def __init__(self):
dataset = data.sp500_returns(num_points=100)
super(StochasticVolatilitySP500Small, self).__init__(
name='stochastic_volatility_sp500_small',
pretty_name=
'Smaller stochastic volatility model of S&P500 returns.',
**dataset)
def __init__(self, use_markov_chain=False):
dataset = data.sp500_returns()
super(StochasticVolatilitySP500, self).__init__(
name='stochastic_volatility_sp500',
pretty_name='Stochastic volatility model of S&P500 returns.',
use_markov_chain=use_markov_chain,
**dataset)
def testStochasticVolatilityModelSP500(self):
num_train_points = 2516
dataset = data.sp500_returns()
self.assertEqual((num_train_points, ),
dataset['centered_returns'].shape)
self.assertAllClose(0.0,
np.mean(dataset['centered_returns']),
atol=1e-5)
def __init__(
self,
centered=False,
use_fft=True,
):
dataset = data.sp500_returns(num_points=100)
super(VectorizedStochasticVolatilitySP500Small, self).__init__(
name='vectorized_stochasticic_volatility_sp500_small',
pretty_name='Smaller stochastic volatility model of S&P500 returns.',
**dataset)
def __init__(
self,
centered=False,
use_fft=True,
):
dataset = data.sp500_returns()
super(VectorizedStochasticVolatilitySP500, self).__init__(
name='vectorized_stochastic_volatility_sp500',
pretty_name='Stochastic volatility model of S&P500 returns.',
**dataset)
def stochastic_volatility_sp500_small():
"""Stochastic volatility model.
This is a smaller version of `stochastic_volatility_model_sp500` using only
100 days of returns from the S&P 500, ending 6/24/2020.
Returns:
target: StanModel.
"""
dataset = data.sp500_returns(num_points=100)
return stochastic_volatility.stochastic_volatility(**dataset)
def stochastic_volatility_sp500():
"""Stochastic volatility model.
This uses a dataset of 2517 daily returns of the S&P 500 index,
representing the time period 6/25/2010-6/24/2020.
Returns:
target: StanModel.
"""
dataset = data.sp500_returns()
return stochastic_volatility.stochastic_volatility(**dataset)
def testMarkovChainLogprobMatchesOriginal(self):
# Use a very short dataset since the non-markov-chain model is slow.
dataset = data.sp500_returns(num_points=10)
model = stochastic_volatility.StochasticVolatility(
use_markov_chain=False, **dataset)
markov_chain_model = stochastic_volatility.StochasticVolatility(
use_markov_chain=True, **dataset)
xs = self.evaluate(model.prior_distribution().sample(
10, seed=tfp_test_util.test_seed()))
xs_stacked = xs[:-1] + (tf.stack(xs[-1], axis=-1), )
self.assertAllClose(
model.unnormalized_log_prob(xs),
markov_chain_model.unnormalized_log_prob(xs_stacked),
atol=1e-2)