Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Improving Portfolio Performance

Linde, Ebba and Rodling, Nina (2014) FMS820 20141
Mathematical Statistics
Abstract (Swedish)
The objective of this thesis is to investigate performance for dierent investment alternatives
for an investor wanting to track a multidimensional stock index. Performance is
measured in terms of transaction cost, active return against the index and tracking error.
The problem is approached by comparing performance for a full replication strategy against
a strategy in which the traded portfolio is a dimension reduction of the index as well as
against a strategy, trading the dimension reduced portfolio, aiming to predict and in turn
minimize transaction costs. The full replication case and the dimension reduction case
trade with a volume-weighted strategy, whilst the last strategy trades at times historically
being least expensive... (More)
The objective of this thesis is to investigate performance for dierent investment alternatives
for an investor wanting to track a multidimensional stock index. Performance is
measured in terms of transaction cost, active return against the index and tracking error.
The problem is approached by comparing performance for a full replication strategy against
a strategy in which the traded portfolio is a dimension reduction of the index as well as
against a strategy, trading the dimension reduced portfolio, aiming to predict and in turn
minimize transaction costs. The full replication case and the dimension reduction case
trade with a volume-weighted strategy, whilst the last strategy trades at times historically
being least expensive to trade at. The dimension reduction is done based on results from
a principal component analysis together with empiric results on transaction costs associated
with trading a certain stock. The transaction cost prediction model implemented is
the PAR-model, presented by Rashkovich and Verma (2012). The results show that when
reducing the dimension of the index, meaning that stocks with undesired characteristics
can be excluded, performance is improved. The transaction cost minimizing strategy show
some improvement against the full replication strategy, but its performance is inferior to
trading a dimension reduced portfolio with a volume-weighted strategy. This highlights
the diculties in predicting stock market behavior. Hence, the strategy recommended for
an investor wanting to track a multidimensional index is to conduct a dimension reduction
according to preferences and use a volume-weighted trading strategy. (Less)
Please use this url to cite or link to this publication:
author
Linde, Ebba and Rodling, Nina
supervisor
organization
course
FMS820 20141
year
type
H2 - Master's Degree (Two Years)
subject
language
English
id
4451458
date added to LUP
2014-05-28 11:38:09
date last changed
2014-05-28 11:38:09
@misc{4451458,
  abstract     = {{The objective of this thesis is to investigate performance for dierent investment alternatives
for an investor wanting to track a multidimensional stock index. Performance is
measured in terms of transaction cost, active return against the index and tracking error.
The problem is approached by comparing performance for a full replication strategy against
a strategy in which the traded portfolio is a dimension reduction of the index as well as
against a strategy, trading the dimension reduced portfolio, aiming to predict and in turn
minimize transaction costs. The full replication case and the dimension reduction case
trade with a volume-weighted strategy, whilst the last strategy trades at times historically
being least expensive to trade at. The dimension reduction is done based on results from
a principal component analysis together with empiric results on transaction costs associated
with trading a certain stock. The transaction cost prediction model implemented is
the PAR-model, presented by Rashkovich and Verma (2012). The results show that when
reducing the dimension of the index, meaning that stocks with undesired characteristics
can be excluded, performance is improved. The transaction cost minimizing strategy show
some improvement against the full replication strategy, but its performance is inferior to
trading a dimension reduced portfolio with a volume-weighted strategy. This highlights
the diculties in predicting stock market behavior. Hence, the strategy recommended for
an investor wanting to track a multidimensional index is to conduct a dimension reduction
according to preferences and use a volume-weighted trading strategy.}},
  author       = {{Linde, Ebba and Rodling, Nina}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Improving Portfolio Performance}},
  year         = {{2014}},
}