def __init__(self, input_lhs, input_rhs, func, name, length=1, input_key='close', desc="Pairwise"):
super(Pairwise, self).__init__(name, [input_lhs, input_rhs], input_key, length=length, desc=desc)
self.lhs_name = PipeLine.get_input_name(input_lhs)
self.rhs_name = PipeLine.get_input_name(input_rhs)
self.func = func
self.is_input_pipeline = False
if isinstance(input_lhs, PipeLine) and not isinstance(input_rhs, PipeLine):
raise TypeError("input_lhs has to be the same type as input_rhs as Pipeline")
if isinstance(input_lhs, Indicator) and not isinstance(input_rhs, Indicator):
raise TypeError("input_lhs has to be the same type as input_rhs as Indicator")
if isinstance(input_lhs, DataSeries) and not isinstance(input_rhs, DataSeries):
raise TypeError("input_lhs has to be the same type as input_rhs as DataSeries")
if isinstance(input_lhs, PipeLine):
self.is_input_pipeline = True
try:
np.testing.assert_almost_equal(input_lhs.shape(), input_rhs.shape(), 10)
self.__shape = input_lhs.shape()
except AssertionError as e:
raise ValueError("input_lhs shape should be the same as input_rhs in Pairwise Pipeline operation!")
else:
self.__shape = np.array([1, 1])
super(Pairwise, self).update_all()
def __init__(self, input_lhs, input_rhs, func, name, length=1, input_key='close', desc="Pairwise"):
super(Pairwise, self).__init__(name, [input_lhs, input_rhs], input_key, length=length, desc=desc)
self.lhs_name = PipeLine.get_input_name(input_lhs)
self.rhs_name = PipeLine.get_input_name(input_rhs)
self.func = func
self.is_input_pipeline = False
if isinstance(input_lhs, PipeLine) and not isinstance(input_rhs, PipeLine):
raise TypeError("input_lhs has to be the same type as input_rhs as Pipeline")
if isinstance(input_lhs, Indicator) and not isinstance(input_rhs, Indicator):
raise TypeError("input_lhs has to be the same type as input_rhs as Indicator")
if isinstance(input_lhs, DataSeries) and not isinstance(input_rhs, DataSeries):
raise TypeError("input_lhs has to be the same type as input_rhs as DataSeries")
if isinstance(input_lhs, PipeLine):
self.is_input_pipeline = True
try:
np.testing.assert_almost_equal(input_lhs.shape(), input_rhs.shape(), 10)
self.__shape = input_lhs.shape()
except AssertionError as e:
raise ValueError("input_lhs shape should be the same as input_rhs in Pairwise Pipeline operation!")
else:
self.__shape = [1, 1]
super(Pairwise, self).update_all()
def __init__(self, inputs, lb, ub, newval, input_key="close", desc="Cross Sessional Tail"):
super(Tail, self).__init__(
inputs=inputs,
np_func=lambda x: np_assign_on_mask(x, lb, ub, newval),
name=PipeLine.get_name(Tail.__name__, inputs, input_key),
input_key=input_key,
length=1,
desc=desc,
)
def __init__(self, inputs, input_key="close", desc="Cross Sessional Average"):
super(Average, self).__init__(
inputs=inputs,
np_func=np.average,
name=PipeLine.get_name(Average.__name__, inputs, input_key),
input_key=input_key,
length=1,
desc=desc,
)
def __init__(self, inputs, input_key="close", desc="Cross Sessional Sum"):
super(Sum, self).__init__(
inputs=inputs,
np_func=np.sum,
name=PipeLine.get_name(Sum.__name__, inputs, input_key),
input_key=input_key,
length=1,
desc=desc,
)
def __init__(self, inputs, input_key='close', desc="Cross Sessional Log"):
super(Log,
self).__init__(inputs=inputs,
np_func=lambda x: np.log(x),
name=PipeLine.get_name(Log.__name__, inputs,
input_key),
input_key=input_key,
length=1,
