for bt in scenarios:
print(bt[0])
experiment_id, backtest_id, parameters_id = bt[0]
results = bt[1]
experiment = db.load_experiment(experiment_id)
backtest_params = db.load_backtests([backtest_id])
returns = db.load_returns(experiment)
excess_returns = returns.values
train, test = backtest.split_returns(
excess_returns,
backtest_params['update_periods'],
backtest_params['trailing_periods'])
portfolios = pg.decimallistseries2nparray(results.portfolio_weights)
portfolio_returns = pg.decimallistseries2nparray(results.test_returns)
train_mean = ps.mean(portfolios[0,:], train[0])
train_variance = ps.variance(portfolios[0,:], train[0])
eta = results.eta.iloc[0]
backtest_statistics = ps.rkw_statistics(
portfolios, portfolio_returns, eta=eta)
test_statistics = backtest_statistics['statistics']
agg_returns = backtest_statistics['aggregate_returns']
output = test_statistics
output['aggregate_returns'] = pg.array2pgarray(agg_returns)
output['train_mean_return'] = train_mean
output['train_variance'] = train_variance
output['parameters_id'] = parameters_id
output['experiment_id'] = experiment_id
output['backtest_id'] = backtest_id
outputs.append(output)
returns = pd.DataFrame(return_values, index=returns_index)
# Subset based on experiment start and end dates.
returns = returns.loc[experiment['start_date']:experiment['end_date']]
# Load portfolios associated with the experiment.
cur.execute(
'''
SELECT * FROM mv_portfolios WHERE experiment_id = \'{0}\'
'''.format(experiment['id'])
)
column_names = tuple(d[0] for d in cur.description)
values = cur.fetchall()
# Convert records to data frame.
portfolios = pd.DataFrame(values)
portfolios.columns = column_names
portfolio_values = np.array(portfolios['portfolio_weights'].apply(lambda x: list(map(float, x))).values.tolist())
portfolios_index = portfolios['id']
portfolios = pd.DataFrame(portfolio_values, index=portfolios_index)
# Compute statistics for all portfolios.
excess_returns = returns.values[:, experiment['asset_columns']]
mean_returns = portfolios.apply(lambda x: ps.mean(x, excess_returns), axis=1)
variances = portfolios.apply(lambda x: ps.variance(x, excess_returns), axis=1)