English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

To diversify, investors should avoid adding assets to their portfolio when their prices exhibit high correlation. Industry diversification is a common portfolio diversification method. It is likely that the notion of investing across different industries can help achieve portfolio diversification, as companies in different industries are likely to have different revenue and cost drivers. However, results across various studies have been mixed. This study seeks to identify a novel application to diversify portfolios to overcome the mixed results of industry diversification, through the use of unsupervised time series clustering-based machine learning technique. There are various ways of clustering time-series, namely shape-based, model-based and feature-based. Feature-based approach faces limitation for the need of equal-length feature vectors, and model-based approach faces limitation in terms of scalability. In this research, a shape-based clustering approach, which overcomes the aforementioned limitations, and specifically agglomerative hierarchical clustering algorithm (AHC-DTW), with dynamic time warping as the distance measure, is utilized to perform diversification. AHC-DTW allows clustering to be conducted across different temporal lengths, many-to-one point comparison to measure distances rather than one-to-one point comparison for euclidean distance. Further, AHC-DTW remains robust with scaling and shifting, unlike for instance, euclidean approach which requires clustering of the same time length, and is highly sensitive to outliers, noise, and transformations. The shape-based clustering approach implemented seeks to match the shapes of time series data as closely as possible. Since shape-based clustering technique groups together cumulative stock returns that trends closely across time, it will be intuitive that investors investing in more than one stock in the same cluster will not be better off, in contrast to diversifying investments across different clusters. Research found clear outperformance of shaped-based cluster diversification against industry diversification. Annualized mean return improved by 598 basis points, and Sharpe performance measure improved by 337%. Further, research found that AHC-DTW clustering exhibited time persistency. These robust results suggest promise for industry practitioners in the utilization of shape-based cluster diversification for enhanced investment portfolio performance.

Portfolio Management: A Financial Application of Unsupervised Shape-based Clustering-Driven Machine Learning Method