def plotdivratio(code):
sz50code, namedf, price, volume = getdata.GetPriceVolume(code)
div = getdata.GetDiv('divi').ix[code]
reportdate = getdata.GetDiv('report_date').ix[code]
print div, reportdate
##do backfill
divvalue = Series()
for idx in range(0, len(reportdate.index)):
if idx != len(reportdate.index) - 1:
startdate = reportdate[reportdate.index[idx]]
enddate = reportdate[reportdate.index[idx + 1]]
else:
startdate = reportdate[reportdate.index[idx]]
enddate = '2017-04-28'
print startdate, enddate
dr = pd.date_range(startdate, end=enddate, freq='B')
ctr = dr.to_pydatetime()
for date in ctr:
divvalue[date] = div[div.index[idx]]
daterange = pd.date_range('01/01/2012', end='04/28/2017', freq='B')
convertrange = daterange.to_pydatetime()
ratio = Series()
for date in convertrange:
if date in price.index and date in divvalue.index:
ratio[date] = divvalue[date] / price[code][date]
#print ratio
fig, ax = plt.subplots()
ratio.plot(ax=ax)
plt.show()
def plotpe(ax):
global profit, hs300
daterange = pd.date_range('01/01/2012', end='03/03/2017', freq='B')
#daterange = pd.date_range('03/01/2014', end = '05/20/2014', freq='B')
convertrange = daterange.to_pydatetime()
sz50code, namedf, price, volume = getdata.GetPriceVolume()
epsall = getdata.GetProfit('eps')
peseries = Series()
validpe = Series()
for date in convertrange:
if date not in price.index:
continue
combinepe = 0
for code in hs300.index:
weight = hs300['weight'][code]
eps = epsall.ix[code]
quarter = date.month / 4 + 1
if date.year == 2012:
earning = eps[20124]
elif date.year > 2016:
earning = eps[20163] + eps[20154] - eps[20153]
else:
if quarter == 1:
earning = eps[(date.year - 1) * 10 + 4]
else:
earning = eps[date.year * 10 + quarter - 1] + eps[
(date.year - 1) * 10 + 4] - eps[(date.year - 1) * 10 +
quarter - 1]
temppe = float(price[code][date]) / earning
if not isnan(temppe):
validpe[code] = temppe
if isnan(temppe) and code in validpe.index:
temppe = validpe[code]
if (not isnan(temppe)) and temppe > 0:
combinepe = combinepe + weight * temppe
#print combinepe
#if date.year == 2014 and date.month==4 and date.day == 3 and temppe < 0:
# print date, code, temppe, earning
peseries[date] = combinepe
print date, combinepe
#normalizedpe = (peseries - peseries.min()) / (peseries.max() - peseries.min())
ax.set_title("hs300PE")
peseries.plot(ax=ax)
def plotcorrelation():
code1 = '601668'
sz50code1, namedf1, price1, volume1 = getdata.GetPriceVolume()
price1 = price1['601668'].pct_change()[1:]
#price1 = price1.apply(lambda x: (x - np.mean(x)) / (np.max(x) - np.min(x)))
volume1 = volume1.apply(lambda x: (x - np.mean(x)) /
(np.max(x) - np.min(x)))
volume_ma = pd.rolling_mean(volume1, window=30)
#price_ma = pd.rolling_mean(price1, window = 30)
#ss = getdata.GetWindowCorr(price_ma[code1], volume_ma[code1], 90)
#price_ma[code1].plot(color='r')
price1.plot(label="PE")
#volume_ma[code1].plot(color='y')
#ss.plot(color='b')
plt.show()
def GetHighestDiv():
global divi, hs300
sz50code, namedf, price, volume = getdata.GetPriceVolume()
dt = datetime.datetime(2017, 04, 28)
df = DataFrame(columns=[2013, 2014, 2015, 'avg', 'price', 'ratio', 'name'],
index=hs300.index)
for ii in df.index:
for year in [2013, 2014, 2015]:
df[year][ii] = divi[year][ii]
