Nvidia's AI Chip Dominance Sparks Global Race for Alternatives

The meaning of the market sell-off

David Solomon, the chief executive of Goldman Sachs, is not normally found walking the bank's trading floor. Monday, however, was an abnormal day for many on Wall Street. Solomon headed down to the fourth floor of the bank's Tribeca headquarters as its traders grappled with one of the most chaotic days of market action in recent years. He was not the only senior figure stalking the front office. Ashok Varadhan, the co-head of Goldman's global banking and markets business, was spending much of the day in contact with the team that traded securities tied to the Vix -- the volatility index ubiquitously known as "Wall Street's fear gauge". Upstairs, the group's wealth management practice was hosting a call for more than 5,000 investors answering questions on the likelihood of a recession, how weak US economic data had caught markets off-guard, and the hypothetical market impact of a war in Iran. Downtown, officials and traders on the floor at the New York Stock Exchange were discussing whether circuit breakers would force a marketwide trading halt for the first time since the outbreak of the coronavirus pandemic. By the end of the day, almost 90 per cent of stocks on the MSCI All-Country World Index (ACWI) had fallen in an indiscriminate global sell-off. Nvidia -- the chipmaker that had single-handedly driven almost a third of the US stock market's gains in the first half of 2024 -- shed around $400bn in market value in the space of a few minutes, before adding most of it back in the next few hours. Yet within a few days, most of the turmoil seemed to have been forgotten. By Thursday evening the ACWI and S&P were both down less than 1 per cent for the week. On their own, the whipsaw swings of the past week-and-a-half say more about the psychology and structure of modern markets than they do about any fundamental shift in the economic or financial outlook. But the moves did not happen in a vacuum. For some market veterans, the real aberration was an extended post-pandemic period of steady market moves. When this week's events are combined with other cracks that began emerging over the past month, there are signs of a longer-term shift that could lead to a period of increased volatility after years of unusual calm. "The market was so certain there would be a soft landing in the US, that there was complacency that any other outcomes were even possible," says Joe Davis, the global chief economist at Vanguard. "There was so much concentration, too many investors and market participants all having the same view of the world . . . and that view was really warm and fuzzy." Now, he says, there has been a "repricing of that thinking". Most observers believe the sheer scale of the moves over the past 10 days was out of proportion to the initial triggers. The immediate spark for the sell-off was a pair of economic updates on the first two days of August -- a survey of manufacturing companies, followed by official figures on the state of the labour market -- that heightened concerns that the US economy was heading for recession and the Federal Reserve was moving too slowly to cut interest rates. The jobs data in particular was well short of expectations, showing the US economy had added just 114,000 new jobs in July compared with expectations of around 175,000, but the figure was not even the worst result of the year. An initial sell-off spiralled out of control when Asian markets had a chance to respond on Monday, as the bad news about the US economy combined with concerns about the impact of rising interest rates in Japan and a stronger yen. Huge numbers of investors have taken advantage of Japan's low rates in recent years to borrow cheaply in yen and invest in assets overseas, including in large US tech stocks -- the so-called "yen carry trade". The Tokyo stock price index suffered its sharpest fall in almost four decades, and the Vix "fear gauge" peaked at 65, a level only hit or surpassed a handful of times this century -- including in the early days of the coronavirus pandemic in 2020, and at the height of the global financial crisis in 2008. A Vix of 65 implies investors expect the S&P 500 to swing an average of 4 per cent a day over the next month. "At its peak, the ferocity of the selling was very reminiscent of the 2008 global financial crisis, but without the systemic risk fears," says Bruce Kirk, Japan equity strategist at Goldman Sachs. "The breadth and depth of the sell-off appeared to be driven a lot more by extremely concentrated positioning coming up against very tight risk limits." Market makers say a lack of supply of derivatives to hedge against price movements in the early hours of Monday morning contributed to the sharp moves in the Vix, but in most areas the biggest driver was hordes of investors moving in the same direction, rather than a structural problem. "There's nothing wrong with the plumbing, the market makers were there, you just haven't had the yin and yang of different views going against each other. Everyone is seeing the market the same way and responding to the data of the day," says Patrick Murphy, head of NYSE market making at GTS, the trading firm. The growth of certain investment strategies can make sudden momentum shifts more likely. Three brokers say multi-manager hedge funds -- which have multiple portfolio managers or "pods" running different strategies -- had been a key driver of the recent growth in Japanese markets. These funds are structured to be closed down or have positions liquidated very quickly