName='rForecast', **optx):
# get data
datax = pull_stock_data(ticker)
asof = int(datax['pbdate'].iloc[-1])
# idxtm=map(lambda x:datetime.datetime.strptime(str(x),"%Y%m%d"),datax['pbdate'])
# datax.set_index(pd.DatetimeIndex(idxtm),inplace=True)
if debugTF is True:
print datax.tail()
# get r-code
pandas2ri.activate()
rstring = 'source("./_alan_ohlc_fcs.r")'
r(rstring)
# convert to r-data
#df=pandas2ri.py2ri(datax[['pbdate','close']])
df = pandas2ri.py2ri(datax['close'][:])
# run r-function
opts = {
'nfcs': 30,
'dwmTF': True,
'autoArima': False,
'difTF': True,
'funcname': 'rAR',
'logTF': True,
'plevel': 0.7,
'freq': 'W'
}
opts.update(optx)
optR = subDict(opts, [
'nfcs', 'plevel', 'funcname', 'autoArima', 'logTF', 'difTF', 'freq',
'fcsLst', 'dwmTF'
])
if debugTF:
print >> sys.stderr, "==Input Args:{}".format(optR)
print >> sys.stderr, "==asof {},df:\n{}".format(
asof, datax['close'][-5:])
if funcName in robj.globalenv:
funcArg = robj.globalenv[funcName]
ret = funcArg(df, asof, debugTF=debugTF, **optR)
if opts['dwmTF'] is True:
dwm = pandas2ri.ri2py(ret[1])
dwm['ticker'] = ticker
else:
dwm = pd.DataFrame()
dd = pandas2ri.ri2py(ret[0])
dd['ticker'] = ticker
return (dd, dwm, datax)
def get_csvdata(args,
sep='|',
src=None,
days=730,
start=None,
end=None,
columns=None,
hdrLst=None):
"""
Get data in datafram with selected [columns]
"""
if isinstance(args, pd.DataFrame):
df = args
if columns is not None and df.size > 0:
df = df[list(set(df.columns) & set(columns.split(',')))]
if hdrLst is not None:
xLst, yLst = hdrLst.split('=')
xyD = dict(zip(xLst.split(','), yLst.split(',')))
df.rename(columns=xyD, inplace=True)
return df
if len(args) < 1:
return None
filename = args[0]
if filename == '-':
df = pd.read_csv(sys.stdin, sep=sep)
elif src is not None:
from _alan_calc import pull_stock_data
df = pull_stock_data(filename,
days=days,
src=src,
start=start,
end=end)
else:
df = pd.read_csv(filename, sep=sep)
if df.size < 1:
print >> sys.stderr, "**ERROR: Data not found!"
return {}
if columns is not None:
df = df[list(set(df.columns) & set(columns.split(',')))]
df.dropna(inplace=True)
if hdrLst is not None:
xLst, yLst = hdrLst.split('=')
xyD = dict(zip(xLst.split(','), yLst.split(',')))
df.rename(columns=xyD, inplace=True)
return df
def get_csvdata(args,sep='|',src=None,days=730,start=None,end=None,colLst=None,hdrLst=None,**optx):
"""
Get data in datafram with selected [colLst]
"""
if isinstance(args,pd.DataFrame):
df = args
if colLst is not None and df.size > 0:
df = df[ list(set(df.columns) & set(colLst.split(','))) ]
if hdrLst is not None:
xLst,yLst = hdrLst.split('=')
xyD = dict(zip(xLst.split(','),yLst.split(',')))
df.rename(columns=xyD,inplace=True)
return df
if len(args)<1:
return None
filename=args[0]
if filename=='-':
df=pd.read_csv(sys.stdin,sep=sep)
elif src.upper()=='FILENAME':
df=pd.read_csv(filename,sep=sep)
elif len(args)>1:
from yh_hist_batch import batch_yh_hist
df = batch_yh_hist(args,days=days,src=src,start=start,end=end,**optx)
elif src is not None:
from _alan_calc import pull_stock_data
df = pull_stock_data(filename,days=days,src=src,start=start,end=end,**optx)
else:
df = pd.read_csv(filename,sep=sep)
if df.size < 1:
sys.stderr.write("**ERROR:{} @{}()".format('Data Not Found!','get_csvdata'))
return {}
if colLst is not None:
df = df[ list(set(df.columns) & set(colLst.split(','))) ]
df.dropna(inplace=True)
if hdrLst is not None:
xLst,yLst = hdrLst.split('=')
xyD = dict(zip(xLst.split(','),yLst.split(',')))
df.rename(columns=xyD,inplace=True)
return df
def run_ohlc_fcs(ticker, opts=None, debugTF=False, pgDB=None, **kwargs):
"""
forecast 'nfcs' periods based on raw data 'datax'
return (dd,dwm,datax)
where
dd: forecast values
dwm: forecast values of next day, week, month (optional)
datax: data used for forecast calculation
Note, dwm can be optional if dwmTF is False
"""
if opts is None:
(opts, _) = opt_ohlc_fcs([])
if len(kwargs) > 0:
opts.update(kwargs)
if debugTF:
pqint(opts, file=sys.stderr)
days = getKeyVal(opts, 'days', 730)
freq = getKeyVal(opts, 'freq', 'D')
# get data
if isinstance(ticker, pd.DataFrame):
datax = ticker
ticker = ''
else: # get data
optx = subDict(opts, ['src', 'days', 'start', 'end'])
datax = pull_stock_data(ticker, pgDB=pgDB, **optx)
if 'ticker' in datax:
ticker = datax['ticker'].iloc[0]
if datax is None or len(datax) < 1:
return (None, None, None)
#idxtm=map(lambda x:datetime.datetime.strptime(str(x),"%Y%m%d"),datax['pbdate'])
#datax.set_index(pd.DatetimeIndex(idxtm),inplace=True)
if debugTF is True:
pqint(opts, file=sys.stderr)
pqint(datax.tail(), file=sys.stderr)
nobs = days
if 'epochs' in datax:
asof = int(
datetime.datetime.fromtimestamp(
int(datax['epochs'].iloc[-1]) / 1000).strftime('%Y%m%d'))
fcsLst = np.array([5, 10, 30])
else:
asof = int(datax['pbdate'].iloc[-1])
fcsLst = np.array([1, 5, 23])
vprc = datax['close'][-nobs:]
# get r-code
pandas2ri.activate()
fpath = os.path.dirname(__file__)
if len(fpath) < 1:
fpath = '.'
