There’s elegance in simplicity.
Analysts, strategists and market participants of all shapes and sizes spent a lot of time last month attempting to explain a mild swoon for US equities, which stumbled in the face of higher yields and… well, let me stop there before I launch into just the sort of belabored explanation I was in the process of suggesting isn’t necessary.
“The near -5% pullback in August looks typical,” Deutsche Bank’s Parag Thatte wrote, in a weekly asset allocation update. If you’re familiar with Thatte’s weeklies, you know he’s fond of reminding investors that shallow selloffs historically occur every two or three months — it’s not necessarily something to be alarmed about, but it is something you should expect.
As the figure below shows, the length of time elapsed between 3% selloffs ranked in the 93%ile looking back seven decades.
“A pullback was overdue,” Thatte flatly remarked.
The question now, obviously, is whether the “worst” is over. That’s a somewhat laughable way to contextualize the situation, hence the scare quotes. There really is no “situation.” Stocks are doing fine. Indeed, they rebounded smartly during the last week of August. Yes, risk was on the back foot again as September kicked off, but it’s fair to say things are going swimmingly under the circumstances in 2023.
So, a better question looking ahead goes something like this: The S&P trades at 19x (17x if you exclude the “Magnificent 7”), real rates are higher than they’ve been in decades, there’s a lot of geopolitical angst and although equities are currently trading in a “bad news is good news” regime, that script can flip quickly in the face of evidence to suggest a recession wasn’t averted, just fashionably late.
Thatte noted that pullbacks of over 5% generally happen every four or five months, and it’s been seven since the last one. “A modest extension of the current pullback would not be surprising,” he wrote.
And yet, recall that equity positioning on Deutsche Bank’s aggregate measure is now back to neutral. Discretionary investors retreated after begrudgingly taking up their exposure over the course of the summer melt-up.
So, it could go either way: There’s room for de-leveraging, but there’s also scope for fundamentals-based investors to get forced back onto the stock bandwagon (Cue the only memorable moment from the Godfather III: “Just when I thought I was out, they pull me back in!”)
It’s all about the incoming data. “Surprises hold the key,” Thatte said, on the way to noting that consensus sees a meaningful deceleration. Again.
The figure on the right shows how long economists and strategists have waited for a slowdown that so far hasn’t bothered to show up.
The implication: There’s scope for the economy to defy expectations for another quarter or two, and that could put a floor under equities.
As Thatte put it, “With the consensus expecting a sharp slowing in growth next quarter, as it has expected every quarter this year, the bar is not very high.”
If we believe that AI fever and Nvidia in particular are responsible for much of the NASDAQ rally we’ve enjoyed, this is worth a read.
Perhaps demand may accelerate further, but that ain’t worth scat if supply cannot be ramped up to fill it. Or for any of the competitors like AMD who also 100% rely on TSMC for all of their advanced logic chips.
Note this is from the chief of TSMC itself.
https://www.tomshardware.com/news/tsmc-shortage-of-nvidias-ai-gpus-to-persist-for-15-years
18 months is an eternity in the tech world.
Derek-
I copied this from another source- it sounds like NVIDIA isn’t sitting there waiting on TSMC to meet their demand.
According to a report from Taiwan’s Commercial Times, NVIDIA is aggressively establishing a non-TSMC CoWoS [advanced packaging] supply chain. Sources in the supply chain reveal that UMC (UMC) is proactively expanding silicon interposer capacity [the nuts and bolts of advanced packaging], doubling it in advance, and now planning to further increase production by over two times. The monthly capacity for silicon interposers will surge from the current 3 kwpm (thousand wafers per month) to 10 kwpm, potentially aligning its capacity with TSMC’s next year, significantly alleviating the supply strain in the CoWoS process.
Thanks. It’s inevitable. But “Sometime next year”? The devil’s in the details between announcements and commercial production amounts.
That said, Nvdia has long been less committed to soley using TSMC than some of their competitors. Kudos to Nelson.
Just remember that every month of delay keeps the doors open to many competitors to launch comparable or better products.
For instance: https://www.techradar.com/pro/microsoft-backed-ai-startup-beats-nvidia-h100-on-key-tests-with-gpu-like-card-equipped-with-256gb-ram
H
I know you know this but I’m surprised to see an “expert” say this or that happening is “overdue.” That’s what sportscasters and amateur gamblers say. I heard two very good announcers backpedaling like mad when Texas was messing up Nick Saban’s day yesterday. Nothing is due or overdue in this situation. When a continuous distribution of outcomes, which is a sample from a large universe, is the set containing a great many values, the probability of correctly forecasting an inflection point is easy. It is one divided by infinity, or what the little boy shot at, nothing. This distribution, however compelling it seems, theoretically contains an infinite number of points, each with probability of occurrence of zero. Inflection points have the same probabilities. This distribution is the same as the one describing the time for a broken clock.
Ehh, not really. Stocks can absolutely be “overdue.” Maybe not in a strict mathematical sense, but you can pretty much take it to the bank that if you get up there in the 90%ile and above in terms of how long it’s been since a minor pullback, there’s going to be a pullback sooner than would otherwise be the case. It’s a function of a lot of things, including sentiment, but also stretched positioning and vol dynamics where you get setups that make a ~3% correction all but certain due to — for example — a pile-up of unstable equity exposure with vol-sensitive systematic strats that would get mechanically de-risked on even a small range / vol expansion. At various intervals last month, pretty much any movement in the S&P (so anything other than an unchanged session, including rallies) had the potential to trigger vol-control selling sufficient to pull the rug out. As for the apparent college football parallel, I’m afraid I can’t offer much — I haven’t watched sports on television since the late-90s.