You wanted more Charlie McElligott, you got it.
On Tuesday afternoon, just after the S&P took a sudden lunchtime dive, Charlie sent out a client blast that put at least part of the blame on CTAs.
We documented that here as soon as it started making the rounds in a quick post called “On Wall Street: Midday Massacre“.
“Our CTA Trend model is again deleveraging massive notional in ‘long US Equities’ expressions across SPX, RTY and NDX”, McElligott wrote, adding that the bank’s “SPX model was triggered to sell down at 2763 with $32.8B notional for sale, reducing from “+100% Max Long” down to “+65% Long”.”
That quick note from McElligott quickly became the talk of the proverbial town. Other strategists subsequently weighed in, suggesting that according to their models, CTA selling likely wasn’t the cause of sudden drop in equities. By Tuesday evening, the debate had reached a fever pitch.
Read our full coverage of the Tuesday rout
As we pointed out pretty much immediately, a simplistic look at the SG CTA index’s beta to SPX certainly seemed to suggest that trend followers weren’t “Max Long”, as Charlie suggested.
Fast forward roughly 24 hours and McElligott found himself starring in a Dani Burger special called “Quant Blame Game Over Stock Sell-Off Pits Nomura Against Nomura.” You can read that post for yourself, but suffice to say Charlie’s colleagues in Tokyo didn’t necessary agree with his assessment of CTAs’ culpability in the Tuesday bloodbath.
Well, as you might expect given last night’s S&P futures crash, McElligott is out on Thursday with a brand new missive and let me tell you what, he is in rare form.
First of all, here’s a quick look at S&P futs with Tuesday’s lunchtime dive and the Wednesday evening flash crash highlighted just to set the stage.
McElligott kicks things off by documenting the events that contributed to the Wednesday night plunge.
“Clearly the shocking (and curiously delayed) news of the arrest of Huawei’s CFO (and daughter of the company’s founder) by US request in Canada on Dec 1st over potential violations of US sanctions on Iran blasted risk-sentiment out of the gates last night, as it only further complicates the already incredibly fragile state of US / China trade negotiations, on top of the assorted risks across slowing global growth- and inflation- data, OPEC and Brexit uncertainties which are paralyzing investors”, he writes, before delivering his take on the flash crash.
You might recall how we opened our post on Wednesday evening. To wit, from “What’s Behind The Flash Crash In S&P Futures?”:
And now, a moment of silence for broken markets.
Charlie echoes that sentiment almost verbatim.
“More proof of ‘broken / un-tradeable markets’ at a time into year-end where there is already a limited incentive to deploy risk due to brutal YTD performance dynamics and ugly seasonal illiquidity”, he writes, before delivering the play-by-play (and we’ll just go ahead and give you the extended verbatim quotes here):
Wednesday’s modest +0.6% rally in S&P in the abbreviated US Equities futures session was based off of confirmation of confirmation of Chinese movement on positive trade-related matters, as the Chinese Ministry of Commerce signaled action on specific items where there is consensus—i.e. preparing to restart imports of US soybeans and LNG as an obvious risk-asset +++
But after we saw this Huawei news hit, it seemingly (and inordinately) contributed to an almost impossible 65 handle drop (~1.9%) in Spooz on the overnight reopen, with 36,700 contracts trading in the first 10 minutes (vs the usual 1000), and triggered more than 40 unique “velocity logic” curb events (a CME tool to prevent severe price swings) in the first six minutes of trading.
All of that was documented at length here on Wednesday evening and one thing we would quickly note about Charlie’s take is that he expresses the same sentiment we did when it comes to whether that Huawei news “should” have caused that dramatic a move. He calls it “inordinate”.
“After the gap-lower inevitably hit more US Equities stop loss limit orders and further bludgeoned trader sentiment, the modest recovery thereafter lost further steam over the very early US hours”, he goes on to document on the way to underscoring how Fed odds are now completely capitulating. EDZ8-9 is now at a cycle low and EDZ9-0 is increasingly pricing in easing in 2020.
