This code makes up the analysis part of a project aimed at pitching a Convertible Bond Arbitrage (CBA) fund to a Pension fund. The Pension fund holds long-only positions in equities and bonds. We use monthly return data. The goal is to demonstrate how well the strategy diversifies a portfolio and holds up during adverse market conditions (hedge efficiency).
- ARBIX — Monthly return data for the ARBIX fund, sourced from Absolute Investment Advisers
- Equities — MSCI World Index (broad global equity proxy)
- Bonds — iShares 7–10 Year Treasury Bond ETF (chosen as a middle ground between cash-like behaviour and interest rate sensitivity)
A clustered bar chart showing the Pearson correlation between ARBIX and each of the two asset class proxies (equities and bonds). Correlations are calculated across:
- A full sample (entire available date range)
- Four sub-periods to test behaviour across different market regimes:
- GFC (Dec 2007 – Jun 2009) — stress
- Post-GFC Expansion (Jan 2012 – Dec 2013) — growth
- Covid Stress (Feb 2020 – Dec 2020) — stress
- CB Renaissance (Jan 2023 – Dec 2025) — growth
Stress periods are shaded red; growth periods are shaded green. A correlation near zero indicates the strategy moves independently of that asset class.
A horizontal bar chart comparing the annualised return volatility (standard deviation) of ARBIX, equities, and bonds. Lower volatility relative to equities indicates the strategy is less prone to large swings in value, which is useful for portfolio diversification.
A vertical bar chart comparing ARBIX's average monthly return across all months versus only equity down months (months where the equity index fell). This tests how well the strategy holds up when equity markets are falling.
| Statistic | Value | Based On |
|---|---|---|
| Downside Capture Ratio (ARBIX vs Equities) | 2.88% | 104 equity down months |
| Down-Market Beta (ARBIX vs Equities) | 0.1827 | 104 equity down months |
- Downside Capture Ratio — measures how much of the equity benchmark's decline ARBIX captures during down months. A value closer to 0% is better, indicating the fund loses far less than equities when markets fall.
- Down-Market Beta — a version of standard Beta calculated only during equity down months. A value below 1 (and especially near 0) indicates low sensitivity to adverse equity movements.
The code automatically checks for the following before running any analysis:
- All required columns (
Month,Year,Return) are present in each data sheet - No duplicate dates exist within a series
- The dataset is trimmed to the common date range shared across all three series
- Any months with missing values are flagged and excluded, with a count printed to the console
- Sub-periods with fewer than 11 observations are flagged and excluded from the correlation results
Running Code.py produces the following files:
Correlation_Chart.pngVolatility_Chart.pngDown_Market_Chart.pngClean_Data.xlsx— contains three sheets: cleaned return data, correlation results, and downside statistics