qis.ModelLayerCumulativeAlphaAttribution

class qis.ModelLayerCumulativeAlphaAttribution(alpha_returns, cumulative_alpha, base_date, first_alpha_date, warmup_periods, freq, beta_span, beta_lag, beta_init_value, mean_adj_type)[source]

Bases: object

Post-warm-up cumulative realised alpha under point-in-time EWMA betas.

Variables:

  • alpha_returns (pandas.DataFrame) – Lagged-beta alpha returns after the warm-up base date.

  • cumulative_alpha (pandas.DataFrame) – Unannualised cumulative alpha with a zero row at the base date.

  • base_date (pandas.Timestamp) – Date on which the cumulative paths are rebased to zero.

  • first_alpha_date (pandas.Timestamp) – First return date accrued after the warm-up.

  • warmup_periods (int) – Minimum number of estimator returns through the base date.

  • freq (str) – Return frequency inherited from the underlying attribution.

  • beta_span (int) – EWMA beta span in return periods.

  • beta_lag (int) – Number of periods between beta estimation and application.

  • beta_init_value (float) – One-observation beta prior of the underlying estimator.

  • mean_adj_type (qis.models.linear.ewm.MeanAdjType) – Point-in-time mean adjustment used by the underlying estimator.

Parameters:
__init__(alpha_returns, cumulative_alpha, base_date, first_alpha_date, warmup_periods, freq, beta_span, beta_lag, beta_init_value, mean_adj_type)
Parameters:
Return type:

None

Methods

__init__(alpha_returns, cumulative_alpha, ...)

get_cumulative_alpha([is_net])

Return additive display paths, omitting identically zero realised costs.

Attributes

alpha_returns

cumulative_alpha

base_date

first_alpha_date

warmup_periods

freq

beta_span

beta_lag

beta_init_value

mean_adj_type

alpha_returns: DataFrame
cumulative_alpha: DataFrame
base_date: Timestamp
first_alpha_date: Timestamp
warmup_periods: int
freq: str
beta_span: int
beta_lag: int
beta_init_value: float
mean_adj_type: MeanAdjType
get_cumulative_alpha(is_net=True)[source]

Return additive display paths, omitting identically zero realised costs.

Parameters:

is_net (bool) – Prefer net total alpha when a net NAV was supplied.

Returns:

Total, risk, signal, integration and optional nonzero cost paths. Complete gross/net audit data remain in cumulative_alpha.

Return type:

DataFrame