Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
A self-normalizing neural network approach to bond liquidity classication
KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
2018 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesisAlternative title
Självnormaliserade neurala nätverk för klassificering av obligationer likviditet (Swedish)
Abstract [en]

Bond liquidity risk is complex and something that every bond-investor needs to take into account. In this paper we investigate how well a selfnormalizing neural network (SNN) can be used to classify bonds with respect to their liquidity, and compare the results with that of a simpler logistic regression. This is done by analyzing the two algorithms' predictive capabilities on the Swedish bond market. Performing this analysis we find that the performance of the SNN and the logistic regression are broadly on the same level. However, the substantive overfitting to the training data in the case of the SNN suggests that a better performing model could be created by applying regularization techniques. As such, the conclusion is formed as such that there is need of more research in order to determine whether neural networks are the premier method to modelling liquidity.

Abstract [sv]

Likviditeten hos obligationer är komplicerad och ett fenomen som varje obligationsinvesterare måste ta itu med. I den här rapporten undersöks hur pass väl ett själv-normaliserande neuralt nätverk kan användas för att klassifiera obligationer med avseende på deras likviditet, samt jämförs detta resultat med när en simplare logistisk regression används. Detta görs genom att analysera de två algoritmernas prediktiva kapacitet på den svenska obligationsmarknaden. Efter genomförd undersökning finner vi att SNN och logistisk regression presterar på liknande nivåer. I fallet med SNN finns dock en stor overfit till träningsdatan, vilket indikerar att en bättre modell möjligtvis skulle kunna nås om vanliga regulariseringsmetoder skulle användas. Slutsatsen blir därmed att det finns behov av mer forskning på ämnet för att dra en konklusion huruvida neurala nätverk är den bäst lämpade samlingen av algoritmer för modellering av likviditet.

Place, publisher, year, edition, pages
2018.
Series
TRITA-SCI-GRU ; 2018:231
National Category
Probability Theory and Statistics
Identifiers
URN: urn:nbn:se:kth:diva-229405OAI: oai:DiVA.org:kth-229405DiVA, id: diva2:1212535
External cooperation
Kidbrooke Advisory
Subject / course
Mathematical Statistics
Educational program
Master of Science - Applied and Computational Mathematics
Supervisors
Examiners
Available from: 2018-06-02 Created: 2018-06-02 Last updated: 2018-06-02Bibliographically approved

Open Access in DiVA

fulltext(684 kB)52 downloads
File information
File name FULLTEXT01.pdfFile size 684 kBChecksum SHA-512
8d9227aefc9a634ba958df52fd8d62d1bbeca4d01c8832fa45157b87c3d8ec5b308a9cca568347b6c78f50b2dc0a7500c28b8d221f3db05e6ab6fc2c5f858fb1
Type fulltextMimetype application/pdf

By organisation
Mathematical Statistics
Probability Theory and Statistics

Search outside of DiVA

GoogleGoogle Scholar
Total: 52 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 193 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf