Two months ago, the US long-end couldn’t sustain a bid to save its life and a hawkish Fed risked an over-tightening accident.
Now, bonds can’t sell off and market pricing suggests high odds of a rate cut as soon as the March FOMC meeting.
A relentless rally in US duration continued on Wednesday, as the Treasury curve bull flattened following a soft read on private sector hiring and amid an ongoing decline for crude.
This is by now one of the more impressive bond rallies I can personally remember, at least in terms of how quickly the tide turned and how unrelenting the subsequent reversal turned out to be.
10-year yields in the US were below 4.11% at one juncture Wednesday. 30-year yields were as low as 4.213%. Both are down ~90bps in six weeks.
There was some debate last month about the extent to which CTA covering in legacy STIR and bond shorts was driving the rally and, relatedly, about how much gas was left in the tank in that regard.
Regular readers will recall that a key piece of nuance was often lost in the short unwind discussion: There was a distinction between model signals. Specifically, longer-term windows for legacy shorts were still deep in-the-money, and because those windows had all the weight, the overall signals were still short. That’s changed now.
The aggregate CTA position in STIRS is now net long. The cumulative aggregate CTA position across G10 bonds is back to neutral.
As the figures, from Nomura, show, these positions haven’t been at current levels in two years.
“[The] legacy CTA trend short across fixed income has been bot-to-cover, first and most violently in the front-end, but as witnessed [Tuesday], the larger regime change felt in bonds is finally pushing into the long-end for mechanical trend investors, as those legacy shorts in 10s/20s/ 30s are now too seeing adjustments, with high-profile short-covering in USD 20-year/30-year driving local long-end outperformance,” Charlie McElligott said Wednesday.
McElligott described a “tectonic shift” for CTA positioning and a veritable “wipe out” of legacy trend shorts in G10 bonds and STIRS.
On Nomura’s estimates, CTAs bought nearly $105 billion of STIRS and more than $165 billion of G10 bond futures over the past month alone.
So, is it over? Not necessarily. “[A] further rally could still elicit plenty more CTA long buying demand, particularly in STIRS, although locally for bonds there isn’t a ton to buy proximately here,” McElligott went on, before cautioning that there’s some risk that any additional CTA accelerant flow could knock-on into pockets of negative convexity from dealers via what he described as “massive customer positions across SOFR- and UST- futures upside.”
For what it’s worth, TD’s rates team on Wednesday suggested taking profits in 10-year longs ahead of Friday’s US jobs report, which the bank’s Gennadiy Goldberg and Molly McGown said could put yields “at risk of backing up sharply.”




Makes one wonder about information value of market-derived forecasts of rate cuts.
Thanks for the post. Was wondering about this topic.
If I was a bond trader, but I buy & hold to mat., … I’d be taking $ off table tomorrow… velocity and overshoots are the new norm, imho. Money makers for the brave and nimble.
well, if not jobs report, then likely hawkish hold and hawkish leaning “killer” dots may frustrate the current uber longs…definitely seems that this cycle bottom is in for long suffering long bond holders, but a reasonable correction may be in order given recent strength…