Flash Crashes, Flash Rallies And A Trip Across Fragile Markets

by: The Heisenberg

As I was perusing some articles ahead of the Fed on Wednesday, I ran across a post documenting a bold bet on Treasury yields falling to 2.60% by late August.

That would represent a pretty notable rally and I wanted to put that in the context of the three recent "flash" rallies we've seen in USTs.

That effort leads naturally to a discussion of flash crashes and flash "smashes" (if you will) more generally.

There's a ton of new research out recently on this subject and I wanted to touch on some of it here to make a series of points about liquidity provision and fragility in modern markets.

Well, we got the Fed today, and the message was hawkish. But I don't want to talk about that here.

Or actually, I do want to talk about it, but only tangentially.

As I was pondering what I wanted to write about in my daily column for Dealbreaker, I ran across a Bloomberg article detailing what, all told, amounts to a $75 million options bet on 10-year Treasury yields falling to 2.60% by the end of August.

That would be notable on its own considering 10-year yields were at their highest levels since 2011 just weeks ago, but it was all the more notable considering the bets were placed on Monday, Tuesday and Wednesday. In other words, they were placed ahead of the Fed.

It should go without saying that those bets run counter to the massive spec short in TY:


In my Wednesday post for Dealbreaker I basically expanded on what Bloomberg suggested in their post detailing the trades. Namely that it's possible traders are betting on some manner of exogenous shock forcing a Fed relent or, perhaps more to the point, on something like political turmoil in Italy, an escalation in trade tensions and/or emerging market turmoil catalyzed by a hawkish Fed sparking a flight to safety that would manifest itself in demand for Treasurys (TLT).

Clearly, there are all manner of other explanations, but considering the circumstances (i.e., considering recent Fed-related volatility in EM and the distinct possibility that the ECB is about tip their hand when it comes to slapping a sell-by date on the asset purchases that have helped to keep a lid on periphery yields and spreads for years), it's at least plausible to suggest that sizable bullish bets on Treasurys are an expression of a someone's view that a meaningful risk-off episode is in the cards over the next couple of months.

That reminded me of last Thursday, when a leg lower (weaker) in the Brazilian real triggered a "flash rally" in U.S. Treasurys. I documented that on my site and then at length in these pages in a post called "Hang 'EM High: Fed Watches As The Emerging World Burns". Here's the chart:


As I noted in the Dealbreaker post mentioned above, Bloomberg's account of the BRL plunge detailed the extent to which dollar offers “vanished” on the local futures market last Thursday. The market witnessed “a total absence of USD sellers” on multiple occasions, to quote Bloomberg again.

While there were factors at play there that are unique to BRL, what's interesting is how quickly it morphed into the "flash rally" in Treasurys shown above and clearly evident in a 10-year yield chart:


That came just a little over a week after liquidity disappeared completely for Italian bonds amid acute concerns about the prospect of new elections. That same day, U.S. Treasurys rallied the most since Brexit and German bund yields plunged in what amounted to a 7-standard deviation move:

(Source: Citi Research)

That episode is what ultimately blew up Bill Gross, and if you don't like the hyperbole there, then suffice to say his Janus Henderson Global Unconstrained Bond Fund suffered its worst day since inception as the spread between U.S. Treasury yields and German bund yields ballooned further to fresh multi-decade wides.


I postulated that he (Gross) had some short volatility positions on as well that might have hurt him. And while that was pure speculation, it's worth noting that this kind of volatility is the stuff of nightmares for VaR-sensitive investors whose position sizes are a function of volatility.

So where am I going with all of this? Well, without extrapolating too much into the motives for that large bullish bet on Treasurys mentioned here at the outset, we've seen at least three instances over the past three weeks of a dramatic drop in Treasury yields (the plunge that accompanied the Italian bond selloff, the quick drop that coincided with the BRL plunge and a sharp move lower last Friday morning when a headline about Apple hit around 4:00 a.m., New York time).

