def __init__(self, data, tii, ti):
self.data = data
self.tii = tii
self.ti = ti
graphdata = tind.display_indicator(self.data, self.tii.indicator.name, self.tii)
self.ma1 = graphdata[0]['y']
self.ma2 = graphdata[1]['y']
def train(self):
cols = []
params1 = []
for ii in self.tii.indicator.indicatorinputs.all():
params1.append(get_input_value(self.tii, ii.parameter))
cols.append(ii.parameter)
if len(params1) > 0:
params1 = prepare_params_for_psar(*params1)
cols.extend(['Returns%', 'MaxR'])
pret = pd.DataFrame(columns=cols)
arglist = [[]] # make all possible parameter combinations
for p in params1:
newarglist = []
for pp in p:
for argset in arglist:
newargset = argset.copy()
newargset.append(pp)
newarglist.append(newargset)
arglist = newarglist
signalsmap = dict()
ipret = 0
for argset in arglist:
graphdata = tind.display_indicator(self.data,
self.tii.indicator.name,
self.tii, True, *argset)
for pltdt in graphdata:
if pltdt['name'] == 'psar':
self.psar = pltdt['y']
if pltdt['name'] == 'psarbull':
self.psarbull = pltdt['y']
if pltdt['name'] == 'psarbear':
self.psarbear = pltdt['y']
signal_graph, signals, traderet = self.trigger()
if traderet['winlossratio'] is None or traderet[
'winlossratio'] == np.inf:
traderet = 0
else:
traderet = traderet['winlossratio']
parama = argset.copy()
parama.extend([traderet, ''])
pret.loc[ipret] = parama
signalsmap[ipret] = signals
ipret += 1
retcol = pret['Returns%']
imax = np.array(retcol).argmax()
pret.at[imax, 'MaxR'] = 'MR'
psetb = pret.loc[imax]
return psetb, pret, signalsmap[imax]
def __init__(self, data, tii, ti):
self.data = data
self.tii = tii
self.ti = ti
graphdata = tind.display_indicator(self.data, self.tii.indicator.name, self.tii)
for pltdt in graphdata:
if pltdt['name'] == 'fish':
self.fish=pltdt['y']
break
def __init__(self, data, tii, ti):
self.data = data
self.tii = tii
self.ti = ti
graphdata = tind.display_indicator(self.data, self.tii.indicator.name, self.tii)
for pltdt in graphdata:
if pltdt['name'] == 'adx_pos':
self.adx_pos=pltdt['y']
if pltdt['name'] == 'adx_neg':
self.adx_neg=pltdt['y']
def __init__(self, data, tii, ti):
self.data = data
self.tii = tii
self.ti = ti
graphdata = tind.display_indicator(self.data, self.tii.indicator.name,
self.tii)
for pltdt in graphdata:
if pltdt['name'] == 'ic_a':
self.ic_a = pltdt['y']
if pltdt['name'] == 'ic_b':
self.ic_b = pltdt['y']
def __init__(self, data, tii, ti):
self.data = data
self.tii = tii
self.ti = ti
graphdata = tind.display_indicator(self.data, self.tii.indicator.name,
self.tii)
for pltdt in graphdata:
if pltdt['name'] == 'aroon_up':
self.aroon_up = pltdt['y']
elif pltdt['name'] == 'aroon_down':
self.aroon_down = pltdt['y']
def __init__(self, data, tii, ti):
self.data = data
self.tii = tii
self.ti = ti