desc=desc)
def __init__(self, inputs, input_key='close', desc="Cross Sessional Sign"):
super(Sign,
self).__init__(inputs=inputs,
np_func=lambda x: np_sign_to_value(x),
name=PipeLine.get_name(Sign.__name__, inputs,
input_key),
input_key=input_key,
length=1,
desc=desc)
def __init__(self, inputs, length, input_key="close", desc="Cross Sessional Delta"):
super(Delta, self).__init__(
inputs=inputs,
np_func=lambda x: x[-1, :] - x[0, :],
name=PipeLine.get_name(Delta.__name__, inputs, input_key),
input_key=input_key,
length=length,
desc=desc,
)
def __init__(self, inputs, length, input_key="close", desc="Cross Sessional Product"):
super(Product, self).__init__(
inputs=inputs,
np_func=lambda x: np.prod(x, axis=0),
name=PipeLine.get_name(Product.__name__, inputs, input_key),
input_key=input_key,
length=length,
desc=desc,
)
def __init__(self, inputs, input_key="close", desc="Cross Sessional Log"):
super(Log, self).__init__(
inputs=inputs,
np_func=lambda x: np.log(x),
name=PipeLine.get_name(Log.__name__, inputs, input_key),
input_key=input_key,
length=1,
desc=desc,
)
def __init__(self, inputs, input_key="close", desc="Cross Sessional Scale"):
super(Scale, self).__init__(
inputs=inputs,
np_func=lambda x: x / np.sum(np.abs(x)),
name=PipeLine.get_name(Scale.__name__, inputs, input_key),
input_key=input_key,
length=1,
desc=desc,
)
def __init__(self, inputs, input_key='close', desc="Cross Sessional Abs"):
super(Abs,
self).__init__(inputs=inputs,
np_func=np.abs,
name=PipeLine.get_name(Abs.__name__, inputs,
input_key),
input_key=input_key,
length=1,
desc=desc)
def __init__(self, inputs, length=20, input_key="close", desc="Cross Sessional DecayLinear"):
super(DecayLinear, self).__init__(
inputs=inputs,
np_func=lambda x: np.dot(np.arange(length, 0, -1), x) / np.sum(np.arange(length, 0, -1)),
name=PipeLine.get_name(DecayLinear.__name__, inputs, input_key),
input_key=input_key,
length=length,
desc=desc,
)
def __init__(self, inputs, f=0.9, length=20, input_key="close", desc="Cross Sessional DecayExp"):
super(DecayExp, self).__init__(
inputs=inputs,
np_func=lambda x: np.dot(np.power(f, np.arange(length)), x) / np.sum(np.power(f, np.arange(length))),
name=PipeLine.get_name(DecayExp.__name__, inputs, input_key),
input_key=input_key,
length=length,
desc=desc,
)
def __init__(self, inputs, ascending=True, length=20, input_key="close", desc="Cross Sessional Timeseries Rank"):
super(TsRank, self).__init__(
inputs=inputs,
np_func=lambda x: timeseries_rank_helper(x, ascending=ascending),
name=PipeLine.get_name(TsRank.__name__, inputs, input_key),
input_key=input_key,
length=length,
desc=desc,
)
def __init__(self, inputs, input_key="close", desc="Cross Sessional SignPower"):
super(SignPower, self).__init__(
inputs=inputs,
np_func=lambda x: np_sign_to_value(x) * np.power(x, e),
name=PipeLine.get_name(SignPower.__name__, inputs, input_key),
input_key=input_key,
length=1,
desc=desc,
)
def __init__(self,
inputs,
input_key='close',
desc="Cross Sessional Scale"):
super(Scale,
self).__init__(inputs=inputs,
np_func=lambda x: x / np.sum(np.abs(x)),
name=PipeLine.get_name(Scale.__name__, inputs,
input_key),
input_key=input_key,
length=1,
desc=desc)
def __init__(self,
inputs=None,
ascending=True,
input_key='close',
desc="Rank"):
self.ascending = ascending
super(Rank, self).__init__(PipeLine.get_name(Rank.__name__, inputs,
input_key),
inputs,
input_key,
length=1,
desc=desc)
def __init__(self,
