Work detail

Predicting Financial Market Crashes using Log-periodic Oscillation and Critical Slowing Down

Author: Bc. Daniel Štancl
Year: 2019 - summer
Leaders: prof. PhDr. Ladislav Krištoufek Ph.D.
Work type: Bachelors
Language: English
Pages: 98
Awards and prizes:
Abstract: This bachelor thesis concerns itself with multiple objectives. First, to compare
two apparently contradictory frameworks, namely the Log-periodic Power Law
model and the Critical Slowing Down, suggested as being able to detect the end
of financial bubbles. Second, to enrich current literature dedicated to the Logperiodic Power Law model with a comprehensible description of the non-linear
optimization methods in one piece of work. This work, furthermore, aims to compare the performance and the robustness of two versions of this model. Regarding
the Critical Slowing down, the correlation across the world market over time prior
to a crash is investigated as an addition to two already studied indicators, 1-lag
serial correlation and standard deviation of detrended fluctuations. Eventually,
both the Log-periodic Power Law models were proved to be able to identify the
time of the burst of the financial bubble, while the modified version of the model
was found to be more proficient over the initial one in terms of computational
efficiency and robustness. In the case of the Critical Slowing Down, obeying autocorrelation of residuals and cross-correlation of intermarket residuals came out to
be misleading, and only variance was supported as an appropriate indicator of an
imminent tumble, and it was proposed as an aspirant for a potential completion
of the Log-periodic Power Law model framework.


Česká Spořitelna


Patria Finance