Predicting US Treasury Returns
Compute 4 z-scored signals from ETF returns (120-month rolling window). Backtest max drawdown: -10.4%.
Strategy & methodology
Compute 4 z-scored signals from ETF returns (120-month rolling window); Bond trend is binary (+1/-1); other 3 are continuous z-scores clipped to [-1, +1]; Average all 4 signals; map to continuous bond allocation: (avg + 1) / 2; Allocate…
- Strategy type:
- Tactical asset allocation
- Rebalance frequency:
- Monthly
- Original publication:
- 2020-06-19; results after that are out-of-sample for the original research. All results are backtest simulations.
- Data through:
- Backtest data through 2026-10-22.
Simulated history
Stand-in funds and until when (1)
- IEF: VFITX before Jul 26, 2002
Before these dates the backtest uses a stand-in, not the fund itself, so the results over those stretches show how the rules would have behaved, not what the fund returned.
Research data and disclosures
Compute 4 z-scored signals from ETF returns (120-month rolling window). Backtest max drawdown: -10.4%. This is a tactical asset allocation strategy. BestFolio supplies the public rule or approach and backtest context; current signals, allocations, and paid interactive data remain restricted to Pro access. Users review the published signal and place any resulting trades in their own brokerage. Displayed returns remain hypothetical and do not represent a customer's brokerage record. The facts above show how current the data is.
- BestFolio supplies
- The monthly signal email and this strategy page; current signals and email alerts require Pro access.
- Customer action
- Review the published signal and place any required trades in your own brokerage. BestFolio does not execute orders.
- Costs and exclusions
- Backtests are net of a modeled one-way transaction cost (10 bps, scaled up to 3x under stress); taxes, fund-expense drift, or market impact are not modeled. No tax, no slippage beyond the stated cost. Methodology limitations
Published result: Engine drift-until-flip-v1, data version 93e9352e, published 2026-10-17
Is Predicting US Treasury Returns still working in 2026?
Predicting US Treasury Returns returned 1.35% over the trailing 12 months and 11.33% over 36 months through 2026-10-22, compared with a full-backtest annualized return of 5.44%. Its full-backtest maximum drawdown was -10.42%. The full sample contains 10533 daily NAV observations from 1985-01-02. These are model results, not investor account returns or a promise. As of 2026-10-22 it is -0.33% below its high-water mark of 2026-02-27, 8 months ago, and its longest run below a previous high was 3.8 years. Recent returns do not establish that the strategy will keep working.
| Period | Return | CAGR | Max drawdown | Observations | Dates |
|---|---|---|---|---|---|
| Trailing 12 months | 1.35% | Not annualized | -1.90% | 252 | 2025-10-22 to 2026-10-22 |
| Trailing 36 months | 11.33% | Not annualized | -3.10% | 754 | 2023-10-20 to 2026-10-22 |
| Full backtest | 814.37% | 5.44% | -10.42% | 10533 | 1985-01-02 to 2026-10-22 |
Last verified
Common questions about these results
Are these live investor returns?
No. These are the latest model NAV results from the published backtest. A recent date alone does not make a result an independently observed live record. Investor costs, taxes and execution can differ.
Why can a strategy lag for a year?
A tactical model can hold defensive assets during a rally or change positions during reversals. A short window can differ substantially from its full history. Compare cumulative returns over matching dates and inspect drawdowns as well.
How long has it spent below a previous high?
Its last high-water mark was 2026-02-27, 8 months before 2026-10-22, and it is -0.33% below that level now. The longest run below a previous high in the full backtest was 3.8 years. Recovering from a drawdown can take years, and a strong trailing return does not mean a past high has been regained.
Where can I check the signals behind these results?
The Signals tab on this page lists each dated model decision for the selected variant; for Pro strategies they are visible to Pro members. The methodology page explains the backtest assumptions.
Predicting US Treasury Returns at a glance
Predicting US Treasury Returns is a tactical asset allocation (TAA) strategy by Baltussen, Martens & Penninga across Treasuries, T-Bills, rebalanced monthly. Backtested 1985-01-02 to 2026-10-22 (41.8 years): 5.4% CAGR, 0.50 Sharpe, -10.4% max drawdown, 4.2% volatility.
- Type
- Tactical (TAA)
- Author
- Baltussen, Martens & Penninga
- Rebalancing
- Monthly
- Period
- 1985-01-02 to 2026-10-22
- CAGR
- 5.4%
- Sharpe
- 0.50
- Max Drawdown
- -10.4%
- Volatility
- 4.2%
Predicting US Treasury Returns — Tactical Asset Allocation Strategy
Predicting US Treasury Returns (Baltussen, Martens & Penninga, 2021) combines four signals into an ensemble model for bond timing: yield spread (IEF-BIL return proxy), bond trend (12-month IEF excess return), equity returns (inverted: low stocks = bullish for bonds), and commodity returns (inverted: low commodities = bullish for bonds). Each signal is z-scored over a 10-year rolling window and capped at [-1, +1]. The average signal maps to a continuous allocation between IEF and BIL.
Predicting US Treasury Returns: frequently asked questions
- What is Predicting US Treasury Returns?
- 4-signal ensemble (yield spread, bond trend, equity returns, commodity returns) for continuous bond/cash allocation.
- Who created the Predicting US Treasury Returns strategy?
- Predicting US Treasury Returns was developed by Baltussen, Martens & Penninga. It is based on Baltussen, G., Martens, M. & Penninga, O. (2021). Predicting Bond Returns: 70 Years of International Evidence. FAJ 77(3)..
- What is the historical return and maximum drawdown of Predicting US Treasury Returns?
- Backtested from 1985-01-02 to 2026-10-22, Predicting US Treasury Returns returned 5.4% CAGR with a -10.4% maximum drawdown and a Sharpe ratio of 0.50. Past performance does not guarantee future results.
- How often is Predicting US Treasury Returns rebalanced?
- Predicting US Treasury Returns is rebalanced monthly. BestFolio publishes the updated allocation signal each period.
- Is Predicting US Treasury Returns a tactical asset allocation strategy?
- Yes. Predicting US Treasury Returns is a tactical asset allocation (TAA) strategy: it adjusts its holdings based on market signals each period rather than holding a fixed allocation.
Backtest Performance (1985-01-02 to 2026-10-22)
| Metric | Predicting US Treasury Returns |
|---|---|
| CAGR | 5.4% |
| Max Drawdown | -10.4% |
| Sharpe | 0.50 |
| Sortino | 0.89 |
| Volatility | 4.2% |
| Calmar | 0.52 |
| Total Return | 812.9% |
| Backtest Period | 41.8 years |
Every rebalance fills at the signal-day close, net of modeled transaction costs. Followers trade at the next open; the delayed-close line in the Rebalance Frequency Sensitivity card shows the effect of trading one session later. Execution assumption
Strategy Details
- Type
- Tactical (TAA)
- Rebalancing
- monthly
- Variants
- 1
- Author
- Baltussen, Martens & Penninga
- Source
- Baltussen, G., Martens, M. & Penninga, O. (2021). Predicting Bond Returns: 70 Years of International Evidence. FAJ 77(3).
Asset Classes
- Treasuries
- T-Bills
Further reading
New to this approach? Read what tactical asset allocation is and how it works.
Holding Predicting US Treasury Returns alongside another strategy? Use the free portfolio overlap calculator to see how much of the two portfolios actually differs.
Track Predicting US Treasury Returns in Your Portfolio
Sign up for BestFolio to get monthly rebalancing signals, blend strategies into custom portfolios, and receive alerts when allocations change.