🍔We've Seen This Downturn Before
Big portfolio changes to defend against the market selloff
Massive selloff in SPX this week. If you've been following our blog, our plan was to move to cash as the market broke key support levels. This saved our portfolio from a lot of damage. Our call to buy bonds has also played out well. What’s the next phase and when do we step back in? Read on to find out.
The Mag 7 posted its worst week of 2025 so far, down -5%. Nvidia tanked -9% after its earnings and Tesla is down -40% from its highs. We predicted the -5% decline in SPX from the highs, which it reached on Thurs. Could it go any lower?
Changing face of Mag 7
Mag 7 stocks ended the week down -5%. Readers will know we’ve been hounding on Mag 7 since last December (yes, right at the peak). Obviously we didn’t predict DeepSeek, but it didn’t matter — our thesis of falling earnings, rising capex, and intensifying competition took these stocks down more than 10%. The market is coming around to our view that the structures of these companies have changed.
It used to be that each of the Mag 7 (known as FANG back in the day) had their own monopoly that didn’t cross each other’s turfs. Facebook was in social media, Amazon was in e-commerce, Netflix was in TV, Google was in search. Attempts to enter each other’s businesses failed (remember Google+?). When you have a monopoly in a high-margin business like software, you get lucrative shareholder value which allowed their stocks to take over the world.
Nowadays, Mag 7 all compete against each other. Meta, Google, Amazon, Apple, Microsoft all sell ads in some form. Microsoft, Amazon, Google compete in cloud infrastructure. Nvidia, Amazon, Google, Meta all have their own AI language model and fight to acquire chips. There’s more examples (consumer electronics, video streaming, quantum computing, self-driving cars) but you get the point.
Each of those markets were too small to sustain the hypergrowth of these juggernauts, so they had no choice but to encroach on each other. Now they have to outspend each other by the billions in what we called an arms race or “winner’s curse”. Add in the new formidable competitors from China and you have a recipe for rapid commoditization across chips, AI, cars, electronics, e-commerce, social media, and software.
Then there’s the capital intensity issue. You see, software and internet businesses are asset-light and scalable. That means they don’t require much investment to grow. On the flip side, infrastructure businesses like AI data centers are expensive. Especially when the internal components become obsolete as soon as they’re installed. And all Mag 7 companies want to build the same thing, data centers. A capital-intensive business is worth far less than a capital-light business.
This cycle of increasing competition, shrinking margins, and rising capital requirements will spell the slow death of Mag 7 stocks. We are currently watching that realization hit. It will only get harder to justify their extreme valuations.
Overbuilt AI infrastructure
Analysts reported that Microsoft is cancelling leases for a significant number of data centers, while pulling back signed commitments on new leases. This is the same tactic that Meta used when canceling the Metaverse program. MSFT is realizing that they, along with the rest of Mag 7, have built too much computing capacity.
Microsoft did attempt to downplay the concerns by reiterating its $80B infrastructure spend, but this rang hollow.
We think this is related to the divorce from OpenAI, which we covered before. As language models become cheaper and commoditized, there is less strategic value to monopolizing OpenAI. They can get the same outcome for less investment. The tricky part now is figuring what to do with all of the excess capacity, which is not able to be repurposed to their Azure cloud business.
We now wait for the other major cloud providers to follow suit in scaling back expansion plans. It will be an awkward admission that the CEOs wasted hundreds of billions of dollars, and that DeepSeek really did change the economics of AI. This could be bad news for Nvidia and other chip and server providers.
DOGE is killing the economy
Investors thought Trump/DOGE would be good for business since taxes would be cut and inefficiencies eliminated. Instead, the eliminations are undermining economic activity. For example if you cut government service contracts, the American businesses that provide those services lose money and lose trust in future deals. They have no choice but to downsize their workforce. Similarly, cuts to regulatory bodies and research grants reduce confidence and spending.
DOGE-driven fiscal contraction was a key pillar of our growth slowdown thesis. Estimates suggest that 300K federal jobs will be cut, along with 600K private contractors. So we could potentially see 1M more layoffs in a country with 7M unemployed. This could have devastating impacts on spending and savings. Let’s see if jobless claims spike more meaningfully in the coming weeks, but so far credit risk remains benign:
The continuous stream of political and business threats from the President is causing uncertainty around policies, trade, government funding, and the labor pool. The Sales Managers Indexes recorded sharply falling sales growth and hiring expectations from just a month ago:
Trump has acknowledged that there will be short-term economic pain as a result of his policies, particularly tariffs. In other words, we should not expect the “Trump put” anytime soon.
Macro continues to deteriorate
US econ data continues to get worse. This week we got a dismal consumer confidence survey, new home sales falling, and jobless claims rising. Jobless claims might be a small sign that DOGE cuts are starting to bite. We wait a bit more evidence, but secondary data points (e.g. the employment subcomponent of services PMI, or the jobs question in consumer confidence survey) are corroborating the slowing jobs market.
Here we make a distinction between soft and hard data. Soft data is based on surveys of how consumers or business managers feel. Hard data is based on quantifiable metrics like sales data and goods orders. Soft data has been wonky as it is muddled by politics, sample sizes, and response rates. These have nothing to do with the real economy. Last week we showed this chart of consumers’ inflation outlooks being tied to political affiliation:
Unfortunately that’s the nature of humans, and is a drawback of using survey data. Consumer confidence, business outlook, and purchasing managers’ index (PMI) are all survey based. Needless to say, with the headline barrage we’ve been getting lately, surveys haven’t been great.
