def optimise_using_correlation(mean_list: list,
avg_correlation: float,
std: float):
corr_matrix = boring_corr_matrix_values(len(mean_list), offdiag=avg_correlation)
stdev_list = np.full(len(mean_list), std)
sigma = sigma_from_corr_and_std(stdev_list, corr_matrix)
weights = optimise_from_sigma_and_mean_list(sigma, mean_list)
return weights
def covariance_from_stdev_and_correlation(
correlation_estimate: correlationEstimate,
stdev_estimate: stdevEstimates) -> covarianceEstimate:
list_of_assets = list(correlation_estimate.columns)
aligned_stdev_list = stdev_estimate.list_in_key_order(list_of_assets)
sigma = sigma_from_corr_and_std(aligned_stdev_list,
correlation_estimate.values)
return covarianceEstimate(sigma, list_of_assets)
def get_annualised_risk_given_inputs(std_dev, cmatrix, weights):
weights = np.array(weights)
std_dev = np.array(std_dev)
std_dev, cmatrix, weights = clean_values(std_dev, cmatrix, weights)
sigma = sigma_from_corr_and_std(std_dev, cmatrix)
portfolio_variance = weights.dot(sigma).dot(weights.transpose())
portfolio_std = portfolio_variance**.5
return portfolio_std
def get_annualised_risk(
std_dev: stdevEstimates, cmatrix: correlationEstimate, weights: portfolioWeights
) -> float:
weights_as_np = weights.as_np()
std_dev_as_np = std_dev.as_np()
cmatrix_as_np = cmatrix.as_np()
std_dev_as_np, cmatrix_as_np, weights_as_np = clean_values(
std_dev_as_np, cmatrix_as_np, weights_as_np
)
sigma = sigma_from_corr_and_std(std_dev_as_np, cmatrix_as_np)
portfolio_variance = weights_as_np.dot(sigma).dot(weights_as_np.transpose())
portfolio_std = portfolio_variance ** 0.5
return portfolio_std
def combine_list_of_correlations_and_stdev(
rolling_correlations, rolling_stdev):
corr_index_dates = rolling_correlations.fit_dates
std_index_dates = list(rolling_stdev.index)
sigma_list = []
for corr_index, index_date in enumerate(corr_index_dates):
std_dev = rolling_stdev.loc[index_date].values
std_dev[np.isnan(std_dev)] = 0.0
cmatrix = rolling_correlations.corr_list[corr_index]
cmatrix = pd.DataFrame(cmatrix)
cmatrix[cmatrix.isna()] = 1.0
cmatrix = np.array(cmatrix)
sigma = sigma_from_corr_and_std(std_dev, cmatrix)
sigma_list.append(sigma)
covariance_estimates = CorrelationList(
sigma_list, rolling_stdev.columns, corr_index_dates
)
return covariance_estimates
def covariance_from_stdev_and_correlation(
correlation_estimate: correlationEstimate, stdev_estimate: stdevEstimates
) -> covarianceEstimate:
all_assets = set(list(correlation_estimate.columns) + stdev_estimate.list_of_keys())
list_of_assets_with_data = list(
set(correlation_estimate.assets_with_data()).intersection(
set(stdev_estimate.assets_with_data())
)
)
assets_without_data = list(all_assets.difference(list_of_assets_with_data))
aligned_stdev_list = stdev_estimate.list_in_key_order(list_of_assets_with_data)
aligned_corr_list = correlation_estimate.subset(list_of_assets_with_data)
sigma = sigma_from_corr_and_std(aligned_stdev_list, aligned_corr_list.values)
cov_assets_with_data = covarianceEstimate(sigma, columns=list_of_assets_with_data)
cov = cov_assets_with_data.add_assets_with_nan_values(assets_without_data)
return cov