There’s been no shortage of chatter recently about the near-record spec short in Treasurys and, more to the point, whether it presages a short squeeze.
After paring bearish bets a week after the 10Y short pushed to a new record, the position was rebuilt in the week through last Tuesday and now sits at 680K contracts.
Again, the prevailing narrative is that spec positioning is a contrarian indicator and that when things get as purportedly stretched as they are now on the short side, anything that catalyzes a quick rally in the long-end could squeeze that position, turbocharging the rally and leading to a sharp drop in yields.
Last month, Jeff Gundlach warned about just such a squeeze in a characteristically bombastic tweet. “Massive increase this week in short positions against 10 & 30 yr UST mkt”, Jeff tweeted on August 17. “Highest for both in history, by far”, he continued, before warning that it “could cause quite a squeeze.”
Or not. Because maybe what you think you’re seeing there isn’t what you’re actually seeing.
On Friday, for instance, Deutsche Bank suggested that what’s behind current CFTC positioning is a massive basis trade (short futures against long cash). In other words, there might not be a large duration short here at all.
We highlighted some excerpts from Deutsche’s analysis on Sunday evening, but for anyone who missed it, DB’s Steven Zeng notes that “there is a growing sense that positioning is not as stretched as the data would suggest and rather than outright duration, we think the data reflect basis positions held by leveraged funds and dealers”. Here are the relevant excerpts from the note:
The strongest evidence comes from our beta analysis of CTA returns. After controlling for curve changes, the partial beta of CTA returns to duration changes does not capture a large outright short bias. In fact, the beta has rose into positive territory since August, which suggests that CTA funds are either modestly long or at least neutral on duration. This diverges from the positioning data for leveraged funds which show near-record short positions. Our analytics show that the net basis for the front WN contracts has steadily cheapened since late spring, which is consistent with the timing of divergence in the CTA beta and positions data series. The same basis cheapening is also observed for FV and TY contracts.
Dealers have been running a large basis position as well. End of August data show that dealers were net long roughly $130bn 10yr equivalents of coupon Treasuries hedged by short positions of roughly $119bn 10yr equivalent of Treasury futures. This basis position is about twice as big compared to the start of this year. The position likely reflects the increase in inventory dealers had to take on due to increases in auction sizes this year.
The bank’s Aleksandar Kocic presented evidence to support that contention in a note to clients that made the rounds on August 29.
“While ED are ‘real’ shorts’ (reflecting possibly the Fed momentum trade), CFTC futures positioning reflects, at least partially, a basis trade that’s roughly duration neutral with slight directionality”, Kocic wrote, adding that there’s evidence of this in the vol. skew.
“The receiver premium increases as tenors shorten [which] can be interpreted as the price of insurance in the context of the current market positioning”, he said, elaborating.
Zeng reiterates that in the Friday note excerpted above. “Cheap low strike skew for longer rate tenors is not consistent with extreme outright bearish positions in the 10yr and 30yr sectors”, he contends.
So that’s something to think about.
Having put that out there, I wanted to kind of fire off a quick tangential factoid, in keeping with the generally freewheeling character of this post. Last week, the iShares 20+ Year Treasury Bond ETF raked in more than $1.6 billion in inflows. That’s the most in a year.
Make of it all what you will, but generally speaking, be wary of the contention that long-end yields are primed to snap lower at the first opportunity due to an assumed squeeze of an “outright” bet that might not in fact be as straightforward as some folks are suggesting.