These moves seem to underscore the contention that thanks in no small part to modern market structure, Treasurys are prone to rallying sharply (yields careening lower) on bad news and if you go up and look at that crowded spec short, it is entirely possible that one of those "flash rallies" could trigger a short squeeze. In fact, it looks like that's already happened a couple of times over the past several weeks.

Commenting early last month in a note ostensibly written to explain why the stretched positioning in USTs wasn't/isn't actually as extreme as it looks, Goldman added the following caveat:

Momentum-driven swings in aggregate positions can significantly amplify yield movements, over and above the levels required to price in new fundamental information and for investors with tight stops, even mild shifts can be consequential.

So the question becomes whether a safe haven bid (i.e., Treasury buying) catalyzed by something like a liquidity vacuum in the Brazilian real or a bad morning for Italian bonds could end up being exacerbated by an unwind in short positions and exaggerated further by HFTs. I explicitly discussed this over the weekend on this platform. Here's what I said:

When algos see something like these panic bids for Treasurys, they're likely to chase down the rabbit hole, potentially exaggerating the move or worse, pull back altogether on the assumption that something is going on that they, by virtue of being algos and not humans, aren't capable of understanding. That latter point sounds funny, but lest you should think I'm just making it up, do note that it's a real phenomenon. You can read more than you would ever care to know about HFTs and "adverse selection" here.

The link at the end of that excerpt is to a lengthy post on my site documenting a recent Goldman note on market fragility, liquidity provision and the "adverse selection" problem for HFTs.

Well, Goldman was out on Tuesday with a new installment of their "Top Of Mind" series and it's an expanded version of that discussion. Guess what gets mentioned? If you said the recent turmoil in Italian bonds, you win a blue star. Here's the bank's Charles Himmelberg, to explain:

The recent political headlines in Italy are a good example. In circumstances like these, HFTs appear to recognize that their algorithms cannot understand or process the information as well as human traders. This puts HFTs at a disadvantage since it exposes them to adverse selection by those better-informed buyers or sellers. In response, it only makes sense for the HFTs to widen out their bid-ask spreads, perhaps dramatically; in some cases they withdraw from the marketplace altogether.

Here's where I get to remind you that generally speaking, I don't just pen these long missives from no reason. Recall this excerpt from another post I wrote for readers here late last week:

Thin liquidity last Monday (New York and London were on holiday) contributed to choppy price action in Italian assets and then when everyone came back online Tuesday morning, traders just chased it all lower, prompting dealers to pull quotes and leading to an explosion in bid-asks.

(Heisenberg, Bloomberg)

That's precisely what Goldman's Himmelberg is talking about in the bolded passage above and he reiterated it in a separate note on Wednesday that includes the following color and accompanying visual:

Maybe [the] spike in BTP yields was not as systemic as it seemed? With the benefit of hindsight, the moderate degree of spillovers and rapid recovery of systemically related markets leads us to doubt that the May 29 sell-off was as fundamentally motivated as it initially appeared. Instead, consistent with the widespread media reports of extremely poor trading liquidity on the day of the sell-off and shortly thereafter, we suspect that the magnitude of the sell-off may have had as much to do with a collapse in trading liquidity.

Exhibit 1 shows evidence that seems to support this thesis. Not only did the front end (1-3 year) of the Italian bond market experience its largest one-day spike in recent history (top chart), it also experienced its largest one-day spike in bid-ask spreads (bottom chart).

Do note that while you could try and draw a parallel with the eurozone sovereign debt crisis, you wouldn't be entirely correct, at least if what you were aiming to do was argue that this kind of thing happens naturally when people are concerned about a possible deterioration in the fundamental outlook. As Himmelberg goes on to write, "the one-day spike in both bond yields and bid-ask spreads was larger than at any time during the European sovereign debt crisis in 2012."

That, in turn, echoes the sentiments of former Goldmanite Robin Brooks (he's now Managing Director and Chief Economist at the IIF) who, just hours before the bottom really fell out for Italian bonds, warned that it was the speed of the selloff (yield rise) that was concerning.