inputs,
input_key='close',
desc="Bundle and Sync DataSeries to Vector"):
super(MakeVector,
self).__init__(PipeLine.get_name(MakeVector.__name__, inputs,
input_key),
inputs,
input_key,
length=1,
desc=desc)
super(MakeVector, self).update_all()
def __init__(self,
inputs,
ascending=True,
length=20,
input_key='close',
desc="Cross Sessional Timeseries Rank"):
super(TsRank, self).__init__(
inputs=inputs,
np_func=lambda x: timeseries_rank_helper(x, ascending=ascending),
name=PipeLine.get_name(TsRank.__name__, inputs, input_key),
input_key=input_key,
length=length,
desc=desc)
def __init__(self,
input_lhs,
input_rhs,
input_key='close',
desc="Pairwise Max"):
super(Max, self).__init__(
input_lhs,
input_rhs,
func=lambda x, y: np.max(np.vstack([x, y]), axis=0),
name=PipeLine.get_name(Max.__name__, [input_lhs, input_rhs],
input_key),
input_key=input_key,
desc=desc)
def __init__(self,
inputs,
length=20,
input_key='close',
desc="Cross Sessional DecayLinear"):
super(DecayLinear, self).__init__(
inputs=inputs,
np_func=lambda x: np.dot(np.arange(length, 0, -1), x) / np.sum(
np.arange(length, 0, -1)),
name=PipeLine.get_name(DecayLinear.__name__, inputs, input_key),
input_key=input_key,
length=length,
desc=desc)
def __init__(self,
inputs,
length,
input_key='close',
desc="Cross Sessional Product"):
super(Product,
self).__init__(inputs=inputs,
np_func=lambda x: np.prod(x, axis=0),
name=PipeLine.get_name(Product.__name__, inputs,
input_key),
input_key=input_key,
length=length,
desc=desc)
def __init__(self,
inputs,
length,
input_key='close',
desc="Cross Sessional Delta"):
super(Delta,
self).__init__(inputs=inputs,
np_func=lambda x: x[-1, :] - x[0, :],
name=PipeLine.get_name(Delta.__name__, inputs,
input_key),
input_key=input_key,
length=length,
desc=desc)
def __init__(self,
inputs,
f=0.9,
length=20,
input_key='close',
desc="Cross Sessional DecayExp"):
super(DecayExp, self).__init__(
inputs=inputs,
np_func=lambda x: np.dot(np.power(f, np.arange(
length)), x) / np.sum(np.power(f, np.arange(length))),
name=PipeLine.get_name(DecayExp.__name__, inputs, input_key),
input_key=input_key,
length=length,
desc=desc)
def __init__(self,
input_lhs,
input_rhs,
input_key='close',
desc="Pairwise Divdes"):
super(Divides,
self).__init__(input_lhs,
input_rhs,
func=lambda x, y: x / y,
name=PipeLine.get_name(Divides.__name__,
[input_lhs, input_rhs],
input_key),
input_key=input_key,
desc=desc)
def __init__(self,
input_lhs,
input_rhs,
input_key='close',
desc="Pairwise Greater"):
super(Greater,
self).__init__(input_lhs,
input_rhs,
func=lambda x, y: x > y,
name=PipeLine.get_name(Greater.__name__,
[input_lhs, input_rhs],
input_key),
input_key=input_key,
desc=desc)
def __init__(self,
input_lhs,
input_rhs,
input_key='close',
desc="Pairwise NotEquals"):
super(NotEquals,
self).__init__(input_lhs,
input_rhs,
func=lambda x, y: x != y,
name=PipeLine.get_name(NotEquals.__name__,
[input_lhs, input_rhs],
input_key),
input_key=input_key,
desc=desc)
def __init__(self,
inputs,
lb,
ub,
newval,
input_key='close',
desc="Cross Sessional Tail"):
super(Tail, self).__init__(
inputs=inputs,
np_func=lambda x: np_assign_on_mask(x, lb, ub, newval),
name=PipeLine.get_name(Tail.__name__, inputs, input_key),
input_key=input_key,
length=1,
desc=desc)
def __init__(self,
input_lhs,
input_rhs,
length,
input_key='close',
desc="Pairwise PairCorrelation"):
super(PairCorrelation, self).__init__(
input_lhs,
input_rhs,