#print ii, df[2013][ii] , df[2014][ii] , df[2015][ii]
df['avg'][ii] = float(df[2013][ii] + df[2014][ii] + df[2015][ii]) / 3
df['price'][ii] = price[ii][dt]
df['ratio'][ii] = df['avg'][ii] / price[ii][dt]
df['name'][ii] = hs300['name'][ii]
#print df.ix[ii]
topcode = df.sort(columns='ratio', ascending=False)[0:50]
for code in topcode.index:
if not np.isnan(divi[2016][code]):
print code, divi[2016][code] / price[code][dt], df['name'][code]
currentquarter = int((month - 1) / 3) + 1
pair = [(year - 1, 3), (year - 1, 4), (year, 1), (year, 2), (year, 3)]
return pair[currentquarter]
def getsecondlastquarter(date):
year = date.year
month = date.month
currentquarter = int((month - 1) / 3) + 1
pair = [(year - 1, 3), (year - 1, 4), (year, 1), (year, 2), (year, 3)]
return pair[currentquarter - 1]
sz50code, namedf, price, volume = getdata.GetPriceVolume()
daterange = pd.date_range(startdate, end=genddate, freq='B')
convertrange = daterange.to_pydatetime()
price = price.ix[convertrange[0]:convertrange[-1]]
def getvalue(tsseries, date):
quarter = int((date.month - 1) / 3) + 1
lastq = getlastquarter(date)
if date.year == 2011:
earning = tsseries[20114]
## data for most recent days, first check wheter value of last quarter is valid, if yes, use last quarter;
## if not, use second last quarter
elif date > convertrange[-180]:
ret[d] = firstunzero
return ret
date = pd.date_range('01/01/2012', end='06/30/2017')
#sz50 = ts.get_sz50s()
sz50 = getdata.GetHs300()
#sz50 = getdata.GetHs300()
#sz50 = sz50.append({'code':'sh', 'name':'index','date': '2016-08-01','weight':0}, ignore_index=True)
sz50code = sz50.index
df = DataFrame(columns=sz50code)
rawdict = {}
namedf = {}
sz50code, nameseries, price, volume = getdata.GetPriceVolume()
corrdf = price.corr()
vvdf = volume.corr()
#print corrdf['000625']['sh']
#print corrdf['000423']['sh']
#print corrdf['000963']['sh']
#codelist = ['000625', '000423', '000963', '600887', '600048', '600066']
codelist = download.LoadCodelist()
for code in codelist:
for c2 in codelist:
if code != c2:
print code, c2
print "%s: %s : %f, %f" % (sz50['name'][code].decode('UTF-8'),
sz50['name'][c2].decode('UTF-8'),
corrdf[code][c2], vvdf[code][c2])
def plotpb(code, ax, name):
global profit, hs300, basic, roe, retpbarray
sz50code, namedf, price, volume = getdata.GetPriceVolume(code)
print "PB: %s: %s" % (code, name)
daterange = pd.date_range('01/01/2012', end=genddate, freq='B')
convertrange = daterange.to_pydatetime()
netprofit = profit.ix[code] / basic[code] / 100
roeix = roe.ix[code]
#print netprofit
#print roeix
#price = price.ix[convertrange[-1]:convertrange[0]]
pbseries = Series()
earningseries = Series()
for date in convertrange:
if date in price.index:
if date.month % 3 == 0:
quarter = date.month / 3
else:
quarter = date.month / 3 + 1
if date.year == 2011:
earning = netprofit[20114]
temproe = roeix[20114]
elif (date.year == 2017 and quarter == 4) or date.year > 2017:
earning = netprofit[20173]
temproe = roeix[20173]
# temproe = roeix[20163] + roeix[20172] - roeix[20162]
# if np.isnan(earning):
# earning = netprofit[20163] + netprofit[20154] - netprofit[20153]
# temproe = roeix[20164]
# if np.isnan(temproe):
# temproe = roeix[20163] + roeix[20154] - roeix[20153]
else:
if quarter == 1:
earning = netprofit[(date.year - 1) * 10 + 4]
temproe = roeix[(date.year - 1) * 10 + 4]
else:
earning = netprofit[date.year * 10 + quarter - 1]
temproe = roeix[date.year * 10 + quarter - 1]
book = earning * 100 / temproe
#if code == '600340' and date.year == 2017 and quarter > 2 :
# print date, earning, temproe, earning * 100 / temproe, price[code][date], price[code][date] / book
temp = price[code][date] / book
#print code, date, price[code][date]
pbseries[date] = temp
#print date, earning, temproe, book, price[code][date], temp
earningseries[date] = earning
normalizedpe = pbseries
print code, normalizedpe.ix[-1], normalizedpe.mean(), normalizedpe.ix[
-1] / normalizedpe.mean(), normalizedpe.std() / normalizedpe.mean()
sortedpe = normalizedpe.sort_values()
print code, sortedpe.index[0], sortedpe[
sortedpe.index[0]], sortedpe.index[-1], sortedpe[sortedpe.index[-1]]
retpbarray[code] = normalizedpe.ix[-1] / sortedpe[sortedpe.index[0]]
ax.set_title(code + ' PB')
normalizedpe.plot(ax=ax, label="PB")
#normalizedpe = (peseries - peseries.min()) / (peseries.max() - peseries.min())
#normalizedearning = (earningseries - earningseries.min()) / (earningseries.max() - earningseries.min())
#ax.set_yticks(np.arange(0.,1.3,0.1))
#normalizedprice.plot(ax = ax, label="Price")
print '------------------------------------------'
def plotpe(code, ax, name):
global profit, hs300, basic, retpearray
sz50code, namedf, price, volume = getdata.GetPriceVolume(code)
print "PE: %s: %s" % (code, name)
daterange = pd.date_range('01/01/2012', end=genddate, freq='B')
convertrange = daterange.to_pydatetime()
netprofit = profit.ix[code] / basic[code] / 100
#print netprofit
#price = price.ix[convertrange[-1]:convertrange[0]]
#print price
peseries = Series()
earningseries = Series()
for date in convertrange:
if date in price.index:
if date.month % 3 == 0:
quarter = date.month / 3
else:
quarter = date.month / 3 + 1
if date.year == 2011:
earning = netprofit[20114]
elif (date.year == 2017 and quarter == 4) or date.year > 2017:
earning = netprofit[20173] + netprofit[20164] - netprofit[20163]
# earning = netprofit[20164]
# if np.isnan(earning):
# earning = netprofit[20163] + netprofit[20154] - netprofit[20153]
else:
if quarter == 1:
earning = netprofit[(date.year - 1) * 10 + 4]
else:
earning = netprofit[
date.year * 10 + quarter -
1] + netprofit[(date.year - 1) * 10 + 4] - netprofit[
(date.year - 1) * 10 + quarter - 1]
temp = price[code][date] / earning
#print date, price[code][date], earning
peseries[date] = temp
earningseries[date] = earning
#print peseries.min(), peseries.max(), peseries.mean(), peseries.median()
#normalizedpe = (peseries - peseries.min()) / (peseries.max() - peseries.min())
normalizedpe = peseries
#print normalizedpe
print code, normalizedpe.ix[-1], normalizedpe.mean(), normalizedpe.ix[
-1] / normalizedpe.mean(), normalizedpe.std() / normalizedpe.mean()
sortedpe = normalizedpe.sort_values()
print code, sortedpe.index[0], sortedpe[
sortedpe.index[0]], sortedpe.index[-1], sortedpe[sortedpe.index[-1]]
retpearray[code] = normalizedpe.ix[-1] / sortedpe[sortedpe.index[0]]
ax.set_title(code + ' PE')
normalizedpe.plot(ax=ax, label="PE")
print '------------------------------------------'