The hard data is weakening but not disastrous. For example, orders for durable goods (a.k.a. big ticket items like cars and appliances) are still trending up. Ultimately, it’s the hard data that will tell us if what we’re experiencing is a mere growth scare or full-blown recession. Regardless, the market has started pricing-in recession odds, hence this week’s selloff:
Tariffs looming large
Tariff threats were in full force this week. Trump announced additional 10% tariffs on China, on top of the existing 10%. He reiterated the 25% on Canada/Mexico to start next Tues. Any retaliatory tariffs by trading partners would be answered in kind. He also unofficially announced 25% tariffs on European cars and “other things”.
Bear in mind, the first round of tariffs were extended at the final hour, and Trump did leave room for another extension or deal. We continue to think the administration wants to do deals; comments by officials in Mexico and Canada are optimistic, and Europe is scrambling to increase defense spending.
China, too, is open to discussions. So far they have responded by expanding tariffs on US imports and tightening critical mineral exports. The markets think this is a “clash of the titans” scenario but in reality, Trump has stated he would rather avert a clash. Trump wants to stop fentanyl from entering its borders, which should be an easy way to reach an agreement (or at least the start of one).
There was a notable pickup in CEOs’ mentions of tariffs during earnings calls. Management teams say they don’t want to invest until there is certainty around tariffs. There was little talk about reshoring manufacturing capacity, and more about frontloading imports. Therefore, the economic boost from reshoring might not materialize as expected.
A study by Goldman Sachs shows that companies with broad exposure to tariffs and/or high sales to Canada, Mexico, and China have already reduced capex expectations for 2025. Time will tell whether this reduced impulse begins to hit the economy.
We’ve seen this downturn before
As mentioned last week, we are currently in the seasonally weak period of Feb which lasts into early March. It’s not a coincidence that the selloff started on Feb 21, SPX options expiry day. That leaves the SPX unpinned and vulnerable to negative catalysts. Recall that the 2020 Covid selloff also started after Feb expiry — on the 21st, literally the exact same date.
The March seasonal flows look like this:
We were expecting a moderate 5-10% pullback from peak to trough, and we entered that band on Thurs. Our analog is 2022 — below we overlay the price action that led to a 12% decline. In both cases, SPX rallied strongly through Nov before consolidating in a wide range for two months. Then in the new year it teased a break to the upside, only to reject strongly and ignite the selloff. This kind of pattern recognition is helpful for timing, but remember, it’s not exact.
Selloffs like these rarely go straight down. They feature fierce “dead cat bounces” that trap hopeful buyers. This was evident in the March 2020 Covid crash as well as the Aug 2024 growth scare. We await technical confirmation before re-entering the market. A test of the Jan lows and 200-day moving average at ~5750, is not impossible, but we will have to wait and see how next week unfolds.
Volatility signals
We previously introduced volatility indicators Voldex, Taildex, and Skewdex that tell us what the market is thinking through the lens of options curves. They show much more nuance than simply using VIX. While Voldex is picking up the increased volatility as expected, Taildex and Skewdex are muted. This means there is no clamoring for “crash hedges”.
In other words, this is not a panic-driven selloff, but rather a methodical deleveraging. Whereas the Aug 2 and Dec 18 selloffs were panicky, hence they rebounded quickly, this one might take some time to work through. It appears investors already bought put protection in anticipation of this event, perhaps because of the long two months of whipsaws we just went through.
The reason for the deleveraging is the market is revising the outlook of certain momentum stocks, including Mag 7. In light of slowing earnings growth (evidenced by Nvidia’s Q4 results) and capex over-investment (made apparent by DeepSeek), companies’ earnings outlooks are being questioned. That’s why the Mag 7 are down -15% from the peak, with stocks like TSLA down -40%.
One final thing to point out relating to volatility is the gamma level (purple bars below). Recall that gamma tells us how market makers must react to up or down moves in SPX. And right now, market makers are amplifying moves in both directions, causing jerky selloffs but also violent bounces. So be prepared for this heightened volatility.
Portfolio changes
As we alluded to last week, we trimmed our portfolio when SPX broke below $6000 (50D moving average) and again when SPX broke below $5925 (100D moving average). We are now 60% cash, 40% SPX. We don’t intend to sell the remainder to cash unless SPX is breaking the 200D average and thus entering recession territory.
A couple of things need to happen to get us back into the market:
Wait for seasonality to improve in early March
Bottoming pattern such as hammer candles or higher lows
Oversold conditions in Put/Call ratios and investor sentiment
Breadth thrusts and follow-through signals fire
Alternatively, a stop-buy near the all-time highs.
The market is shifting to a defensive posture and imputing the odds of a growth slowdown or recession. We think this is a garden-variety pullback rather than a full-blown recession. Thus the opportunity to buy the dip should emerge soon. Hang in there.
That’s all for this week!
Thanks for being a loyal subscriber. If this post helped your portfolio, please share it with someone else!
Disclaimer: This publication is for educational purposes only. The authors are not investment advisors and nothing here is investment advice. Always do your own due diligence.
Great write up. To be clear, did you liquidate your QQQ position, as part of your going to 60% cash this week? Thanks