Quantitative researchDec. 2022–Jul. 2023

Forecasting the Term Structure of Interest Rates with Dynamic SV Models

Public scope

This is a public project case study, not a publication or article. The manuscript, model details, data, and findings are not hosted on this site.

01Data horizonTen years
02Model setSix time-series models
03MarketChinese bond market
04RoleFirst author
Methods & tools
PythonRSPSSExcelLaTeXTime-series analysis
01 / Question

Modelling the shape and movement of rates

The project examined how the term structure of interest rates could be represented and forecast through dynamic stochastic-volatility approaches. It formed part of my earlier quantitative work in finance and established a foundation in model comparison, empirical testing, and careful interpretation of time-dependent data.

02 / Data

A decade of bond-market observations

I collected, cleaned, aligned, and prepared ten years of Chinese bond-market data. The preparation work included consistency checks, transformation for modelling, and creation of analysis-ready datasets for comparison across model specifications.

03 / Modelling

Six models under a common analysis workflow

I constructed six time-series models and used a consistent workflow to compare their treatment of interest-rate dynamics and forecasting behavior.

  • Data preparation and exploratory analysis
  • Model specification, estimation, and statistical testing
  • Forecast comparison and result visualization
  • Interpretation and academic documentation
04 / Contribution

A quantitative foundation for later evaluation work

As first author, I led the empirical workflow and used Python, R, SPSS, Excel, and LaTeX across data processing, modelling, statistical analysis, visualization, and writing. The project developed the evidence discipline that now informs my work on evaluation and assurance.

05 / Public boundary

Project record, not a publication page

This public case study records the project scope, role, data horizon, model count, and tools. It does not publish or link a manuscript, research article, model appendix, dataset, or detailed findings.

Open channel / Collaboration

Working on agent safety,
evaluation, or multilingual AI?