def forward(self, tensor_x, tensor_y):
if tensor_x.dim() < tensor_y.dim():
tensor_x = tensor_x.expand_as(tensor_y)
residual = calc_residual3d(tensor_x,
tensor_y,
window=self._window,
keep_first_nan=True)
residual = residual.squeeze(-1).transpose(0, 1)
return residual
def forward(self, tensor_x, tensor_y):
if tensor_x.dim() < tensor_y.dim():
tensor_x = tensor_x.expand_as(tensor_y)
residual = calc_residual3d(tensor_x,
tensor_y,
window=self._window_train,
keep_first_nan=True)
residual = residual.squeeze(-1).transpose(0, 1)
return tsf.rolling_cumulation(data_ts=residual, window=self._window)
def forward(self, tensor_x, tensor_y):
if tensor_x.dim() < tensor_y.dim():
tensor_x = tensor_x.expand_as(tensor_y)
residual = calc_residual3d(tensor_x,
tensor_y,
window=self._window_reid,
keep_first_nan=True)
residual = residual.squeeze(-1).transpose(0, 1)
return tsf.rolling_max_drawdown_from_returns(data_ts=residual,
window=self._window)
def forward(self, tensor_x, tensor_y):
if tensor_x.dim() < tensor_y.dim():
tensor_x = tensor_x.expand_as(tensor_y)
residual = calc_residual3d(tensor_x,
tensor_y,
window_train=self._window_train,
window_test=self._window_test,
keep_first_train_nan=True)
residual = residual.squeeze(-1).transpose(0, 1)
return tsf.rolling_max(residual, self._window)
def forward(self, tensor_x, tensor_y):
if tensor_x.dim() < tensor_y.dim():
tensor_x = tensor_x.expand_as(tensor_y)
residual = calc_residual3d(tensor_x,
tensor_y,
window=self._window_train,
keep_first_nan=True)
residual = residual.squeeze(-1).transpose(0, 1)
return tsf.rolling_weighted_std(residual,
window=self._window,
halflife=self._halflife)
def forward(self, tensor_x, tensor_y):
if tensor_x.dim() < tensor_y.dim():
tensor_x = tensor_x.expand_as(tensor_y)
residual = calc_residual3d(tensor_x,
tensor_y,
window=self._window_train,
keep_first_nan=True)
residual = residual.squeeze(-1).transpose(0, 1)
varmom = tsf.rolling_mean_(
residual, window=self._window) / tsf.rolling_std_dof_0(
residual, window=self._window)
return varmom
def forward(self, size):
if size.dim() == 2:
size = size.unsqueeze(-1)
log_size = torch.log(size)
cube_log_size = torch.pow(log_size, 3)
# input order in calc_residual is x,then y
residual = calc_residual3d(log_size,
cube_log_size,
window_train=self._window_train,
window_test=self._window_test,
keep_first_train_nan=True)
return residual.squeeze(-1).transpose(0, 1)
def forward(self, tensor_x, tensor_y, benchmark=None):
import torch
if tensor_x.dim() < tensor_y.dim():
tensor_x = tensor_x.expand_as(tensor_y)
residual = calc_residual3d(tensor_x,
tensor_y,
window=self._window_train,
keep_first_nan=True)
residual = residual.squeeze(-1).transpose(0, 1)
if benchmark is None:
benchmark = torch.zeros_like(residual)
return tsf.rolling_upside_std(tensor=residual,
tensor_benchmark=benchmark,
window=self._window)
