Investment fund performance validation - Vine copulae estimation using a minimum spanning tree

Catching the dependencies between financial time series is a complex exer-
cise with a lot of challenges, both of theoretical and practical nature. We develop a metho-
dology to model portfolio dynamics using the minimum spanning tree methods combined
with econometric models which solves a good part of these challenges. We use a tracking
error that is equivalent to the Euclidean distance, to cluster the closest market indices.
Financial risk is difficult to manage, because risk is evolving constantly and depends
on very different factors like volatility, liquidity, asset class etc. To capture this evolution
we develop a recursive portfolio validation method that reveals the true nature of the
evolution of the risk structure of financial portfolios. We investigate several portfolio
strategies to understand the dynamics behind the holdings. We thus validate the results
of our portfolio analysis.
We use a vine copula construction, which allows us to separate the marginal estimation
from the dependence estimation and calibrate the underlying dynamics very precisely. The
minimum spanning tree method helps us to create a robust tree foundation to support
the entire vine structure. Starting from a portfolio analysis, we are thus able to validate
the portfolio valuation over a specific period of time.