def _get_sharpe_ratio(price_series: NDFrame,
rf_series: Series,
scale_to_annualise: bool):
""""
Returns the Sharpe ratio based on a series of asset prices
and risk-free asset prices. The calculation is based on the
arithmetic mean of actual returns, as appears to be standard
:param price_series - a pandas series or data frame of prices
:param rf_series - a pandas series representing the total
return/price series of the risk free rate
:param scale_to_annualise - bool governing whether or not a
scaling factor is applied to annualise the statistic
Notes
-------
https://en.wikipedia.org/wiki/Sharpe_ratio
"""
# excess_return = get_annualised_excess_return(price_series, rf_series)
# vol = get_annual_vol(price_series)
# sharpe_ratio = excess_return/vol
returns = (price_series / price_series.shift(1)).dropna()
returns_rf = (rf_series / rf_series.shift(1)).dropna()
rel_returns = returns.subtract(returns_rf, axis=0)
avg_excess = rel_returns.mean()
vol = rel_returns.std()
annualising_scaling = TSeriesHelper._get_annualisation_factor(price_series.index) if scale_to_annualise else 1
sharpe_ratio = avg_excess * annualising_scaling / vol
return sharpe_ratio
def get_annualised_vol(price_series: NDFrame):
""""
Returns the annualised volatility of returns based on a stream of asset prices. The function derives
the time period of the price data uses this to calculate a suitable factor to annualise the data with.
Note that the function calculates the standard deviation of the NATURAL LOG of the returns, as is conventional.
:param price_series - a pandas series or dataframe of prices
Notes
--------
https://en.wikipedia.org/wiki/Volatility_(finance)
"""
log_returns = np.log(price_series / price_series.shift(1))
annualising_scaling = TSeriesHelper._get_annualisation_factor(price_series.index)
return np.std(log_returns) * annualising_scaling
def price_to_return(price_series: NDFrame):
"""
convert a series of asset prices to a series
or returns
:param price_series: pandas price series or dataframe
of price series
:return:
"""
if (price_series == 0).any(axis=None).any():
raise ValueError("Cannot convert price series with zeroes to a return")
price_series = price_series / price_series.shift(1)
price_series.dropna(inplace=True)
price_series = price_series - 1
return price_series
def _get_sortino_ratio(price_series: NDFrame,
benchmark_series: Series,
scale_to_annualise: bool):
""""
Returns the Sortino ratio
:param price_series - a pandas series or data frame of prices
:param benchmark_series - a pandas series representing the total
return/price series of the chosen benchmark
Notes
------
https://en.wikipedia.org/wiki/Sortino_ratio
"""
returns = (price_series / price_series.shift(1)).dropna()
returns_rf = (benchmark_series / benchmark_series.shift(1)).dropna()
avg_excess = (returns.subtract(returns_rf, axis=0)).mean()
vol = TSeriesHelper._get_downside_deviation(returns, returns_rf)
annualising_scaling = TSeriesHelper._get_annualisation_factor(price_series.index) if scale_to_annualise else 1
sortino_ratio = avg_excess * annualising_scaling / vol
return sortino_ratio
