Almost exactly a year ago, we solved what at the time was a mysterious meltup in stocks (S&P rose over 6% from mid-Jan while bonds and the dollar did practically nothing). It was the slow-at-first-then-all-of-a-sudden collapse of a little-known, multi-billion-dollar futures fund whose apparent specialty was 'picking up nickels in front of steam-rollers' by selling upside-ratio-call-spreads.
We explained at the time that the firm was basically buying at the money options and funding those purchases by selling even more out of the money options (this is a simplified view of the fund strategy but close enough for examining whether it could have impacted markets).
The strategy is (notably simplified) a bet that pays off if the market drifts higher amid calm volatility (and ends above the lower strike and below the upper strike).
If, however, the market should rally aggressively beyond the upper strike, the initially-long position begins to 'turn short' and requires significant hedging/unwinds to maintain any semblance of control.
As we noted at the time - it appears that is what happened in Feb 2017:
The trade was going well, until the S&P rose above 2,300.
At that point the "convexity seemingly ‘kicked-in’ as witnessed by market participants, the short-gamma ‘take’ since has been nothing short of astonishing."
Charlie McElligott (now of Nomura) said at the time:
I’m worried that this stock ‘melt-up’ move is extraordinarily mechanical right now - almost entirely the aforementioned forced-covering, not high conviction induced-buying - and may be sending a “false signal” which is potentially dragging-in new buying on the breakout to new highs.
While the concern at the time was whether it was possible that a single small fund could bring down the mighty US equity markets; as Peter Tchir noted at the time, in all likelihood this particular fund is just a relatively public example of a more widespread strategy - a strategy that was getting hit across the board.
I am more willing to believe the argument that this fund was just one of many funds trading this strategy and that everyone employing this strategy was hit by the same combination of factors and that this widespread unwind was driving the market.
I want to believe that view, because the alternative, that liquidity has devolved to the point that a relatively small and formerly obscure fund can drive the entire market for days on end is quite scary as both a trader and investor.
But one glimpse at the fund's performance and 'coincidental panic-buys' in stocks suggests that Catalyst - the fund's name - was indeed the catalyst for those odd melt-ups (or at least was the most evident symptom of a strategy that had become dominant in markets).
Fast forward a year and look at what happened in January...
Another sudden unexplained stock market meltup - talking heads explained it all away as 'normal': earnings, tax reform, infrastructure, goldilocks, etc., etc., but just as we saw a year ago - stocks and VIX decoupled completely as that ratio-call-spread went short and forced-buying sent stock prices higher (and vol higher as the sold upside calls were bought back).
And right as the equity market was melting up in January - grabbing every headline with its irrational record-beating ramp - so a managed futures fund was starting to suffer (as its panic'd gamma squeeze sent its NAV notably lower, despite its initially bullish options position).
It won't last right. It can't. Someone will do something. But it did and it appears that Catalyst finally threw in the towel their month-end statement arrived (and likely with it some margin calls). That liquidation event just happens to have coincided with someone needing to buy vol at any cost (to exit their position), crashing XIV - triggering its termination event, and taking the Dow down 1600 points (twice).
So is Catalyst (or its strategy being so broadly-followed as to become 'the market') the culprit again? We shall see, but we are sure the denials will be first.
* * *
Update: Additional sources suggest the fund was also trading the vol of the index against the vol of the individual names (implicitly betting on lower implied correlation)...
That worked for a while...until it didn't...