Bonds are back: Mastering market timing for long-term success

In the (brief) risk-off periods in 2023 and 2024, bonds have rallied when stocks and other risky assets declined. Therefore the strategy can play a valuable role in portfolios with its strong performance in challenging market environments.

We continuously strive to enhance the market-timing strategy. The latest innovation is the addition of duration positions that aim to exploit shorter-term market inefficiencies. These positions should act as a diversifying source of alpha within the existing risk budget while staying true to the nature of the strategy: all active positions are taken in developed government bond markets, based on quantitative strategies.

The flagship of our dynamic duration offering is a global bond strategy, driven primarily by bond market timing through dynamic duration management. The strategy’s duration (interest-rate sensitivity) is increased when our quantitative model has a positive outlook on bond markets, and reduced when the model is negative. The strategy aims to offer protection against rising bond yields and to benefit strongly from declining yields. These active duration positions are implemented using futures in the main developed bond markets: US, Germany and Japan. The global government bond portfolio – on top of which the dynamic duration overlay is applied – is managed with an enhanced index approach, taking sustainability considerations into account. This portfolio is tilted toward government bonds that are deemed attractive based on factors like value, momentum and low-risk. Meanwhile, our portfolio construction algorithm also ensures that, compared to the index, the portfolio’s average ESG rating is better and the average carbon intensity is lower.

Government bond yields declined in the decade after the financial crisis and reached their lowest levels in 2020 during Covid, with negative yields even on long-dated German bonds. From these levels, further declines in yields and hence further gains for bonds were harder to achieve. But global bond yields have recovered: the yield of the JP Morgan Global Government Bond index is now 1% higher than the average since 1998.

As yields have recovered, bonds have regained the ability to rally strongly when yields decline in response to bad news. This was clear during the US regional banks crisis in spring 2023 and again in summer 2024, when the Nikkei index plunged 12% in a day, and weak US labor market data raised fears of a hard landing. Obviously, yields can also rise further. For example, after the strong rise in yields in 2021-2023, government bonds again lost nearly 3% in the first four months of 2024. And in October, US yields rose ca. 50 bps in a single month. So yes, bonds are back; with yields at better levels, bonds again offer diversification in risk-off periods. But keep in mind, bond volatility is back as well.

The return of bond volatility is actually good news for the Dynamic Duration strategy, as the best opportunities to add value through bond market timing arise when bond yields move significantly. While the live track record of the strategy’s quantitative duration management is already 27 years long, until recently, yields had mainly declined during this period. As we employ a quantitative strategy, we can use backtests and deep historical datasets to look back and analyze what the performance of such a strategy would have been in periods with predominantly rising yields, like the 1970s and early 1980s. That is precisely what we did in our paper published in The Financial Analyst Journal. We showed that the bond market timing strategy is both more profitable and more accurate (a higher success ratio) in calendar quarters with large changes in yields – both upward and downward.

Throughout its live track record, Robeco QI Global Dynamic Duration Strategy has performed in line with the pattern described above: it has, on average, delivered outperformance in the quarters with the weakest bond returns and in the quarters with the weakest equity returns, when bond returns tend to be strong. The average outperformance was in both scenarios clearly stronger than the average outperformance over the full track record. In Figure 2 we show the strategy performance in periods with strongly rising yields and in periods with large equity market declines. The right-hand panel shows the performance in the calendar quarters where the JP Morgan Global Government Bond Index declined more than 1% (now 18 quarters of the 108 quarters since the inception in 1998¹ ). Government bonds lost an average of 2.5% in these quarters. Robeco QI Global Dynamic Duration Strategy outperformed by 0.6% on average (not annualized!) in these quarters.

On the left-hand side of Figure 2, we focus on the calendar quarters where the MSCI World index declined more than 5%. This happened in 17 out of 108 calendar quarters since 1998. As the graph shows, government bonds on average gained 2.5% in these periods. Robeco QI Global Dynamic Duration’s Strategy average outperformance in these quarters was 1.1% (again not annualized).

Although our duration-timing research started more than 30 years ago, we keep striving for further enhancements. Our recent research has focused on inefficiencies that result in predictable shorter-term patterns in government bond markets. For example, we observe a V-shaped price pattern around government bond auctions: prices tend to decline in the run-up to the auction and to recover afterward. Figure 3 shows an example of bond futures prices around a US Treasury bond auction, where the vertical line denotes the time of the release of the auction results. As illustrated, the futures price declined before the auction, and recovered immediately after.

We can benefit from this pattern with a short position in government bond futures ahead of the auction, which is then closed and replaced by a long position beforehand. Markets generally absorb the new bond supply quickly, allowing prices to recover and the overweight to be again closed. We thus buy bonds when the Treasury is supplying new bonds to the market, benefiting from the temporarily lower prices. Our research shows how and when to exactly position to efficiently harvest the premium for this liquidity provision, taking risk considerations and transaction costs into account.

This auction strategy requires multiple trades within a matter of days. Compared to our regular duration positions, which tend to last for a few weeks to a few months, this is clearly a more short-term strategy. This results in a low correlation with the performance of these regular positions, providing diversification benefits within the strategy. We expect to generate some additional return without materially increasing absolute or relative risk.

The auction strategy is the first shorter-term signal that we have implemented in the dynamic duration strategy. We aim to add more signals in the future, capturing other market inefficiencies. When we do, we will not allow the combined shorter-term signals to take active positions larger than one year of duration. We will not increase the strategy’s tracking error limit; the new signals will be added within the existing risk budget.