rstring = 'source("{}/{}")'.format(fpath, "_alan_ohlc_fcs.r")
if debugTF:
pqint(rstring, file=sys.stderr)
r(rstring)
# convert to r-data
df = pandas2ri.py2ri(vprc)
# run r-function [rGARCH|rAR]
optx = subDict(opts, [
'nfcs', 'plevel', 'funcname', 'autoArima', 'logTF', 'difTF', 'freq',
'fcsLst', 'dwmTF'
])
if debugTF:
pqint("==Input Args:{}".format(optx), file=sys.stderr)
pqint("==df\n:{}".format(vprc.tail()), file=sys.stderr)
ret = robj.globalenv['rForecast'](df, asof, debugTF=debugTF, **optx)
#ret=robj.globalenv['rForecast'](df,asof,plevel=plevel,funcname=funcname,autoArima=autoArima,debugTF=debugTF,logTF=logTF,difTF=difTF,freq=freq,fcsLst=fcsLst)
if opts['dwmTF'] is True:
dwm = pandas2ri.ri2py(ret[1])
dwm['ticker'] = ticker
else:
dwm = pd.DataFrame()
dd = pandas2ri.ri2py(ret[0])
dd['ticker'] = ticker
dd['freq'] = freq
return (dd, dwm, datax)
def run_alan_plot(tkLst, opts=None, optx=None):
if opts is None:
opts, _ = opt_alan_plot([])
if optx is not None:
opts.update(optx)
for ky, va in opts.items():
exec("{}=va".format(ky))
colorUD = ['red', 'green'] if lang == 'cn' else ['green', 'red']
if pngDIR is None or os.path.isdir(pngDIR) is False:
pngDIR = "/home/web/bb_site/html/images"
if debugTF is True:
pqint("===options:{}".format(opts), file=sys.stderr)
for stock in tkLst:
try:
if chartType == 'minute':
optx = {'gap': '1m', 'ranged': '1d', 'outTF': False}
datax = pull_stock_data(stock, **optx)
pqint(datax.tail(), file=sys.stderr)
else:
datax = pull_stock_data(stock, start, end)
except Exception as e:
pqint("***Data ERROR:", str(e), file=sys.stderr)
continue
if chartType == 'minute' and 'epochs' in datax:
datax['pbdatetime'] = datax.epochs.apply(epoch_parser)
dtfmt = '%H:%M %b %d'
else:
datax['pbdatetime'] = datax.pbdate.apply(ymd_parser)
dtfmt = '%m-%d-%y'
try:
block = False if bscTF is True else True
datax.dropna(inplace=True)
pqint(datax.tail(), file=sys.stderr)
if chartType != 'minute':
datax.reset_index(inplace=True)
datax['mpldatetime'] = datax.pbdatetime.apply(mdates.date2num)
#- calc macd & rsi
datax = run_tech(datax,
pcol='close',
winLst=[ma1, ma2],
debugTF=debugTF,
nanTF=True)
fig, axes = plot_candlestickCombo(datax,
stock,
ma1,
ma2,
savePng=savePng,
block=block,
pngDIR=pngDIR,
chartType=chartType,
trendTF=trendTF,
npar=npar,
debugTF=debugTF,
colorUD=colorUD,
title=title)
if savePng is False:
plt.show(axes)
# DEPRECATED, use run_tech() + plot_candlestickCombo()
#graphData(datax,stock,ma1,ma2,savePng=savePng,block=block,pngDIR=pngDIR,chartType=chartType)
if bscTF is True:
run_ohlc(datax)
bsc_plot(datax,
stock,
days=7,
savePng=savePng,
block=True,
pngDIR=pngDIR,
dtfmt=dtfmt)
except Exception as e:
pqint(str(e), file=sys.stderr)
return datax