Tactical Asset Allocation in Fixed Income: Timing Bonds with Value, Momentum, and Carry

So you own bonds. You collected your term premium and credit premium from Chapter 2. Good. But can you do better than just holding them? Can you turn the dial up when bonds look attractive and dial it down when they don’t?

That’s what Chapter 3 is about. Richardson walks through a framework for tactically timing your bond exposure. And the honest answer is: yes, you can do a little better. But only a little.

The chapter borrows its working title from a research paper by Asness, Ilmanen, and Maloney: “Sin a Little.” That phrase sets the tone for everything here. You can time bonds. But don’t bet the farm on it.

The Three Signals

Richardson uses three signals to time bonds. These aren’t new ideas. They show up in stock investing all the time. But they work in fixed income too.

Value asks: are bonds cheap or expensive right now compared to some fundamental anchor?

Momentum asks: have bonds been going up or down recently, and will that trend continue?

Carry asks: how much do you earn just by holding the bond, assuming nothing changes?

Each signal gets applied to two different problems. First, timing the term premium (government bonds). Second, timing the credit premium (corporate bonds). Same three signals, different applications.

Timing the Term Premium

For government bonds, the expected return breaks down into two parts. There’s the initial yield you lock in when you buy the bond. And there’s what happens to yields after you buy.

Carry is measured as the term spread. That’s just the difference between long-term government bond yields and short-term bill yields. When the yield curve is steep, carry is high. You earn more for holding longer bonds.

Momentum is the trailing 12-month average of government bond excess returns. If bonds have been doing well recently, the momentum signal says keep going.

Value is trickier. It’s measured as the real bond yield, which is the nominal yield minus a survey-based forecast of long-term inflation. When real yields are high compared to history, bonds look cheap. When they’re low, bonds look expensive.

How They Turn Raw Data Into Signals

Here’s where it gets practical. You can’t just use raw numbers. Richardson walks through a careful signal construction process.

First, cap and floor extreme values at the 5th and 95th percentiles using only data available at that point in time. No peeking at the future.

Second, subtract the expanding median to benchmark the signal. This tells you whether today’s reading is high or low relative to history.

Third, scale by volatility so the signal strength means something consistent over time.

Fourth, apply a timing curve that maps the signal to a portfolio weight between 50% and 150% of your base allocation. So even in the most extreme case, you’re never going to zero or doubling up.

This is important. The framework is conservative by design. You stay invested. You just tilt a bit.

Results for Term Premium

Using a century of data (1920 to 2020), the results are modest but real.

Carry was the strongest signal for timing the term premium, with a Sharpe ratio of 0.31. Momentum came in at 0.15. Value was the weakest at 0.07.

Combining all three signals produced a Sharpe ratio of 0.29. The combination worked because the signals have low correlations with each other. Value and momentum were slightly negatively correlated at -0.17. Value and carry were basically uncorrelated at 0.01. Momentum and carry had a positive correlation of 0.48, partly because total return momentum includes realized carry.

But here’s the key point. The average excess return from the full timing model was about 0.42% per year. Compare that to the term premium itself, which averaged around 2% per year. The timing adds something, but it’s small relative to just showing up and holding the bonds.

And these returns are before trading costs.

Timing the Credit Premium

Corporate bonds bring a different challenge. The unique return source here is the credit spread, what you earn above risk-free yields. So the framework focuses on credit excess returns.

The same three signals apply, but they’re measured differently.

Carry is just the credit spread itself. A wider spread means more carry.

Momentum is the trailing 12-month average of corporate bond excess returns.

Value is more involved. Richardson runs a regression of credit spreads on three fundamental variables: corporate leverage, profitability, and stock market volatility. The residual from that regression (what the spread “should” be versus what it is) becomes the value signal. If the actual spread is wider than what fundamentals justify, bonds look cheap.

Investment Grade vs High Yield

This is where it gets interesting. The signals work very differently depending on which part of the credit market you’re looking at.

For US investment-grade bonds, the results were pretty weak. The correlations of value, momentum, and carry signals with next-month credit excess returns were basically zero: 0.01, -0.01, and 0.06, respectively.

For US high-yield bonds, the picture was much better. Value and momentum both showed meaningful correlations with future returns (0.09 and 0.08). The combined signal produced a Sharpe ratio of 0.48.

Why the difference? There’s simply more credit premium in high yield, and more variability in credit excess returns. You need that variability for timing to matter. Investment-grade bonds are just too stable for tactical signals to grab onto.

The combined timing model generated active returns of about 0.82% per year in high yield, compared to just 0.10% in investment grade.

One interesting pattern: momentum and carry have a very strong negative correlation, around -0.85 in both IG and HY markets. This makes sense. Momentum is positive when spreads have been tightening (prices going up). Carry is positive when spreads are wide. They’re basically opposite signals. Combining them with value helps smooth things out.

The Merton Model Connection

Richardson makes a cool detour into how credit and equity markets are linked through option pricing theory. Think of a company that has both debt and equity. If the company’s assets are worth way more than its debt, the debt is super safe. Changes in asset value barely affect the debt price.

But as the company’s assets shrink toward the value of its debt, something changes. The debt starts behaving more like equity. Small changes in asset value now move both the stock and the bond price in similar ways.

This has real implications for tactical timing. For riskier credits (high yield, emerging markets), equity market signals become more relevant. For safer credits (investment grade), less so. If you’re building a credit timing model, equity momentum and equity value signals might be worth including, especially for the high-yield part of your portfolio.

Data Mining: The Big Warning

Richardson is refreshingly honest about the risks here. With timing models, you only have one history. One data series. For the term premium, that’s a century of data. For the credit premium, it’s less than three decades.

And there’s a subtler risk beyond just running hundreds of regressions until something fits. Every choice you make when building a signal is a degree of freedom. Which data source for inflation expectations? Rolling or expanding window? Mean or median for benchmarking? Parametric or nonparametric volatility scaling?

Each individual choice might not matter much. But they compound. A study by Kessler, Scherer, and Harries (2020) looked at 3,168 alternative implementations of equity value strategies and found Sharpe ratios ranging from -0.10 to 0.78. Same concept, wildly different outcomes based on design choices.

There’s also the memory problem. Experienced investors remember what went wrong in the past. They add new variables to their models that weren’t available (or weren’t being used) at the time. That’s a subtle form of cheating.

The remedy? Be consistent. Use the same choices across geographies and asset classes. Don’t customize each model to fit its own history.

The Bottom Line

Chapter 3 gives you a complete framework for tactically timing bond exposure. Three signals. Two applications. Real out-of-sample results.

But the message is humble. The timing returns are small. They come before trading costs. And the risk of overfitting is real.

Richardson’s advice boils down to this: stay invested, stay diversified, and sin a little. Tactical timing can add some value at the margin. But the bulk of your return still comes from the strategic decision to hold bonds in the first place.

That’s a refreshingly honest take in a world where everyone wants to sell you a market-timing system.

Previous: Why Bonds Belong in Your Portfolio

Next: How Active Fixed Income Managers Really Perform

Book: “Systematic Fixed Income: An Investor’s Guide” by Scott A. Richardson, Ph.D. Published by John Wiley & Sons, 2022. ISBN: 9781119900139.