And then Charlie delivers what everyone was waiting for. The granular details of what exactly happened from Monday through Tuesday, including his take on how upside G20 hedges were quickly monetized, flipping the script on the directionality of the gamma effect on the way to triggering stop-losses for CTAs, leaving them “Max Long” headed into Tuesday. Here it is:
A point I’ve tried to hammer home over the past week is this: once those few brave macro funds who were fortunate enough to have performance in order to play for ‘Equities Upside’ around Powell / G20 event risk then began to monetize their trades immediately out of the gates Monday, in-turn created an avalanche of both large notional delta for sale as well as outright futures selling which clearly outsized buy demand flows
This “real” flow from fundamental / discretionary universe into very poor year-end liquidity then conspired with the latest cross-asset confirmation of the “end-of-cycle” growth-scare which picked-up Monday and Tuesday—this time from US front-end inversions across a number of curves and “corroborated” the slowdown / recession-type pricing which has already been made evident in US Equities “cyclical” sectors over the past few months
As monetization then triggered simple ‘stop-losses’ in light of the rapidly deteriorating risk-sentiment, we then hit the selling levels for the recently re-accumulated “Max Long” in SPX from CTA trend universe (who added $22B last week and another $16B+ on Monday this week to start Tuesday “Max Long”)
In light of the well-flagged “tight ranges” for systematic CTA buy- and sell- trigger levels from this year’s “chop trade” across the spectrum of horizon windows we track, this meant that despite again turning “Max Long” that we remained dangerously proximate to “deleveraging” levels which would in-turn “flip” both 3m- and 1m- horizons back from “long” to “short” around 2764 (cutting from +100% Long” to 82.5 to 65 to 47 to 30 to ultimately just 22% long as of the Tuesday close after clubbing through 2711).
And listen, if you thought Charlie wasn’t going to mount an ostensibly trenchant defense of all the pushback he got from his Tuesday CTA call, you were wrong.
After detailing exactly what he thinks transpired on Monday and Tuesday, he says he “received a number of pass-alongs notes from around the Street which questioned the validity / accuracy of our CTA model ‘deleveraging’ call from Tuesday”.
He then reminds everyone that from where he’s sitting, his model “again ‘nailed’ the S&P futures level where the market would come under significant notional selling pressure as well as Russell and Nasdaq trigger levels.”
McElligott goes on to put part of the blame for the pushback on what he’s characterizing as a “game of telephone” on the Street (that’s a reference to a children’s game where the original message gets distorted as the intermediaries proliferate). In order to avoid perpetuating that dynamic, I’ll just quote Charlie directly again:
There were further mis-categorizations / inaccuracies on a number of fronts which need to be highlighted:
- Misinformation as to the notional size of the selling which we estimated vs what was reported
- Inaccuracies of the AUM scale of the CTA universe and position sizing / leverage allocation therein
- Whether this “trigger” was a “deleveraging” of a long (size reduction of the long which it was) vs outright “shorting”(which was misreported / misinterpreted by some)
- Omissions / lack-of-context surrounding macro- and positioning- / performance- catalysts which I’ve been documenting in recent notes that actually “kicked off” the move to said trigger levels
- A general lack-of-awareness as to the make-up of the model–i.e. that we incorporate 2w, 1m, 3m, 6m and 12m windows to capture the broad-spectrum of CTA lookback periods across the trend universe
- No context as to the incredible accuracy of the model—not just via the success that the tool has had in identifying “market inflections,” but with regards to our CTA replication model’s incredible track-record vs benchmark
And he wasn’t done. Not by a long shot. Having thus addressed what he feels are rampant mischaracterizations and inaccuracies, Charlie then says that when you take a step back and consider what it is he’s asking his model to do, it (the model) is “brilliant.” Take it away, Charlie:
Considering the incredible task of prognosticating 58 unique cross-asset futures contracts as our QIS team’s model does—and inherent requirements of trade direction / sizing / leverage of course—the index replication model is brilliantly accurate on performance-matching vs the index:
Since 2016, the model has exhibited an average deviation from benchmark of just 54bps, with a median deviation from benchmark of 48bps
Over the incredibly volatile last 6m both from a cross-asset realized volatility- and CTA performance- perspective, the average deviation of the replication model from benchmark is just 56bps, with the median deviation from benchmark being 58bps.
In the absence of viable alternative theories being presented from elsewhere with regards to the extreme and ‘price-insensitive’ -action and explosive futures volume on the sell-down—as well as an awareness that many competing CTA models on the Street grossly oversimplify to just a few of the larger asset futures and apparently do nothing more than incorporate a “VWAP breakeven” and / or simple “moving-average” logic—I feel more confident than ever in our TRANSPARENT model—especially as we alone have our performance historically tracked versus benchmark.
And there’s more, if you can believe it, but I think you get the point.
To say McElligott is confident in his methods (whether “methods” means his models or his exceptionally unique delivery style when it comes to writing his daily missives) would be a late candidate for understatement of the year.
This is a man who is “ten toes down” in the game (if you’re up on your street slang).
Finally, for the icing on the proverbial cake, McElligott includes a series of screengrabs with key de-leveraging/re-leveraging levels for nearly a dozen assets and in the interest of brevity, we’ll just include the S&P table because given the events of the last 48 hours, that seems appropriate.
One final thought: We are a little disappointed that McElligott didn’t release this last night when futs were collapsing…
Charlie McElligott is probably doing about 95 on the Turnpike right now trying to get to his desk so he can send out a client blast about what's going on with futures pic.twitter.com/iDMOnlYyRO
— Walter White (@heisenbergrpt) December 6, 2018