Systematic macro across 21 equity, bond, currency and commodity markets (2008–2026): volatility targeting, risk parity across asset classes, turnover costs, a carry sleeve that did not pay, a crisis record that did, and a gradient-boosted ML overlay tested under purged walk-forward validation that underperformed the simple rule. Benchmarked against Goldman Sachs' Managed Futures Strategy Fund.
Read the note: docs/part3_macro.md · Part 3 of the GS QIS product review · companion to the multi-factor strategy.
| Program | 60/40 | GMSSX (since 2012) | |
|---|---|---|---|
| Sharpe | 0.41 | 0.71 | 0.24 |
| Max drawdown | -20.7% | -29.7% | -19.4% |
| Corr. to S&P 500 | -0.15 | 0.98 | -0.09 |
| 2008 / 2020 / 2022 | +15.2% / +15.0% / +13.7% | -28.8% / -11.5% / -15.8% | n/a / +4.9% / +20.6% |
src/data.py prices (yfinance) and short rates (FRED) src/ml.py features, purged walk-forward GBM, importance
src/strategy.py signals, sizing, simulation, sleeves src/analysis.py metrics, benchmarks, crisis table, charts
docs/ the note output/ tables (csv), figures/
Run order: data -> strategy -> ml -> analysis. Requires the factor-strategy repo's returns_v2_betaneutral.parquet for one correlation column.
Hugo Hoenn, September 2026. Not investment advice.