when markets turn against them. "The amount of money that is sitting outside of the regulated system, they can really move markets," says one senior Wall Street bank executive. Japanese stocks had also been boosted by domestic retail traders using high amounts of leverage; when those positions began losing value, margin calls for more collateral led to additional forced selling. In the US, funds that use algorithms to follow market trends were particularly caught out by the string of disappointing economic data. Société Générale's CTA index, which tracks the performance of 20 of the largest such funds, fell 4.5 per cent in the first week of this month, adding to selling pressure as funds scrambled to cover short positions. When funds that invest in momentum reverse course you can expect a "sharp reaction", says Shep Perkins, the chief investment officer for equities for Putnam Investments, an asset manager owned by Franklin Templeton. "There's a saying: stability breeds fragility," he adds. "The stability led to complacency and the market was testing folks to say, 'hey do you know what you own?'" A rebound on Thursday highlighted the lack of fundamental clarity in what had come before. Less than a week after the disappointing payrolls data, a separate -- and traditionally less important -- update on the US jobs market encouraged the S&P 500 to its best day since November 2022. "The market is so fascinated with what is the latest data point," says Jim Tierney, a portfolio manager at AllianceBernstein. "The ties between fundamentals and day-to-day stock price moves, I'm not sure they've ever been more disconnected than they are today." But there have been some signs of a more meaningful shift in the background. Pressure is building on multiple fronts including the US economy, corporate earnings, global interest rates and politics. Growth in the US is clearly trending downward, and the second-quarter earnings season has been dominated by warnings about consumers cutting back on spending. Investors had also been voicing concerns about stretched stock market valuations for months, particularly in the technology sector. Big gains for the largest tech companies, driven by enthusiasm about artificial intelligence, had helped prop up the wider US stock market, but most have so far shown little return on the hundreds of billions they have invested. "The bloom is off the AI rose a little bit," Putnam's Perkins says. "Even Nvidia, the poster child for AI, announced a delay in a new chip. Given how much excitement was underpinning Nvidia, a delay like that matters." Investors were caught off guard by the Bank of Japan's interest rate rise on July 31, but that also tied into a longer-term trend, with the BoJ having finally ended its negative interest rate policy a few months earlier. Meanwhile global politics has grown more uncertain, with tensions flaring up again in the Middle East, and the entry of Kamala Harris as the Democratic candidate making the US presidential race more unpredictable. Each change may have been manageable on its own, but the cumulative impact is beginning to have an effect. Stocks had been unusually calm since a brief sell-off last October, but this week would mark the fourth consecutive week of declines for the S&P 500, and the Vix had been slowly trending higher even before this week's spike. "There has been a decent amount of new information -- when you start to stack four or five concerning things, it's not unreasonable for the market to have done what it did" since mid-July, Tierney adds. The shakeout caused by the unwinding of the yen carry trade in particular could be seen as one of the last gasps of the pandemic era of easy returns facilitated by easy money. The beginning of the Federal Reserve's rate hikes in 2022 was seen as the end of an era during which low rates dampened volatility and encouraged investments in risk assets, but in a globalised market, investors need not be bound by the rates of their home country. With the last holdout central bank finally -- gradually -- moving away from its low rate policy, another long-term volatility dampener has been removed. Investors are still trying to disentangle the ultimate impact of all these different factors, but few expect a return to the steady gains of the past 18 months. "From a trader's perspective, we feel volatility is back in the market," says GTS's Murphy. Markets have identified a host of potential set-pieces that could cause sharp swings in the weeks to come, according to analysis by Citibank. Options markets have priced in daily moves in the S&P 500 of around 1.5 per cent to coincide with the release of inflation data, the next payrolls update, the annual meeting of global central bankers in Jackson Hole, Wyoming, and Nvidia's second-quarter results. In a historical context, it was the almost two years without a 3 per cent daily drop in the S&P 500 that was more unusual than Monday's price action in the US. "You see these events every once in a while. It was a reminder that when there is consensus thinking, the market can turn on its head in a very short amount of time," says David Giroux, a chief investment officer and head of investment strategy at T Rowe Price. Even after a pullback from the high the market hit in early July, he points out that the S&P 500 is still up around 9 per cent year to date. "It only [feels] horrible because we've had a really good run in the marketplace and only had modest corrections and people got complacent," he adds. "At the start of the year people would have happily signed up for 9 per cent."

Kolanovic redeemed?