graphdata = tind.display_indicator(self.data, self.tii.indicator.name,
self.tii)
for pltdt in graphdata:
if pltdt['name'] == 'smi':
self.smi = pltdt['y']
if pltdt['name'] == 'smi_sig':
self.smi_sig = pltdt['y']
def __init__(self, data, tii, ti):
self.data = data
self.tii = tii
self.ti = ti
graphdata = tind.display_indicator(self.data, self.tii.indicator.name,
self.tii)
for pltdt in graphdata:
if pltdt['name'] == 'psar':
self.psar = pltdt['y']
if pltdt['name'] == 'psarbull':
self.psarbull = pltdt['y']
if pltdt['name'] == 'psarbear':
self.psarbear = pltdt['y']
def train(self):
cols = []
params1 = []
for ii in self.tii.indicator.indicatorinputs.all():
params1.append(get_input_value(self.tii, ii.parameter))
cols.append(ii.parameter)
psetb, pret = train_for_kst(self.data, *params1)
graphdata = tind.display_indicator(self.data, self.tii.indicator.name, self.tii, True, *psetb)
for pltdt in graphdata:
if pltdt['name'] == 'KST':
self.kst=pltdt['y']
break
signal_graph, signals, traderet = self.trigger()
return psetb, pret, signals
def train(self):
cols = []
params1 = []
for ii in self.tii.indicator.indicatorinputs.all():
params1.append(get_input_value(self.tii, ii.parameter))
cols.append(ii.parameter)
if len(params1) > 0:
params1 = prepare_params_for_bb_bw(*params1)
cols.extend(['ovb', 'ovs', 'Returns%', 'MaxR'])
pret = pd.DataFrame(columns=cols)
arglist = [[]] # make all possible parameter combinations
for p in params1:
newarglist = []
for pp in p:
for argset in arglist:
newargset = argset.copy()
newargset.append(pp)
newarglist.append(newargset)
arglist = newarglist
signalsmap = dict()
ipret = 0
for argset in arglist:
graphdata = tind.display_indicator(self.data,
self.tii.indicator.name,
self.tii, True, *argset)
for pltdt in graphdata:
if pltdt['name'] == 'bb_bw':
pltdt1 = pltdt['y']
break
maxv = max(pltdt1)
minv = min(pltdt1)
ovba = np.arange((maxv + minv) / 2,
maxv, (maxv - minv) / 20,
dtype=float)
ovsa = np.arange((maxv + minv) / 2,
minv,
-(maxv - minv) / 20,
dtype=float)
ovba = np.delete(ovba, 0)
ovsa = np.delete(ovsa, 0)
for ovb in ovba:
for ovs in ovsa:
self.bb_bw = pltdt1
signal_graph, signals, traderet = self.trigger(ovb, ovs)
if traderet['winlossratio'] is None or traderet[
'winlossratio'] == np.inf:
traderet = 0
else:
traderet = traderet['winlossratio']
parama = argset.copy()
parama.extend([ovb, ovs, traderet, ''])
pret.loc[ipret] = parama
signalsmap[ipret] = signals
ipret += 1
retcol = pret['Returns%']
imax = np.array(retcol).argmax()
pret.at[imax, 'MaxR'] = 'MR'
psetb = pret.loc[imax]
return psetb, pret, signalsmap[imax]
def train_for_threshold(df, tii, plt, recact=False):
""" Train for plot vs plot in.
Return:
psetb: Parameter set with highest trade returns.
pret: Frame with parameters and returns values.
