def test_bias(self):
""" Test forward bias of algo. Test on a portion of given data set, then add several
data points and see if weights has changed. """
S = self.random_portfolio
k = 10
B1 = CRP().run(S.iloc[:-k]).B
B2 = CRP().run(S).B
self.assertTrue((B1 == B2.iloc[:-k]).all().all())
def test_equity(self):
""" Make sure that equity of a portfolio [1,0,...,0] with NaN values
is the same as asset itself. """
S = self.random_portfolio
b = [1.] + [0.] * (len(S.columns) - 1)
result = CRP(b).run(S)
self.assertAlmostEqual(result.total_wealth, S[S.columns[0]].iget(-1),
10)
def hedge(self, result=None):
""" Hedge results with results of other strategy (subtract weights).
:param result: Other result object. Default is UCRP.
:return: New AlgoResult object.
"""
if result is None:
from algos import CRP
result = CRP().run(self.X.cumprod())
return AlgoResult(self.X, self.B - result.B)
def plot(self,
ucrp=False,
bah=False,
residual=False,
assets=False,
**kwargs):
""" Plot strategy equity.
:param ucrp: Add uniform CRP as a benchmark.
:param bah: Add Buy-And-Hold portfolio as a benchmark.
:param residual: Add portfolio minus UCRP as a benchmark.
:param assets: Add asset prices.
:param kwargs: Additional arguments for pd.DataFrame.plot
"""
# NOTE: order of plotting is important because of coloring
# plot portfolio
d = self.to_dataframe()
D = d.copy()
# add individual assets
if isinstance(assets, bool):
if assets:
assets = self[0].asset_equity.columns
else:
assets = []
if list(assets):
D = D.join(self[0].asset_equity)
ax = D.plot(color=_colors_hash(D.columns), **kwargs)
kwargs['ax'] = ax
ax.set_ylabel('Total wealth')
# plot residual strategy
if residual:
d['RESIDUAL'] = self[0].residual_r.cumprod()
d[['RESIDUAL']].plot(**kwargs)
# plot uniform constant rebalanced portfolio
if ucrp:
from .algos import CRP
crp_algo = CRP().run(self[0].X.cumprod())
crp_algo.fee = self[0].fee
d['UCRP'] = crp_algo.equity
d[['UCRP']].plot(**kwargs)
# add bah
if bah:
from .algos import BAH
bah_algo = BAH().run(self[0].X.cumprod())
bah_algo.fee = self[0].fee
d['BAH'] = bah_algo.equity
d[['BAH']].plot(**kwargs)
return ax
def plot(self, ucrp=False, bah=False, assets=False, **kwargs):
""" Plot strategy equity.
:param ucrp: Add uniform CRP as a benchmark.
:param bah: Add Buy-And-Hold portfolio as a benchmark.
:param assets: Add asset prices.
:param kwargs: Additional arguments for pd.DataFrame.plot
"""
# NOTE: order of plotting is important because of coloring
# plot portfolio
d = self.to_dataframe()
portfolio = d.copy()
ax = portfolio.plot(linewidth=3., legend=False, **kwargs)
kwargs['ax'] = ax
ax.set_ylabel('Total wealth')
# plot uniform constant rebalanced portfolio
if ucrp:
from universal.algos import CRP
crp_algo = CRP().run(self[0].X.cumprod())
crp_algo.fee = self[0].fee
d['UCRP'] = crp_algo.equity
d[['UCRP']].plot(**kwargs)
# add bah
if bah:
from universal.algos import BAH
bah_algo = BAH().run(self[0].X.cumprod())
bah_algo.fee = self[0].fee
d['BAH'] = bah_algo.equity
d[['BAH']].plot(**kwargs)
# add individual assets
if isinstance(assets, bool):
if assets:
assets = self[0].asset_equity.columns
else:
assets = []
if list(assets):
self[0].asset_equity.sort_index(axis=1).plot(
color=_colors(len(assets) + 1), **kwargs)
# plot portfolio again to highlight it
kwargs['color'] = 'blue'
portfolio.plot(linewidth=3., **kwargs)
return ax
def ucrp_sharpe(self):
from universal.algos import CRP
result = CRP().run(self.X.cumprod())
return result.sharpe
def ucrp_sharpe(self):
from .algos import CRP
result = CRP().run(self.X.cumprod())
result.set_rf_rate(self.rf_rate)
return result.sharpe