Doctor of theoretical high-energy physics and erstwhile JPMorgan chief strategist Marko Kolanovic spent the months before his departure warning that overcrowded momentum trades, like all things, tend towards disorder. After the events of this past week, you'd forgive the man a moment of quiet self-satisfaction. Kolanovic's former team, now led by Dubravko Lakos-Bujas (who may be adopting a Horus / Northman Alexander Skarsgård / Inigo Montoya stance) had this to say about equity rotation, Japan and the carry trade unwind in a note published on Thursday: Equities no longer a one-way upside trade, instead increasingly a two-sided debate on growth downside risks, Fed timing, crowded positioning, rich valuation, and rising election and geopolitical uncertainties. While the market focus in 1H was largely tied to the path of inflation, 2H focus is quickly turning to growth risks given elevated earnings expectations for 2H24 (+9%) and 2025 (+14%). In our prior strategy reports... we highlighted extreme positioning and momentum crowding that historically led to violent unwinds. We stressed the importance of portfolio diversification and emphasized the abandoned Low Vol equity complex (i.e. defensives such as Utilities) offering attractive orthogonal properties and superior risk/reward (see Figure 6). The current market pullback, in our view, was primarily driven by fears tied to weakening growth and the repricing of recession probabilities. Additionally, divergent central bank policies (i.e. Fed / BOJ) further amplified market volatility, disrupting crowded trades such as the G10 FX Carry (~7 STDev event) with USD/JPY at the epicenter (~5 STDev event). This shock reverberated across asset classes with record 1-day spike in VIX, vanishing liquidity, and forced deleveraging by volatility sensitive and trend-following strategies. While the recent market flush took out some of the froth, equity positioning and valuation still remain at risk especially if growth continues to decelerate and the Fed does not show urgency. More investment institutions and shrinking stock markets mean the most lucrative trades are becoming increasingly crowded, according to Ludwig Chincarini, professor of finance at the University of San Francisco and author of The Crisis of Crowding, a 2012 book (recommended to FTAV by two traders this week) on the dangers posed by strategies that neglect to account for their own popularity. "Arbitrageurs trading to exploit anomalies tend to follow the same academic recipe and exhibit correlated trades," Chincarini wrote in a prescient paper published in April: This situation and the frequent use of leverage may exacerbate crash risk in anomaly stocks . . . The time it will take to exit the room will depend on both the number of people in the room and the size of the exit door. Crowding is associated with stronger expected returns but higher liquidity risks, Chincarini notes. Echoing this, JPM write that aggressive deleveraging by volatility targeting strategies stuck in the same handful of trades almost certainly contributed to the market sell-off. And positioning remains extreme even after the recent rush for the exit: Lakos-Bujas et al (typos theirs): Momentum is experiencing its third and most impactful unwind in the past year and contributing to the global equity rout ( Figure 5). We have been cautioning investors of a significant market pullback in an event of a momentum unwind (see Risk for Active Managers, AI Halo Effect, and Right-tail Crowding). The rationale behind the argument being -- (a) Momentum and Market downside risks are tied together -- Momentum mix was and still is dominated by index-moving Mega-caps (i.e. size factor crowding still at 96%ile, Figure 4) and index-sensitive High Beta stocks; (b) Setup ripe for reversal -- Momentum had an extreme right tail non-linear crowding one month ago that receached 35-year record high (100%ile), which after this correction has fallen to 89%ile, Figure 3; (c) AI Halo effect now working in reverse -- Momentum's outsize moves this year were supported by FOMO in AI / LLM related trade that were expected to deliever productivity boosts for businesses almost immediately. During this earnings season, however, some of this enthusiam was partially corrected as revenue ramps from new services were pushed forward. The AI Halo effect is now working in reverse with LLM stocks on average down ~17% from their peaks; and (d) Positioning remains inconsistent with the cycle and recent risks -- Investors are still not positioned for a softening macro backdrop with ample upside opportunity in Defensives ( Figure 6), Low Vol ( Figure 7) and Low Beta ( Figure 8). The current unwind has only partially corrected these extreme dislocations and far from completely mitigating the Momentum tail risk. Tech momentum trades have only partially unwound, the team warn, meaning there's plenty of trouble stored up if and when a real growth scare sparks another violent capitulation into "defensives, bond proxies and low vol". Historical analysis suggests a complete Momentum unwind -- typical of full cycle resets -- results in ~30% drawdown on average based on Long vs Short, cap-wtd sector neutral quintiles of S&P 500. This would imply the current rotation is only ~34% complete. However, we believe we are not yet at the end of the cycle, so expect only a partial unwind in Momentum and not a full flush, although we are gradually approaching towards it. Vindication for Daddy Bear? US and Japanese stocks may have clawed back the bulk of their losses, but it's clear that wherever he may be now, Kolanovic's shadow continues to loom large over Wall Street...

OK, but why *did* stocks fall this week?