All of this points to an increasingly fragile underlying market structure. Liquidity is abundant under normal conditions, but it evaporates instantaneously when non-carbon-based lifeforms (i.e., HFTs) cannot discern what the proximate cause of a given move is.

This is arguably exacerbated by ETFs and other passive flows. Himmelberg takes a more benign view of their role in this equation, but as regular readers know, I do not (take a benign view, that is). Here's a brief excerpt from a longer piece I wrote on Tuesday night:

Passive flows combined with what Howard Marks has described as a “perpetual motion machine” dynamic, have created a self-feeding loop, encouraged by the post-crisis monetary policy regime which kept volatility anchored and transformed “BTFD” from a derisive meme about retail investors into a viable (indeed, an almost infallible) trading “strategy”.

If that self-feeding dynamic ever slams into reverse, there are questions as to the durability of the mechanisms that underpin it (e.g., Will the ETF model actually crack, as it did on August 24, 2015? And what about credit ETFs? What happens to HY and EM bond funds when the underlying liquidity mismatch is exposed?)

This is all part and parcel of the same problem: the current market structure is largely untested and every time it gets a dry run, it fails or at least cracks.

It cracked in February, it cracked (and nearly broke) on August 24, 2015, and frankly, it cracks all the time as demonstrated in the following helpful infographic from one of the Goldman pieces mentioned above:


So while I have no way of knowing what the actual rationale is that motivated someone to make a $75 million bet on 10-year Treasury yields falling to 2.60% within the next two months at a time when everyone seems to think yields are destined to keep rising in light of the rather poor supply/demand dynamic created when you pair Fed balance sheet rundown with massive Treasury supply, what I would say is that the combination of a short squeeze and a "flash" rally like the three we've seen over the past three weeks could make those bets profitable in the blink of an eye (literally).

The problem for everyone else with this setup is the same as it ever was. Exaggerated moves spill over. Sharp rallies in Treasurys ostensibly signal that something is "wrong" and that informs (or actually, "misinforms" is better) trades in other assets. I'm going to use the same quote from Citi's Matt King here that I used in the Dealbreaker post linked above. To wit, from a note dated June 1:

While the specific shocks may have been unpredictable, we don’t think they were unforeseeable. The vulnerability of a complex system – like global markets – owes more to the interconnections between its elements than to any one element individually [and as such] investors should not be looking for the grain of sand which causes the avalanche, or at the match which causes the forest fire, but rather at the height of the sandpile or the dryness of the forest more broadly.

This might all seem rather amorphous or otherwise hard to wrap your head around, but it's highly relevant for all of us. Just look at that Goldman infographic. It's not like they're talking about esoteric assets that no one cares about or markets that traditionally lack depth. You've got the S&P, U.S. Treasurys, cable and the VIX - not exactly "obscure".

And look, even if you don't remember those episodes and even if you want to pretend like you weren't affected (directly or indirectly) by what happened in Italian bonds a couple of weeks ago, you most assuredly were affected by what happened in early February and that episode was yet another manifestation of modern markets failing a stress test. Remember, part of why equities reacted like they did during the week of February 5 was because systematic strats (a fixture of modern markets) were forced to de-risk to the tune of some $200 billion. And that was itself caused by the realization of the rebalance risk inherent in leveraged and inverse VIX ETPs, another "miracle" or modern markets.

So what does this mean for you? Well, as usual, if your investment horizon is "forever" (or a period of time that might as well be forever), then this is just an academic exercise which, to the extent you learned something from it, was well worth your time.

But for anyone who wants to be honest with themselves and admit that if your investment horizon was "forever" you probably wouldn't be spending a lot of time on investment-related websites, then what you might consider taking away from the above is that, as Goldman puts it, "markets themselves" are now the biggest risk to, well, to markets.

That means you should steel yourself for more instances where price action seems to be exaggerated and detached from fundamental drivers.

That's obviously relevant to anyone who cares about managing open positions (see the reference to "tight stops" above) and more broadly, it's especially important for asset managers running models that assume the type of subdued volatility that's persisted since the dawn of ultra-accommodative, post-crisis monetary policy.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.