Fixed-income research · Asset allocation

Regime-Aware Fixed-Income Allocation

A transparent allocation framework that combines macro-policy conditions and market stress to adjust exposure across duration, credit, inflation protection, liquidity, and safe-haven assets—with an emphasis on portfolio governance and practical implementation.

Institution: Frankfurt School of Finance & Management
Study period: December 1969–December 2024
Primary objective: Capital preservation

Full presentation (PDF) Client cover letter Source notebook

Executive summary

A defensive allocation process for changing regimes

Primary mandate

The framework was designed to reduce downside risk relative to a static strategic allocation. In the full-period backtest, maximum drawdown was −10.1%, compared with −16.2% for the strategic benchmark.

Secondary mandate

The strategy sought positive risk-adjusted returns through interpretable regime signals and explicit risk controls. The full-period return-to-volatility ratio was 1.89 versus 1.07 for the benchmark.

Annualised return5.89%Benchmark 5.18%

Annualised volatility3.11%Benchmark 4.85%

Maximum drawdown−10.1%Benchmark −16.2%

Return / volatility1.89Benchmark 1.07

Data coverage

Analysis period: 661 monthly observations
Asset universe: 15 return series
Strategic categories: 8 allocation sleeves
Macro indicators: 12 policy and market signals
Regime states: 4 classifications
Historical tests: 6 selected periods

Technical reference

Regime classification and allocation rules

A policy-based primary score and a market-based secondary score are dynamically weighted. Rolling percentile thresholds map the combined score to four states. Those states determine base allocations, which are then adjusted by volatility, drawdown, correlation, and tail-risk controls.

1Observe

Update policy, macro, volatility, liquidity, and geopolitical inputs.

2Classify

Translate the combined signal into one of four market regimes.

3Allocate

Map the regime to transparent asset-class weights and limits.

4Control

Apply drawdown, volatility, correlation, and tail-risk adjustments.

Regime classification thresholds

Regime	Combined-score threshold	Historical frequency	Months
Crisis	Above 80th percentile	20.9%	208
Risk-Off	60th–80th percentile	19.8%	197
Neutral	40th–60th percentile	18.8%	187
Risk-On	Below 40th percentile	40.6%	404

Frequencies refer to the longer regime-classification history, not the 661-month investable backtest window.

Three-panel regime classification framework showing the combined score, component scores, and historical regime timeline

Average asset allocation by regime

Asset class	Crisis	Risk-Off	Neutral	Risk-On
Cash / short	41.3%	33.4%	18.3%	10.9%
Government long	25.0%	28.9%	26.5%	18.4%
Investment grade	18.1%	26.0%	38.1%	42.4%
High yield	1.1%	3.9%	11.6%	23.3%
TIPS short	2.0%	0.6%	0.2%	0.04%
TIPS long	3.2%	0.9%	0.1%	0.02%
Gold	4.5%	2.8%	1.7%	1.5%
FX / currency	4.8%	3.6%	3.4%	3.4%

Average asset-class weights across four market regimes

Performance by regime

Regime	Annualised return	Volatility	Return / volatility	Maximum drawdown
Crisis	7.98%	3.44%	2.32	−3.59%
Risk-Off	5.58%	2.96%	1.88	−3.55%
Neutral	3.09%	3.39%	0.91	−7.83%
Risk-On	6.35%	2.85%	2.23	−3.94%

Full-period correlation matrix across the asset universe with coefficients in each cell

Backtest results

Full-period performance and implementation audit

Full-period results

Metric	Regime-aware	Strategic allocation	Difference
Annualised return	5.89%	5.18%	+0.71 pp
Annualised volatility	3.11%	4.85%	−1.74 pp
Return / volatility	1.89	1.07	+0.82
Sortino ratio	3.08	1.64	+1.44
Calmar ratio	0.58	0.32	+0.26
Maximum drawdown	−10.10%	−16.23%	+6.13 pp
95% conditional VaR	−1.53%	−2.69%	+1.16 pp
Monthly win rate	73.2%	66.3%	+6.9 pp
12-month loss probability	7.1%	13.2%	−6.1 pp

Conservative implementation check

Applying a one-month signal lag, limiting the comparison to February 2005–December 2024, and charging an illustrative 10 bp one-way cost lowers the strategy’s return / volatility ratio to 0.99. Maximum drawdown remains shallower than the benchmark at −11.05% versus −16.23%.

Portfolio growth and drawdown comparison between the backtest implementation, a lagged cost scenario, and the strategic benchmark

Historical stress testing

Performance across six market environments

Test periods

Period	Dates	Duration	Purpose
Dot-com bust	Jan 2000–Dec 2002	36 months	Equity and credit repricing
Global financial crisis	Jul 2007–Dec 2009	30 months	Systemic liquidity shock
Calm period	Jan 2017–Dec 2019	36 months	Low-volatility control window
COVID-19	Jan 2020–Dec 2020	12 months	Rapid risk-off and policy response
Inflation surge	Jul 2021–Dec 2022	18 months	Rates and inflation shock
2024 volatility	Jan 2024–Dec 2024	12 months	Recent out-of-sample-style review

Performance by historical period

Period	Portfolio	Cumulative return	Annualised return	Return / volatility	Maximum drawdown
Dot-com bust	Regime-aware	25.8%	7.7%	2.92	−1.7%
Dot-com bust	Strategic allocation	26.5%	7.9%	2.22	−2.1%
Global financial crisis	Regime-aware	14.7%	5.5%	1.67	−3.5%
Global financial crisis	Strategic allocation	17.9%	6.8%	1.09	−7.6%
Calm 2017–2019	Regime-aware	14.6%	4.6%	1.61	−2.0%
Calm 2017–2019	Strategic allocation	14.8%	4.7%	1.35	−2.9%
COVID-19	Regime-aware	7.6%	7.3%	3.00	−0.7%
COVID-19	Strategic allocation	9.5%	9.3%	1.90	−1.7%
Inflation 2022	Regime-aware	−7.4%	−5.0%	−1.56	−9.4%
Inflation 2022	Strategic allocation	−12.6%	−8.7%	−1.20	−16.2%
2024	Regime-aware	4.7%	4.6%	2.08	−0.6%
2024	Strategic allocation	1.8%	1.9%	0.34	−2.9%

Maximum drawdown comparison across six historical periods

Complete presentation

Research, architecture, validation, and implementation

The original 44-page team presentation is included in full.

Download presentation (PDF)