Markets go up sober and go down drunk. The workaday business of asset appreciation is interrupted sporadically by intoxicating moments of disorder. Then comes the morning after, and the crapulous task of piecing together what happened. FT Alphaville has sided with consensus when trying to explain this week's multi-asset wassail. The gradual unwinding of carry trades that started in March, most notably but not exclusively involving the yen, became much less gradual and cascaded through markets after hitting a mostly American wall of worry (weak labour market, Fed timing doubts, overcrowded positions in AI and momentum more generally, election uncertainty, etc). Other perspectives are available. Our colleagues at Unhedged rightly note that the Bank of Japan had telegraphed its policy shift, and the equity sell-off didn't get going until two days after an unexpectedly hawkish BOJ rate hike, so a repricing of recession probabilities was probably the catalyst and the acceleration of carry unwind was just a symptom. Economist John Cochrane questions why forced selling by leverage speculators should cause prices to fall, rather than attracting opportunistic buying by leveraged speculators. Justin Wolfers in the NYT suggests that in the short term markets just act silly, like children, so are best ignored, like children. None of these are bad arguments, though they could all use more supporting evidence. Probably the best thing we've read this week is an August 7 note from the global markets strategy desk at JPMorgan, in which Nikolaos Panigirtzoglou and team pull together fund-flows data to show who was selling what. Rather than try to paraphrase the findings we'll quote the note at length, starting with the big, pink, shōchū-tinged elephant in the room: The so-called "yen carry trade" has been so big in size that if inflation induces further BoJ rate hikes in the future, there would be more unwinding of the yen carry trade over the medium term. How big has the yen carry trade been? In our opinion, there are three main components of the yen carry trade. 1) One component of the yen carry trade directly related to equities is the purchases of Japanese equities by foreign investors on a currency-hedged basis. Currency hedging implies a short yen leg which is combined (as a package) with the Japanese equity leg. As the short yen leg backfired recently, investors were forced to unwind the whole package, i.e. to unwind the Japanese equity leg also. Assuming that much of the cumulative increase in Japanese equity holdings of foreign investors since end-2022 has been currency hedged given the seemingly relentless yen weakness, before the past week's correction the cumulative size of this yen carry trade was around $60bn (cumulatively since the BoJ trade emerged at the end of 2022). Cumulative inflows into Japanese equities from foreign residents: 2) Second, there is an implicit yen carry trade when foreign investors borrow in yen to fund purchases of foreign assets including equities and bonds. Borrowing in yen by foreign non-bank investors was worth around $420bn at the end of Q1 2014, according to BIS. Unfortunately, it is difficult to gauge this second component of the yen carry trade on a high frequency basis as the data are quarterly. Total credit in yen to non-bank borrowers outside Japan: 3) Thirdly, there is a domestic (and the biggest) component to the yen carry trade which involves Japanese investors buying of foreign equities and bonds. For example, Japanese pension funds buying foreign equities or bonds do so to meet their yen-denominated future liabilities making it an implicit carry trade. Japan's net international investment position, excluding banks, showed portfolio investments into foreign equities and bonds were worth around $3.5tr before the correction, 60% of which reflected foreign equities. [ . . . ] Adding all the above three components together, the total size of the yen carry trade is estimated at around $4tr. Four trillion dollars is a lot. Forced sales would reset a global market that has overdosed on equities but, notably, that hasn't happened: Implied equity allocation by non-bank investors globally: Implied bond allocation by non-bank investors globally: Implied cash allocation by non-bank investors globally: What happened instead is that momentum-driven quants were pushed out of their long-equity positions, often paired with shorts of the yen, Bunds and Japanese government bonds. JPMorgan's team calculates that the z-scores for the main indices (which measure standard deviations above or below a mean value) have since the end of July collapsed to neutral or lower, having previously been drifting from elevated levels. Momentum signals for major equity indices: Market moves were amplified by momentum chasers including commodity trading advisers (meaning anyone registered with the National Futures Association to handle derivatives trading) and to a lesser extent risk parity funds, JPMorgan says. But retail investors' retrenchment "has been rather limited so far compared to previous equity market corrections", say Panigirtzoglou et al: On our calculations for the equity allocation at global level to return to post-2015 average levels, equity prices would have to decline by a further 8% from here. As a result, equity and credit is not yet reflecting the recession risk that's priced into rate markets. JPMorgan's method is to compare the current cycle peak-to-trough moves in various asset prices to those seen during previous recessions, with the broker using small caps as its preferred gauge due to their higher cyclicality and greater reliance on floating-rate debt. Here, as of August 6, were the scores: From all of the above we offer three possible conclusions: