We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

q-fin.CP

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Quantitative Finance > Statistical Finance

Title: Predictability of Volatility Homogenised Financial Time Series

Abstract: Modelling financial time series as a time change of a simpler process has been proposed in various forms over the years. One of such recent approaches is called volatility homogenisation decomposition, and has been designed specifically to aid the forecasting of price changes on financial markets. The authors of this method have attempted to prove the its usefulness by applying a specific forecasting procedure and determining the effectiveness of this procedure on the decomposed time series, as compared with the original time series. This is problematic in at least two ways. First, the choice of the forecasting procedure obviously has an effect on the results, rendering them non-exhaustive. Second, the results obtained were not completely convincing, with some values falling under 50% guessing rate. Additionally, only nine Australian stocks were being investigated, which further limits the scope of this proof. In this study we propose to find the usefulness of volatility homogenisation by calculating the predictability of the decomposed time series and comparing it to the predictability of the original time series. We are applying information-theoretic notion of entropy rate to quantify predictability, which guarantees the result is not tied to a specific method of prediction, and additionally we base our calculations on a large number of stocks from the Warsaw Stock Exchange.
Comments: 8 pages, 6 figures, 1 table
Subjects: Statistical Finance (q-fin.ST); Computational Finance (q-fin.CP)
Cite as: arXiv:1406.7526 [q-fin.ST]
  (or arXiv:1406.7526v1 [q-fin.ST] for this version)

Submission history

From: Paweł Fiedor [view email]
[v1] Sun, 29 Jun 2014 17:14:22 GMT (2396kb)

Link back to: arXiv, form interface, contact.