def test_sharpe_iid():
data = np.array([
0.259, .198, .364, -.081, .057, .055, .188, .317, .24, .184, -.01, .526
])
# numpy array
sharpe = perf.sharpe_iid(data, bench=.05, factor=1, log=True)
assert (np.isclose(sharpe, .834364))
sharpe = perf.sharpe_iid(data, bench=.05, factor=1, log=False)
assert (np.isclose(sharpe, .834364))
# below is for computing sharpe ratio with pct returns
# assert(np.isclose(sharpe, 0.8189144744629443))
# turn data to pandas.Series
data = pd.Series(data)
sharpe = perf.sharpe_iid(data, bench=.05, factor=1, log=True)
assert (np.isclose(sharpe, .834364))
sharpe = perf.sharpe_iid(data, bench=.05, factor=252, log=True)
assert (np.isclose(sharpe, .834364 * np.sqrt(252)))
sharpe = perf.sharpe_iid(data, bench=.05, factor=1, log=False)
assert (np.isclose(sharpe, .834364))
def test_sharpe_iid():
data = np.array([
0.259, .198, .364, -.081, .057, .055, .188, .317, .24, .184, -.01, .526
])
# numpy array
sharpe = perf.sharpe_iid(data, bench=.05, factor=1, return_type='log')
assert (np.isclose(sharpe, .834364))
sharpe = perf.sharpe_iid(data, bench=.05, factor=1, return_type='pct')
assert (np.isclose(sharpe, 0.8189144744629443))
# turn data to pandas.Series
data = pd.Series(data)
sharpe = perf.sharpe_iid(data, bench=.05, factor=1, return_type='log')
assert (np.isclose(sharpe, .834364))
sharpe = perf.sharpe_iid(data, bench=.05, factor=252, return_type='log')
assert (np.isclose(sharpe, .834364 * np.sqrt(252)))
sharpe = perf.sharpe_iid(data, bench=.05, factor=1, return_type='pct')
assert (np.isclose(sharpe, 0.8189144744629443))
def dsr_from_returns(test_sharpe, returns_df, risk_free=0):
'''
Calculate DSR based on a set of given returns_df.
Parameters:
test_sharpe : Reported sharpe, to be tested.
returns_df : Log return series
risk_free : Risk free return, default 0.
Returns:
DSR statistic
'''
T, N = returns_df.shape
sharpe = perf.sharpe_iid(returns_df, bench=risk_free, factor=1)
sharpe_std = np.std(sharpe, ddof=1)
skew = returns_df.skew()
kurtosis = returns_df.kurtosis() + 3
dsr = dsr(test_sharpe,
sharpe_std=sharpe_std,
N=N,
T=T,
skew=skew,
kurtosis=kurtosis)
return dsr
def psr_from_returns(returns, risk_free=0, target_sharpe=0):
"""
PSR from return series.
Parameters:
returns:
return series
risk_free:
risk free or benchmark rate for sharpe ratio calculation,
default 0.
target_sharpe:
minimum sharpe ratio
Returns:
PSR probabilities.
"""
T = len(returns)
sharpe = perf.sharpe_iid(returns, bench=risk_free, factor=1)
skew = returns.skew()
kurtosis = returns.kurtosis() + 3
return psr(
sharpe=sharpe,
T=T,
skew=skew,
kurtosis=kurtosis,
target_sharpe=target_sharpe,
)
def metric(x):
return np.sqrt(255) * perf.sharpe_iid(x)
