def __init__(self, trading_rec, param_cols):
self.trarec_grp = trading_rec.groupby(param_cols)
self.trading_rec = trading_rec
self.bt_res = self.trarec_grp.apply(
lambda x: gbt.BT_wCost(x)[0]).unstack(level=-1).fillna(np.nan)
self.trec = self.trarec_grp.apply(lambda x: gbt.BT_wCost(x)[1])
self.trec_unstack = self.trec.unstack(param_cols)
self.underly = self.trec['Underlying'].unique()
self.indiv_tr = {
ud: self.trec[self.trec['Underlying'] == ud]
for ud in self.underly
}
def print_basic(self, coerext='145500'):
idx_coer = self.trec['dt_Exit'].dt.strftime('%H:%M:%S') == coerext
coerext_pct = gfc.true_pct(idx_coer)
print '%Coerce Exit: ', coerext_pct, '\n\n'
if coerext_pct > 0:
emat_coerce, trec_coerce = gbt.BT_wCost(self.trec[idx_coer])
emat_complete, trec_complete = gbt.BT_wCost(self.trec[~idx_coer])
coer_vs_compl = pd.concat([emat_coerce, emat_complete],
1,
keys=['coerce', 'complete']).T
print coer_vs_compl[['TotalReturn', 'WinLoss']]
print coer_vs_compl.filter(like='count')
print coer_vs_compl.filter(like='Long')
print coer_vs_compl.filter(like='Short')
else:
print 'No coerce exit.'
def seperated_heatmap_parts(self, underly=None):
if underly is None:
underly = self.underly
for ud in underly:
tr = self.indiv_tr[ud]
btr = tr.groupby(tr.index.names[:-1]).apply(
lambda x: gbt.BT_wCost(x)[0]).unstack(-1).fillna(np.nan)
self.heatmap_parts(bt_res=btr)
def cummulative_returns(self):
# Cummulative Return
cret0 = self.trec_unstack.loc[:, 'CumRet(RReturn)'].ffill()
# Integration by refine_tradingRecord
self.trarec_refined = gbt.refine_tradingRecord(
self.trading_rec.sort_values('dt_Enter'))
emat_int, trec_int = gbt.BT_wCost(self.trarec_refined)
cret1 = pd.Series(trec_int['CumRet(RReturn)'], name='integrated')
fig, ax = plt.subplots(figsize=(9, 10))
cret0.plot.line(grid=True,
title=', '.join(self.underly),
cmap=cm.rainbow,
ax=ax)
cret1.plot.line(ls='-.', color='k', grid=True, ax=ax, legend=True)
fig.tight_layout(rect=(0, 0, 1, 1))
return fig, ax
tradingsignals = pd.DataFrame(
[openlong, openshort, selllong, buycover],
index=['openlong', 'openshort', 'selllong', 'buycover'],
columns=df.index[win:]).T
abprice = df[['ap1', 'bp1']]
bt_data = pd.concat([abprice, tradingsignals], 1).ffill().dropna()
bt_data[tradingsignals.columns] = bt_data[tradingsignals.columns].shift(2)
bt_data.dropna(inplace=True)
bt_data = bt_data.rename(columns={'ap1': 'ask', 'bp1': 'bid'})
daily_container.append([myc, bt_data])
colns, dfs = zip(*daily_container)
tradingsignals_df = pd.concat(dfs, 1, keys=colns, names=['underly', 'signals'])
tradingsignals_df = tradingsignals_df.ffill().dropna().stack('signals')
# Backtest results
trarec = gbt.generate_trading_records(tradingsignals_df)
trarec['Underlying'] = 'stock'
trarec.drop(['underly_Enter', 'underly_Exit'], axis=1, inplace=True)
cn_ord_parts = [
'dt_Enter', 'Price_Enter', 'Type_Enter', 'dt_Exit', 'Price_Exit',
'Type_Exit'
]
trarec = trarec[cn_ord_parts +
[cn for cn in trarec.columns if cn not in cn_ord_parts]]
trarec_res = gbt.BT_wCost(trarec)
tradingsignals_df = pd.concat(dfs, 1, keys=colns, names=['underly', 'signals'])
tradingsignals_df = tradingsignals_df.ffill().dropna().stack('signals')
# In[]
# Backtest results
trarec = gbt.generate_trading_records(tradingsignals_df)
trarec['Underlying'] = 'stock'
trarec.drop(['underly_Enter', 'underly_Exit'], axis=1, inplace=True)
cn_ord_parts = [
'dt_Enter', 'Price_Enter', 'Type_Enter', 'dt_Exit', 'Price_Exit',
'Type_Exit'
]
trarec = trarec[cn_ord_parts +
[cn for cn in trarec.columns if cn not in cn_ord_parts]]
emat, trec = gbt.BT_wCost(trarec)
print '\nOverview Performance: \n', emat,
print '\nTrading Logs: \n', trec
# In[]
# # Performance
# import RiskAdjustedReturnMetrics as rarm
# rr = trec['RReturn'].values
# mr = np.random.uniform(-1, 1, trec.shape[0])
# # Expected return
# e = np.mean(rr)
# # Risk free rate
# f = 0.06
# # Alpha