Research-Grade Macro-Regime Portfolio Backtesting (All-Weather Strategy, ML Extension )

Title:
R Developer for Research-Grade Macro-Regime Portfolio Backtesting (All-Weather Strategy, ML Extension )

Description:
We're conducting a quantitative finance research project exploring the performance of a macro-regime-aware All-Weather Portfolio, inspired by risk parity and Bridgewater’s approach. The project is fully scoped and led by our internal team — we're seeking a technically skilled R developer with domain knowledge in financial modeling to support the core implementation.

Project Goal
To compare a regime-based All-Weather Portfolio against benchmark allocations (60/40, all-equity), and assess:
Performance across macroeconomic states
Risk-adjusted returns (Sharpe, Sortino)
Robustness of regime logic
Enhancements using Bitcoin allocation and machine learning classifiers

What You’ll Build

1. Data Handling

- Monthly/quarterly data from 1970–2021
- Assets: S&P 500, Gold, GSCI, US Treasuries (5y, 10y, 20y)
- Macro indicators: CPI, real GDP, VIX, yield spreads, bond spreads

2. Regime Classification

- 4 macro regimes based on GDP and inflation deviations from 3-year trend
- Mapping asset class performance to economic states
- Implement OLS regressions & average return analysis

3. Portfolio Modeling

- Construct 4 sub-portfolios with risk parity weights per regime
- Backtest with quarterly rebalancing and risk balancing logic
- Rolling holding period returns (1, 5, 10, 20 yrs)

4. Performance Analytics
- Drawdown analysis
- VaR / CVaR
- Time-decayed Sharpe ratios
- Macro-sensitivity modeling

5. Extensions

- Integrate Bitcoin as an additional asset class
- Implement ML classifiers (Random Forest / XGBoost / Logistic Regression) to classify macro regimes and dynamically adjust allocation weights

Who This Is For

Required:
- Strong skills in R programming (esp. tidyverse, lubridate, PerformanceAnalytics, xts/zoo)
- Solid understanding of portfolio construction, asset pricing, and macro-financial theory
- Experience with regime switching, rolling-window analysis, and data alignment across frequencies

Preferred:
- Exposure to machine learning for time series or classification
- Familiarity with Quarto / RMarkdown for documentation
- Knowledge of cryptocurrency asset behavior, especially BTC post-2010

If your experience is limited to basic data manipulation or general scripting, this project is not a fit. Please apply only if you’ve worked with backtesting, financial time series, or econometrics.

Deliverables:
- Clean, commented R code modules
- Reproducible analytics pipeline
- Final set of result tables and figures
- Documentation of economic rationale used in modeling
- GitHub push or zipped archive