"""
cols = []
params1 = []
for ii in tii.indicator.indicatorinputs.all():
params1.append(get_input_value(tii, ii.parameter))
cols.append(ii.parameter)
if len(params1) > 0:
params1 = globals()['prepare_params_for_'+tii.indicator.id_letter](*params1)
cols.extend(['ovb', 'ovs', 'Returns%', 'MaxR'])
pret = pd.DataFrame(columns=cols)
arglist = [[]] # make all possible parameter combinations
for p in params1:
newarglist = []
for pp in p:
for argset in arglist:
newargset = argset.copy()
newargset.append(pp)
newarglist.append(newargset)
arglist = newarglist
sargs = [trade_with_threshold]
bssigmap = dict()
stopsigmap = dict()
trendsigmap = dict()
sigtypemap = dict()
ipret = 0
for argset in arglist:
graphdata = tind.display_indicator(df, tii.indicator.name, tii, True, *argset)
for pltdt in graphdata:
if pltdt['name'] == plt:
pltdt1=pltdt['y']
break
maxv = max(pltdt1)
minv = min(pltdt1)
ovba = np.arange((maxv+minv)/2, maxv, maxv/10, dtype=int)
ovsa = np.arange((maxv+minv)/2, minv, minv/10, dtype=int)
for ovb in ovba:
for ovs in ovsa:
bsrets, sigtype, stopsig, trendsig, bssig = trade_with_threshold(df, pltdt1, ovb, ovs, True)
if bsrets.empty:
traderet = 0
else:
traderet = bsrets['Returns%'].iloc[-1] # Last index for final return
parama = argset.copy()
parama.extend([ovb, ovs, traderet, ''])
pret.loc[ipret] = parama
bssigmap[ipret] = bssig
stopsigmap[ipret] = stopsig
trendsigmap[ipret] = trendsig
sigtypemap[ipret] = sigtype
ipret+=1
retcol = pret['Returns%']
imax = np.array(retcol).argmax()
pret.at[imax, 'MaxR'] = 'MR'
psetb = pret.loc[imax]
return psetb, pret, sigtypemap[imax], stopsigmap[imax], trendsigmap[imax], bssigmap[imax]
def train_for_cross2(df, tii1, tii2, plt1, plt2, recact=False):
""" Train for plot vs plot in.
Return:
psetb: Parameter set with highest trade returns.
pret: Frame with parameters and returns values.
"""
cols = []
params1 = []
for ii in tii1.indicator.indicatorinputs.all():
params1.append(get_input_value(tii1, ii.parameter))
cols.append(ii.parameter)
params2 = []
for ii in tii2.indicator.indicatorinputs.all():
params2.append(get_input_value(tii2, ii.parameter))
cols.append(ii.parameter)
if len(params1) > 0:
params1 = globals()['prepare_params_for_'+tii1.indicator.id_letter](*params1)
if len(params2) > 0:
params2 = globals()['prepare_params_for_'+tii2.indicator.id_letter](*params2)
cols.extend(['Returns%', 'MaxR'])
pret = pd.DataFrame(columns=cols)
arglist = [[]] # make all possible parameter combinations
for p in params1:
newarglist = []
for pp in p:
for argset in arglist:
newargset = argset.copy()
newargset.append(pp)
newarglist.append(newargset)
arglist = newarglist
arglist1 = arglist
arglist = [[]] # make all possible parameter combinations
for p in params2:
newarglist = []
for pp in p:
for argset in arglist:
newargset = argset.copy()
newargset.append(pp)
newarglist.append(newargset)
arglist = newarglist
arglist2 = arglist
sargs = [trade_with_cross2]
bssigmap = dict()
stopsigmap = dict()
trendsigmap = dict()
sigtypemap = dict()
ipret = 0
for argset1 in arglist1:
graphdata1 = tind.display_indicator(df, tii1.indicator.name, tii1, True, *argset1)
dfnew = df.copy()
for pltdt in graphdata1:
dfnew[pltdt['name']] = pltdt['y']
if pltdt['name'] == plt1:
pltdt1=pltdt['y']
for argset2 in arglist2:
graphdata2 = tind.display_indicator(dfnew, tii2.indicator.name, tii2, True, *argset2)
for pltdt in graphdata2:
if pltdt['name'] == plt2:
pltdt2=pltdt['y']
break
bsrets, sigtype, stopsig, trendsig, bssig = trade_with_cross2(df, pltdt1, pltdt2, True)
if bsrets.empty:
traderet = 0
else:
traderet = bsrets['Returns%'].iloc[-1] # Last index for final return
parama = argset1.copy()
parama.extend(argset2)
parama.extend([traderet, ''])
bssigmap[ipret] = bssig
stopsigmap[ipret] = stopsig
trendsigmap[ipret] = trendsig
sigtypemap[ipret] = sigtype
pret.loc[ipret] = parama
ipret+=1
retcol = pret['Returns%']
imax = np.array(retcol).argmax()
pret.at[imax, 'MaxR'] = 'MR'
psetb = pret.loc[imax]
return psetb, pret, sigtypemap[imax], stopsigmap[imax], trendsigmap[imax], bssigmap[imax]
