U.S. Treasury Yield-Curve Rotation Strategy

Long-duration U.S. Treasuries (TLT, ~17-year duration) suffer severe drawdowns during rapid rate-hiking cycles (losing −27% in 2022 even while the 2s10s yield curve remained positive), making naive yield-curve bucket rules vulnerable to inflation shocks.

Can a systematic, long-only monthly rotation across U.S. Treasury duration ETFs (TLT, IEF, SHY) capture macro regime shifts from the 2s10s spread while protecting capital during rate-shock bear markets, net of realistic execution lag and transaction friction?

23 years of daily market and macro data (Jan 2003 – May 2026, 5,800+ trading days): Yahoo Finance auto-adjusted total-return series for TLT (20+ Yr), IEF (7–10 Yr), SHY (1–3 Yr), AGG, and SPY, merged with FRED DGS10 and DGS2 constant-maturity Treasury yields shifted by 1 day (yields.shift(1)) for strict zero look-ahead.

Classified the macro rate regime monthly using the lagged 2s10s spread (DGS10 − DGS2): Steep (>1.0% → TLT), Flat (0–1.0% → IEF), and Inverted (≤0% → SHY), enhanced with the V4 Price Filter switching to near-cash SHY whenever the selected ETF's 3-month trailing return is negative. Modeled realistic friction (T+1 backtest execution with 2 bps commission + 2 bps slippage; T+2 out-of-sample paper trading from Jan 2024 to May 2026 with stressed 5 bps slippage). Validated robustness via 2D threshold/lookback plateau grids, 2,000-iteration block bootstrap, walk-forward testing, and Newey-West HAC / Probabilistic Sharpe Ratio (PSR) / Jobson-Korkie statistical tests.

Outperformed passive buy-and-hold IEF over 2003–2026 with 4.08% CAGR vs. 3.38% (+70 bps/yr), improved Max Drawdown to −21.2% vs. −23.9% (−26.6% unfiltered), lifted Calmar ratio by +36% (0.19 vs. 0.14), and generated a statistically significant HAC CAPM alpha of +3.51%/yr (t = 2.30, p = 0.02, β_SPY = −0.16). During the 2022 bond crash, the V4 filter exited TLT before macro inversion, limiting calendar losses to −5.4% vs. −15.2% for IEF (+982 bps tail protection). Out-of-sample paper trading (9 rotations, 19.1 bps mean drift) confirmed drawdown protection (−4.68% vs. −6.89% for IEF).

Institutional-grade quantitative research pipeline demonstrating zero-look-ahead backtesting, realistic T+2 execution modeling, non-overfit parameter plateau validation, and transparent reporting of regime limitations.