Abstract
Active Portfolio
Management is a simple yet powerful tool for portfolio selection which has
became a popular strategy among individual investors. Theoretical background of
active management has been established by Fundamental Law of Active Management.
The main postulate of this law suggests that in order to achieve additional
return alpha, investors should apply a model with higher forecasting power (IC)
and should apply it to a maximum number of assets (breadth). Individual
investors are attracted by this straightforward and uncomplicated logic and
that is the reason why active management is so popular strategy among them.
However, the original and later developed variants of the law lead to the same
conclusion – for better results the forecasting model must be applied to as
many as possible assets. However, this contradicts with the concentration
paradox which investors meet in their investment decisions – applying the same
forecasting model to maximum amount of assets results in involving unknown
assets with diverse characteristics in analysis. With contextual approach we
change this perspective. By applying a contextual approach to active management
investors can concentrate the forecasting models to a specific group of assets
with similar characteristics. We justify the existence of concentration paradox
and by using different fundamental contexts to differentiate stocks in
different groups, we test fundamental cross-sectional regression models as
forecasting models for active investment. With data from Taiwan Stock market we
prove that both ex-ante IC and realized return achieved with contextual models
outperform the general portfolios which follow the standard postulate of
fundamental law.
