Quantitative Risk Analytics — R

Overview

A quantitative risk analytics project implemented entirely in R (via IRkernel in Jupyter). It applies CAPM regression, parametric and historical VaR, and CVaR to ASC.L (ASOS) and AV.L (Aviva) benchmarked against the FTSE 100, with automated ggplot2 visualisations throughout.

Data is sourced via quantmod from Yahoo Finance (Jul 2024 – May 2025, ~259 trading days).

---

What's Inside

Section 1: Setup & Library Imports

Installs and loads: quantmod, PerformanceAnalytics, ggplot2, tidyverse, quadprog

Section 2: Data Download

Downloads adjusted close prices for ASC.L (ASOS) and AV.L (Aviva) plus FTSE benchmark using getSymbols() from quantmod.

Section 3: Return & Excess Return Calculation

• Computes daily returns using mutate() with lagged prices
• • Calculates excess returns using rf_annual = 0.045 (4.5% risk-free rate)

  • • Produces a clean tibble with 7 columns (date, raw returns, excess returns) × 259 rows

    • Section 4: CAPM Model Implementation

    • Full CAPM theory markdown with formula:

    • E(Ri) = Rf + βi(E(Rm) - Rf)
      Ri - Rf = αi + βi(Rm - Rf) + εi
      

      OLS regression via lm() for each stock:

      • beta_asc = coef(lm(ASC_excess ~ Market_excess))[2]
      • • beta_av = coef(lm(AV_excess ~ Market_excess))[2]

        • • Computes CAPM-implied expected returns for ASC and AV

          • Section 5: Parametric VaR (95% & 99%)

          • Uses the Gaussian assumption:

          • VaR = μ - z * σ
            

            Computed for both stocks at 95% and 99% confidence.

            Section 6: CVaR / Expected Shortfall

            Computes the mean of returns below the VaR threshold for each confidence level.

            Section 7: ggplot2 Visualisations

            • Return distribution histograms with VaR/CVaR overlaid

            • • CAPM regression scatter plots

              • • Correlation plots

                ---

                Technologies Used

                Tool	Purpose
                R (IRkernel)	Core language — pure R notebook
                quantmod	Yahoo Finance data download
                ggplot2	Professional visualisations
                tidyverse (dplyr, tibble)	Data wrangling
                PerformanceAnalytics	Risk metric utilities
                Jupyter Notebook	Interactive execution via IRkernel

                Note: This notebook runs R natively via IRkernel, not Python with rpy2. You need R and the IRkernel installed to execute it.

                Getting Started

                # In R, install required packages
                install.packages(c("quantmod", "PerformanceAnalytics", "ggplot2", "tidyverse"))
                
                # Install IRkernel for Jupyter
                install.packages("IRkernel")
                IRkernel::installspec()

                Then launch Jupyter and select the R kernel to run the notebook.

                git clone https://github.com/srhimal149/srhimal149-Quantitative-Risk-Analytics-R.git
                jupyter notebook

                Author

                Md Saifur Rahman Himal MSc FinTech (Distinction Candidate) | Quantitative Finance & Risk Analytics LinkedIn | GitHub

                License

                MIT License