Visualization of changes in correlations of stock returns during and after financial crisis
|Author:||Bc. David Zbožínek|
|Year:||2017 - summer|
|Leaders:|| prof. PhDr. Ladislav Krištoufek Ph.D.
|Work type:|| Bachelors
|Awards and prizes:|
|Abstract:||This thesis aims to describe structural changes in US stock markets during
and after global financial crisis. We utilize correlation coefficients of logarithmic
differences in daily closing prices to generate correlation networks.
Minimal spanning tree and hierarchical tree are used to filter out less important
information from correlation network, and thus they enable us to
obtain unique taxonomy of stocks. Daily closing prices from 8 June 2007
to 31 December 2010 for 73 constituents of market index S&P 100 are divided
into nine 100 trading-days-long time intervals. The effect of market
shock after the fall of Lehman Brothers on 15 September 2008 is investigated.
Minimal spanning tree significantly shrinks in the period from 15 September
2008 to 7 January 2009 and afterwards, it gradually reverts back to its precrisis
state. We also describe clustering patterns of stocks and their changes
during the crisis. Clusters of companies from financial, energy, and utilities
sectors are recognized in most time windows with only slight variations.
In the time window after 15 September 2008, several topological shifts are
identified. Additionally, companies from industrials sector are found to form
significantly larger clusters in time windows following 8 January 2009.