length=length,
func=lambda x, y: np.corrcoef(np.transpose(x), np.transpose(y)),
name=PipeLine.get_name(PairCorrelation.__name__,
[input_lhs, input_rhs], input_key),
input_key=input_key,
desc=desc)
def __init__(self, inputs, ascending=True, input_key='close', desc="Rank"):
self.ascending = ascending
super(Rank, self).__init__(PipeLine.get_name(Rank.__name__, inputs, input_key),
inputs, input_key, length=1, desc=desc)
def __init__(self, inputs, input_key='close', length=30, desc="Correlation"):
super(Corr, self).__init__(PipeLine.get_name(Corr.__name__, input),
input, input_key, length, desc)
super(Corr, self).update_all()
def __init__(self, input_lhs, input_rhs, input_key='close', desc="Pairwise Max"):
super(Max, self).__init__(input_lhs, input_rhs, func=lambda x, y: np.max(np.vstack([x, y]), axis=0),
name=PipeLine.get_name(Max.__name__, [input_lhs, input_rhs], input_key),
input_key=input_key, desc=desc)
def __init__(self, input_lhs, input_rhs, length, input_key='close', desc="Pairwise PairCorrelation"):
super(PairCorrelation, self).__init__(input_lhs, input_rhs, length=length,
func=lambda x, y: np.corrcoef(np.transpose(x), np.transpose(y)),
name=PipeLine.get_name(PairCorrelation.__name__, [input_lhs, input_rhs], input_key),
input_key=input_key, desc=desc)
def get_name(indicator_name, input_lhs, input_rhs, input_key, *args):
return PipeLine.get_name(indicator_name,
inputs=[input_lhs, input_rhs],
input_key=input_key, *args)
def __init__(self, input_lhs, input_rhs, input_key='close', desc="Pairwise NotEquals"):
super(NotEquals, self).__init__(input_lhs, input_rhs, func=lambda x, y: x != y,
name=PipeLine.get_name(NotEquals.__name__, [input_lhs, input_rhs], input_key),
input_key=input_key, desc=desc)
def __init__(self, input_lhs, input_rhs, input_key='close', desc="Pairwise Divdes"):
super(Divides, self).__init__(input_lhs, input_rhs, func=lambda x, y: x/y,
name=PipeLine.get_name(Divides.__name__, [input_lhs, input_rhs], input_key),
input_key=input_key, desc=desc)
def __init__(self, input_lhs, input_rhs, input_key='close', desc="Pairwise Greater"):
super(Greater, self).__init__(input_lhs, input_rhs, func=lambda x, y: x > y,
name=PipeLine.get_name(Greater.__name__, [input_lhs, input_rhs], input_key),
input_key=input_key, desc=desc)
def __init__(self, inputs, input_key='close', length=30, desc="Correlation"):
super(Corr, self).__init__(PipeLine.get_name(Corr.__name__, input),
input, input_key, length, desc)
super(Corr, self).update_all()
def get_name(indicator_name, input_lhs, input_rhs, input_key, *args):
return PipeLine.get_name(indicator_name,
inputs=[input_lhs, input_rhs],
input_key=input_key,
*args)
def __init__(self, inputs, input_key='close', desc="Cross Sessional Sign"):
super(Sign, self).__init__(inputs=inputs, np_func=lambda x: np_sign_to_value(x),
name=PipeLine.get_name(Sign.__name__, inputs, input_key),
input_key=input_key, length=1, desc=desc)
def __init__(self, inputs, input_key='close', desc="Cross Sessional Abs"):
super(Abs, self).__init__(inputs=inputs, np_func=np.abs,
name=PipeLine.get_name(Abs.__name__, inputs, input_key),
input_key=input_key, length=1, desc=desc)
def __init__(self, inputs, input_key='close', desc="Bundle and Sync DataSeries to Vector"):
super(MakeVector, self).__init__(PipeLine.get_name(MakeVector.__name__, inputs, input_key),
inputs, input_key, length=1, desc=desc)
super(MakeVector, self).update_all()