At the heart of the AI discussion is one issue and issue alone. Are the hyperscalers spending too much? The Economist provides a nice guide, basically showing that we are twice the dot com boom in half the time!

Goldman Sachs also provides a comparison, which puts total amount of capex on a par with the dot com boom.

So basically we have huge amount of capex - more than any other boom on some measures. Adding to this bearish set up - the Capex is now debt financed

And you are seeing the CDS respond to debt issuance.

And then to cap it all off, the US AI models are getting undercut by Chinese models. If you are bearishly minded, then this is a slam dunk short case.

For me, I understood all this, but I saw the capex spending as defensive. With Elon Musk joining the AI party - Google, Microsoft and even Meta could feel threatened, and so would all need to spend aggressively to defend their turf. Or to put it another way, Volkswagon should have massively raised R&D on EVs when Tesla was launched, and now its too late. The big tech was not going to make the same mistake. From this line of thinking, the bear signal would be someone giving up. But I was reading the Microsoft results, which was excellent, and could not stunned by the future contractual obligations they reported. Well over USD 700bn.

The most recent quarter also saw their cloud business become their largest single business line - with over 30% year on year growth. If I saw this type of “contractual obligation” growth in my own businesses, I would also be spending big.

So the question is can Anthropic and OpenAI actually meet the the US 1 trillion backlog they have built up with the hyperscalers? These two companies alone make up the commitment to the hyperscalers.

And here is where it gets hard. Microsoft, Oracle, Google and Amazon must all know that they are extremely reliant on OpenAI and Anthropic. But they also must have their own visibility of the demand for AI related areas. And they should also have good visibility on the capability of Chinese AI, and yet they still spend. So to be bearish you need to take a view that the tech giants have missed something, or have gotten something wrong. And this does happen. The GFC was basically caused by a belief that AIG could not go bust (the investment banks bought CDS from AIG to hedge their MBS business risk. When AIG failed, they were no longer hedged.).

What could I see that the big tech companies not see? I don’t know. I think the biggest risk could potentially Chinese AI completely upends the AI system. But Chinese cloud pricing has been cheaper that the West for years without slowing cloud growth. I would also think this was more likely if Nvidia GPUs did not still trade at a premium in China. That China still wants and needs GPUs means that their AI is still built along similar lines, and not a completely different infrastructure.

Having lived through big busts in the 1990s and 2000s, I understand the fear that investors have. But I think to really commit to a bearish trade you need to find something that the market has wrong. My personal view is that markets have interest rates expectations wrong - and I have good reasons to think that. But on AI, what I see is that OpenAI and Anthropic have huge obligations to the hyperscalers, which they seem willing to invest on the back off. Given their greater transparency, I find that hard to call that bearish. However, if they are basing their investments on the idea that US 10 year treasury yield is going to stay below 5% - then they have a problem. But that is really not tech specific.

Subscribe now

I have done a LOT of short selling in my life. In my previous career I ran a hedge fund of one sort or another from 2006 to 2021, so a 15 year continuous stretch. I was long from 2006 to 2007, then short 2008 to 2010, when briefly long, and then went short again, where I stayed short until 2020. I probably should have gone long in 2016 - but that’s another story. In 2013 and 2014, we were probably the only fund in London to still be net short - so much so that the GS stock loan desk took us out for Christmas drinks. Apparently we were there best customer by far.

I was short Volkswagon all through its squeeze, and eventually made (a small amount of) money on it. That was unpleasant.

I also tried and failed to short oil in 2008, putting on shorts in Petrobras in early 2008, but getting squeezed out before it collapsed.

Actually my ideal short is a cyclical stock you run long all the way up, and then switch to short when the fundamentals breakdown. The reason I am thinking about this is the ongoing momentum bloodbath in markets. The GS High Beta Momentum Index has lost 43% in just over over a month.

The only comparable move that I can remember was 2008. There was a mini one I was on the right side off in 2016, but 2008 was the really big one.

What happened in 2008, was that as the GFC started to unfold, the Fed starting cutting interest rates, which weakened the dollar. The China boom was still going, so commodities, and specifically oil and iron ore were going to the moon. But as recession took hold, this cyclical stock that has attracted momentum money got destroyed. Looking at SK Hynix and other memory stocks, its easy to see this type of market again - down 50% this month.

The thing is that back in 2008, the commodity stocks moved one for one with commodity prices. Petrobras peaked around the same time as oil. Vale peaked around the same time as iron ore.

This time the memory stocks have pre-empted a fall in the underlying. Memory prices are still at or very near all time highs.

Or to put it another way, Micron stock is falling even as earning expectations are being revised up.

The obvious short case is that hypercalers CDS is beginning to sell off and this will cause Capex to fall. I would be a bigger believer of that if I was seeing long dated bond yields falling - the market would be agreeing that recession is coming.

But if you do want to short memory stocks, I can offer one interesting argument. Before memory stocks became THE AI play, power equipment stocks were THE AI play. In the US, GE Vernova was the go to play. It is weak in July - but not broken - not yet anyway.

Koreans being Koreans, pushed their own electricity generation equipment stock up 700% in a year, but has halved in the last 2 months.

Chinese electricity equipment maker, Harbin Electric, peaked in February, and then halved before bouncing a bit this month.

There is a reasonable argument to say Asian stock markets are a lead on the US. As they make physical goods, they tend to have a longer lead time, and can see when things slow earlier. Sometime the Kospi does seem to turn well before the S&P 500, and sometimes it seem to be simultaneous.

One of the big problems with this analysis is that other reliable “tells” on the market do not really work. Normally the time to buy Korea was when its currency began to appreciate, and sell when it weakened. In this cycle, Kospi has soared as the currency has weakened, and fallen as the currency strengthened.

What I am trying to say is that short selling is hard. I used to only take on shorts when I could see three different catalysts to make money. Right now in memory stocks, you need to memory prices fall. And to make that work, you are basically saying that the hyperscalers will see their rising CDS and decided to cut capex aggressively. I would bet on that if I saw their core businesses going into decline - but I don’t see that (yet). Or if I saw recession - but I don’t see that yet either.

Subscribe now

Bloomberg notes that CDS on hyperscalers are all beginning to sell off. Oracle has by far the highest CDS - because it has so much debt. USD 167Bn of debt, with a big increase in 2026.

I am a bit surprised by the increase in Nvdia CDS, as it has USD 80bn of cash on the balance sheet -with relatively little debt. Google now has 118bn USD of debt from USD 25bn two years ago. Microsoft has USD 125bn of debt - up from USD 100bn a year ago. And Meta has risen from USD 40bn to 87bn this year.

In the GFC - rising CDS was definitely “BAD”. But its usefulness since then has been diminished somewhat. Tesla had Oracle level CDS or higher for most of 2018 to 2022.

The high CDS sent a message about financial issues with Tesla, which led to high short interest, which eventually got obliterated, when Tesla generated cash. Short interest is the middle graph below. Being bearish on Tesla was very popular and very wrong.

My point is that CDS does represent increasing debt levels, but is largely agnostic on the future profitability or lack thereof of that investment. But the movement in CDS is interesting. It matches up with a move lower in HYG US. Which I would consider bearish.

Which is wildly at odds with the behaviour of equal weighted SPX - which is at all time highs.

If I look at the KDP High Yield daily - this is generally correlated to equities. You would think US equities would be broadly weaker.

Following on from my autocallable note, US equities are not following dividend futures.

Something funky is definitely happening. GS Momentum Index getting obliterated again, with longs down and shorts up. Now down 5% for the year, from up 70% in June. Yowser.

If I had to take a view, and I guess I do, the market got over long AI trade, and we have seen some negative pricing in CDS from big bond issuance. This led to some selling, which because of the presence of leveraged ETFs, caused more selling. This hurt long books, so hedge funds degrossed, and so bought back shorts and underweights. You end up with equal weight S&P 500 going up -see Apple at new highs. But actually, credit is not looking great - so short covering is happening at the wrong time. It does make me wonder if another momentum trade that suddenly reversed early this year is about to make another reversal? Gold has underperformed S&P 500 hugely since a February peak? Time for a turn?

This would fit more closely in with my pro-labour view of the world. That is one of volatility, as rising cost of capital fights with growing revenue. And the S&P 500 was very volatile in the 1970s.

In some ways I love this market - because I have a theory about how it should work, and we are going to see if its right or not. That theory is pro-labour politics is great for growth but raises the cost of capital - which should be more problematic for the US. Cost of capital is rising now - lets see if the theory holds.

Subscribe now

Korea has always been a market of extremes. I am not sure why, but somewhere in the investing psyche of Korea, there is a tendency to push everything to a limit. One of the best examples came in the GFC, when the Korean Won just suddenly devalued 74%, after years of appreciation.

It turned out the Koreans thought the Won would keep appreciating forever, so many SMEs over hedged their dollar exposure to make more money. When recession hit in the US, the US dollar revenue was not there and all the SME were forced to buy back dollars hedges, causing the Won to collapse. That this happened only 10 years after the Asian Financial Crisis, I still find extraordinary. It is for this reason, I try and keep an eye on Korean “investing” trends.

Korea has been the biggest market for autocallables (structured products) for well over 10 years. Back in 2015, Koreans were very keen on HSCEI related autocallables. In 2015, there was a rally followed by a 50% fall that caused many autocallables to become unhedged, and led to heavy selling of Chinese stocks.

The signs of problems with autocallables come from two sources. First volatility surges even as the market rises. This happens as autocallable pierce upper barriers, and so need to get restruck at higher prices, and then surges again as low barriers are breached, and autocallables are also unwound. VHSCEI shows this.

But a better tell of autocallable trouble is with dividend futures. If they are not tracking the market higher, then you know trouble beckons. You can see this happened in 2015, when dividend futures fell as the HSCEI went up.

So are autocallables to blame for the current sell off in Korea (and Japan)? Well issuance of autocallables has been high.

Read more

For a long time Goldman Sachs have published a pair trade of momentum longs versus momentum shorts. It tells you what everyone who works in asset management knows - momentum is great until its not. We have once again entered the bloodbath stage of momentum investing.

It may look like this falls apart when the long book crumples. But back in 2002, everyone was short tech names, and these rallied, the short book caused problems back then. You can see that more clearly when you look at the momentum long part of this trade in isolation. Big falls in the long book are highlighted. Hedge fund long investing has generally been good.

But what really hurts is when the shorts move against you. These are highlighted below. And that is what we are seeing today - momentum longs falling and momentum shorts rallying. Keen eyed observers will probably recognise that the only time you really want a short selling fund is AFTER short selling has not worked - that is 2000, 2005, 2011, 2021 and maybe, 2026.

The GS momentum pair has dropped 40% from its June peak, and is back to flat for the year. I don’t know exactly what goes into GS Momentum Long Index, but we can use the data from a iShare Momentum ETF to get a good idea. Broadly speaking - AI trades are about 50%.

Typically in my experience this market action tends to cause weird behaviour all over the market - mainly as sudden arrival of risk in one part of a hedge fund manager book then requires risk to be cut everywhere. One area I am surprised not to see action is in currencies. Carry trade has had its best run since the 2000s.

And Yen shorts are at Max short levels. I would have suspected a Yen rally of some sort.

As are Swiss Franc shorts. I would also have expected a Swiss Franc rally of some sort.

Despite this, the dollar remains near one year highs. Given the centrality of AI to US growth in recent years - if stock prices are to believed, then short dollar trade should make a comeback.

I guess what I am trying to say, is that while all the action is centred on AI, the outlook for short selling and potentially short dollar trades look to be improving.

Alternatively, the AI sell off is purely positional - and this is not causing any problems in the rest of the market, because the weakness in AI stocks is not reflecting any real underlying weakness in the AI trade. If I would have to find a similarity to this market, it was the blow-up in the XIV ETF back in 2018.

This fund gave you twice the negative return of VIX, or essential selling VIX. It did well, but when a small shock came along, VIX doubled over night, and it went to zero. But the reason it doubled overnight was because it was so short VIX. There was no information content in the price action, other than the XIV ETF was a bad product.

The S&P 500 fell 10%, and then carried on up -the VIX spike was the death of a bad product rather than anything genuinely market related.

It does seem to me, that leveraged long memory ETFs are the new XIV. I think you need to see far more confirmation from markets to really believe something has changed, like a weak dollar, or change in bond markets..

Subscribe now

Calling a top in semiconductors has been a tough gig for nearly ten years now. I thought 2018 was a top and was wrong. 2022 was a head fake too, and was 2025 Deepseek sell off. But here we are again, with the SOX down 20% from a peak in June.

The bull market in semis is very easy to understand. Pricing power has been phenomenal. TSMC, which I have been covering for 20 years, has been able to push prices up massively over the last few years. The only other time this happened was the dot com bubble.

As of today, we are seeing a very big fall in leading memory stocks. SK Hynix, which recently listed ADRs, is now down 50% from its peak. Amazingly, it is still above its 200MDA.

So was that the top? I see two different explanations for current price action.

Read more

I have been looking at clearinghouses for close to a decade now. The note below is eight years old - but it does show how clearinghouses price risk. The original note is in italics, and I will add in current observations throughout the note (not in italics).

Clearing Houses and Initial Margins

There is perhaps no market that shows the development and growth of globalisation and internationalisation more than the interest rate derivative market. Total notional outstanding of interest rate derivatives in 1998 was USD 48 trillion ( 1.5 times world GDP). It has now risen to USD 436 trillion as of end of 2018 (or 5 times GDP today).

This amount is now just shy of USD 700 trillion, which is about 6 times global GDP.

During the financial crisis, the size of the derivative market became an issue, as the bankruptcy of Lehman Brothers meant that it was difficult for banks to know their actual exposure. In the post crisis period, regulators have moved from a bank-centric risk model to a clearinghouse-centric model. Regulators have also encouraged this move by imposing a capital charge on uncleared trades. According to BIS, in OTC derivative markets central clearance rates have been running at around 80% since 2013. The effect of these changes has been to move CCPs from a participant in the interest rate derivative market, to the central player through which all other participants trade. These days central clearance rates are well over 90%.

Interest rate derivatives, and how they are priced are an important market variable. Using BIS data, we can get the value of all of these derivatives, and compare that to the notional value of them, to get an idea of how these derivatives can fluctuate in value. As can be seen, valuation tends to be cyclical, with lows coinciding with market volatility in 2000, 2007 and in 2018.

Clearinghouses by design do not to take a view on the market. Most clearinghouse calculations of the level of initial margin are set with a lookback period, and in the case of S&P margin, it is highly correlated to 6-month historic VIX (see chart below).

I have been able to update the above graph to most recent data points. Future margin does look a bit higher than expected - but VIX has probably been more volatile in a day to day basis under the current Trump administration.

There is no doubt that a default by a clearinghouse member is more likely when initial margins are low and may be caused by a sharp unexpected move in the underlying markets. In the event of a default by one member this could trigger a chain reaction. Firstly, initial margins will rise in an event of a default, which will restrict other traders ability to participate in the market possibly acerbating the move in the underlying markets. Secondly, if the defaulting member was particularly large player in one asset (likely, as this would be the cause of the default), the members with opposite positions may not receive variation margin and hence will become unhedged at the very moment they need hedging. Thirdly, the clearinghouse may well have to recapitalise itself from the surviving members.

Major clearinghouse members (including JP Morgan) appear to be seriously worried about a clearinghouse problem after the default by a member of Nasdaq Clearing AB prompted this response https://www.goldmansachs.com/media-relations/press-releases/current/multimedia/ccp-paper.pdf . It is asking for the clearinghouses to raise more capital, and to setup a resolution scheme in case of bankruptcy. They are also seeking to limits the ability of clearinghouses to draw on the resources of its members to recapitalise clearinghouses. The default by a member of the Nasdaq Clearing AB in Sweden left remaining members unhedged. This made resolution more difficult as the surviving members were asked to bid for the defaulting members position as well as recapitalise the clearing house

The big question is now that the financial crisis market volatility is over 10 years ago and mainly excluded from Initial margin calculations then do backward looking margin calculation models mean that there is too little initial margin within the system just when it may be needed ie prior to turning points or market shocks.

Given the centrality of clearinghouses, bankruptcy, although deserved, is unlikely. I am more interested in when the poor design of clearinghouses leads to market problems. For me, this I tend to think the treasury market is where the problems are. We know that treasury basis trade has grown substantially, and in 2020 and in 2025 we saw treasury liquidity dried up. Looking at listed margins to trade treasuries, we can see how the spike in margins drove illiquidity. What I find unusual about this is that here is no margin spike in 2025, during the Liberation Day Tariff sell off.

Looking at the MOVE Index, which tracks fixed income volatility, since 2020, you can see that a brief surge in 2020 lines up with treasury illiquidity. Likewise a brief sure in 2025 matches up with illiquidity during the “liberation day tariff” sell off.

Looking at the past reactions of markets, and current levels of MOVE Index, a change of some some of market problem in treasuries is not non-zero. As a reminder, this trade has grown substantially.

The size of this trade is due entirely to the presence of clearinghouses. If hedge funds tried to get leverage of this size from banks, they would be denied (as the note above points out - banks know what is going on). Now we live in a world of chicken. Hedge funds have huge leverage, but assume the banks and regulators will not call time as they have too much too lose. They are probably right, but I do wonder if Warsh and Bessent want to give them a scare.

Subscribe now

Bear markets are funny things. They tend to start small, and isolated, and then suddenly they become a big problem. Almost all the talk now is about whether the AI “bubble” is bursting. From the numbers being reported out of Google last night, maybe you could argue this is as good as it gets, but bursting seems a stretch at the moment. The backlog number and token usage numbers look great to me.

My working theory on markets was that growth would be good (see Google numbers above), but the cost of capital would rise. And this rising cost of capital would be the problem for markets. This has already been a problem for parts of the market. Private equity and private credit equities have traded poorly. Invesco Private Equity ETF is pretty indicative.

But for broader markets, things have been okay. Generally speaking the S&P 500 tends to follow HYG US - the high yield ETF. When this is falling - equities don’t do so great.

First of all, the 200MDA on this is now inflecting lower, like it did back in 2022. Secondly, HYG is falling even though credit spreads are tight. To restate what HYG says, when credit spreads widen, equities don’t do so great. Currently we are near life time lows on spreads.

Another measure of credit is the leverage loan index. This is pretty heavily tied to private equity, and again when this is going lower, bad things tend to happen. This tends to match up well with movement in the Invesco Private Equity ETF. It was weak at the beginning of the year, stabilised, but not able to get back to old highs as it has done before.

What all of this credit metrics do not need would be a treasury sell off. As pointed out, credit spreads are probably TOO tight. The US 5 year is really having a good look at breaking higher.

What I am saying is that conditions for a bond market led sell off in equities are starting to look ideal. The best example is more like 2022, which was a repricing of assets, rather than collapsing revenue or earnings. How bad could it get? Its a tough question. The relationship between rising government bond yields and US equity valuations has been unstable of late. In 2022 it seemed to matter, but since then, not so much.

The big issue is that the US government would be forced to raise more revenue. Stealing a graph from Grizzle on Substack shows the problem.

Looking at US interest payments highlights the problem.

What I am trying to say is that AI trade will someday come to an end, as all trades do. But bear markets tend to come from areas that people know are problem, but are no longer invested in. People know that private credit is bad, they know that private equity is bad, they know the government is bankrupt, so they have already minimised their exposure there. The problem is that it is very hard for markets to ignore higher sovereign yields. I think the next sell off in sovereigns will be particularly problematic as it will likely send a number of hedge funds doing the treasury basis trade bankrupt overnight. I live in hope.

Subscribe now

I have been talking about GLD/TLT for so long, even I was getting bored of it. Since starting to talk about it in 2022, it has been on absolute tear.

But from a high in February, it has had a chunky 26% fall. And for the first time since 2022, went through the 200 MDA, and completed a dark cross. For technical only investors (CTAs etc) this is very bad. It has had a rally recently - but should this be sold?

Read more

I have been pretty sure that yields on JGBs were going to go much higher, and this has been right.

I did have an expectation that this would drive yields much higher elsewhere. This has not been so true. US yields are higher, but still within a level that we have seen before within living memory.

And the biggest surprise have been the collapse in Chinese bond yields, to close to all time lows.

What disturbs me about this, is that my view on bond yields was driven by a view that the world is becoming more “populist” or pro-labour. Generally speaking, this would also include China - and yet its bond yields remain at low levels. If I just look at Japan, I could speculate that loan demand is what drives bond yields. Japanese loan data inflected in 2013, and have not looked back.

FDIC data for US banks show aggressive loan growth has become in recent years, and in particular in contrast to slower loan growth in the 1990s.

The problem is that Chinese loan growth also seems pretty good.

One thing that is different between China and Japan, is that foreign investment is dropping in China, and rising in Japan. The best was to see rising FDI in China is in the increase in it NIIP. Falling FDI into China, and rising FDI out of China should drive a higher NIIP.

While from Japanese NIIP data, we can see that foreign investment into Japan is rising from a low base.

The problem with FDI as a driver of bond yields is that the US has FDI into the US even through the period of low yields.

So that leaves government spending. Going back to the US, government spending seems to drive yields. Government expenditure rose rapidly in 1970s and in 2020s.

And a similar dynamic is playing out in Japan. The increase in spending matches the inflection in JGB yields higher.

Chinese data is a bit hard to compare but there is definitely a slowdown in government spend in 2025 at least.

There is a way to check this hypothesis. In Europe, the UK has the higher government bond yields. Government expenditure looks like this.

Switzerland has the lowest bond yields in Europe. Its government expenditure looks like this.

This seems to prove what I always thought. Sovereign bond yields reflect markets belief in governments ability to control spending. And in the UK, with Reform on the right and Greens on the left - there is no belief.

Subscribe now

Chapter 6 of my book. Find other chapters here.

Generally speaking, my kids hate playing sports against me. I am lucky enough to have married a very athletic wife, and lucky enough again to have two very athletically gifted children. I am considered the least athletic member of my family. But my kids still hate playing against me.

To compensate for my lack of physically advantages, I almost always develop a strategy to compensate. In racquet sports, if I see you have a weak backhand or don’t like high bouncing balls, then I will make you hit those shots relentlessly. In basketball, I know that younger players want to run, so I aim to drive the ball to get fouled, and then take my free throws very slowly.

In any competitive game, if I am not hated at the end of the game, then I have not tried hard enough.

In finance, I have tried for a similar relentless attitude. I noticed that equity fund managers rarely think about currency risk. At most, they generally assume the past will look like the future. But for me, currencies are that most dangerous of financial assets - very stable for long periods of time, and then suddenly break.

Read more

I always think of this as one of my all time great notes, for two reasons. First, almost no one had looked at this before I did, and second, after I looked at it, I had research providers trying to sell me this as their own research.

In some ways, this is a timely note - as once again ETFs (this time leveraged single stock ETFs) are causing market havoc.

Japanese US REIT funds and the Buy Case for Yen (2015)

The last few years in Japan have seen the emergence of and selling of some innovative high yield funds. These asset management products are designed to pay out very high yields, while sadly almost guaranteeing the destruction of capital. There are numerous examples of this but this note shall deal with only US REIT Funds sold in Japan.

According to Goldman Sachs, the largest 20 Japanese US REIT funds have 50bn USD under management. The single largest is the Shinko US REIT fund (Bloomberg Ticker: 06311049 JP), which has an AUM of around 14bn USD. For comparison the Vanguard US REIT Index fund has 26bn USD of assets, total US REIT market capitalization is 720bn USD.

Japanese US REIT funds tend to offer a dividend yield in the high teens. The Shinko fund has a yield of 17% currently. For reference the US based Vanguard REIT Fund has a yield of 3.4%. The extra yield of 13.6% does not come from leverage. In fact the assets owned by the funds are very similar. Rather Japanese funds will sell down assets to ensure that a higher yield is paid out.

From 2005 to 2010, the Shinko fund and Vanguard fund had very similar performance, after 2010 a huge divergence opened up. In my view, Shinko noticed that assets under management at their fund grew aggressively when funds offered a high dividend yield. Talks with participants in Japanese financial markets indicate Japanese investors prefer funds that offer at least a 1% monthly dividend yield.

As can be seen from above, Shinko AUM exploded higher once it began to offer a yield in excess of 12%, and it continues to rise. The other major funds that I have looked at all offer similar yields and are presumably managed in a similar way.

With such high dividends, assets under management can fall as the fund needs to sell assets to pay dividends. To get an idea of when this can happen we can use the change in the share count of the fund multiplied by the fund price and then subtract the dividend payout to calculate net cashflow movements in the fund. We can see that there have been two times when assets raised have fallen below the cash needed to pay out.

This has a very clear effect on the shares outstanding for the Shinko fund, which fell in 2012, and again in recent months.

I suspect this new fund structure will have a big effect on the Yen. Typically, the flow of investment funds relative to currency movements is unknown. The reason for this is that sometimes investors will see the currency depreciation as a chance to buy an asset cheaper. Certainly US listed Japan equity funds tend to see rising share count (i.e. raising assets) when the yen is weakening. In other cases, weak currencies tend to correlate to weak returns and investors selling assets. The latter is what we have seen more recently with emerging markets funds

What I find most intriguing is that at constant values, ie US REITS remain constant, and the Yen exchange rate does not change, and Shinko does not raise any more assets, then this fund will be forced to sell 14% of its holdings in the US to repatriate to Japan to pay dividends each year. What is true for Shinko would be true for the industry as a whole.

The implications of this are intriguing for me. Firstly, assets in these funds are very unstable, as the high yields destroy assets under management very quickly. Secondly, if there was a shock that caused yen to appreciate or US REITS to fall in value, funds like Shinko would be forced to liquidate their assets very quickly to meet redemptions and dividend payments. Their liquidation could cause yen to appreciate further and possibly REIT to fall in value locking them into a vicious cycle. Even more disturbing, that this is only one of seemingly many assets classes that have this feature.

For investors nervous about the state of global markets, a long yen position seems to offer a good hedge in my view.

Japanese US REIT Fund – an Update 2016

In 2015, we released two market views about Japanese investors being short Yen via fund products. One was long Turkish lira (TRY) and short Yen (JPY), while the other was long US Reits (Real Estate Investment Trusts) unhedged. Both of these assets classes are relatively concentrated, the two largest TRY funds account for 80% of assets, while the two largest US Reit funds make up 50% of assets. This year we have seen Yen rally significantly. While the size of the Turkish lira funds (Amundi European High Yield – TRY Course and Nikko Pimco High Income Soverign Fund – TRY Course) have shrunk dramatically, the US Reit funds (Shinko US Reit and Fidelity US Reit Fund) have increased.

The contrasting fortunes of Turkish lira funds and US Reit funds in Japan are undoubtedly connected to the divergent performance of these assets. The Turkish lira has been very poor, while US Reits have performed well.

While the US Reit funds and the Turkish lira funds are very different assets, they both offer Japanese investors very high dividend yields as the main selling point for Japanese yield hungry retail investors. The largest Turkish lira fund, the Amundi Europe High Yield fund Turkish Lira Course, still offers a 25% indicated yield. The largest US REIT fund in Japan, Shinko, also offers an indicated yield of 25%. This level of dividend yield is typical of the funds in this space.

US Reits do not offer 25% dividend yields, and the Japanese US Reit funds are not leveraged. Instead, they commit to maintaining these dividend yields from capital. Fortunately, for these funds they have been able to raise enough funds and realise enough capital appreciation to grow their funds’ assets significantly. The problem is that with increasing share count, and large yields the two largest funds are paying out 64bn JPY (640m USD) a month. Since 2015, a period of significant capital raising, they have on average raised 80bn JPY a month. If Japanese investors just cease adding capital to these funds as they did in 2012 and 2014, it would precipitate significant selling of US Reit holdings to pay dividends.

It is also interesting that US Reits, in contrast to the trend for most US corporates, have generally been issuers of shares, some quite dramatically. In general terms, Japan based US Reit Fund use the FTSE North America Reit Index as a benchmark. According to the index provider, the top five stocks in this index is Simon Property Group, American Tower Corp, Public Storage, Crown Castle Intl and Prologis.

Reits have various tax advantages, as long as they pay out the majority of their income as dividends. This means that to grow assets, they need to issue shares to buy assets. The Reits above are typical in that share count has a tendency to rise over time. It seems to me that for the Reits to grow, they need to be able to find buyers of new shares at ever higher prices, while Japanese based US Reit funds need to find new investors to not become sellers of Reits to meet dividend commitments. This strikes me as particularly unstable.

Subscribe now

Republican and in particular Trump energy policy can be summed up as “Drill, Baby Drill”. It has become de rigueur in Substack and elsewhere to praise US policy, and the criticise European energy policy. The UK has come in for particular abuse. The main nature of this abuse is that high energy costs in the Europe and a pursuit of net zero has led to business stagnation. This Economist article is pretty much standard. The key graphic is below.

The political view on this is that the UK has been “too woke” and heavy taxation on oil and gas production, as well as trying to upgrade an electricity grid has made UK and Europe uncompetitive. Data from one my favourite publications, the Energy Institute Statistical Review of World Energy 75th edition provides compelling evidence. North American natural gas production has surged, while European gas production has collapsed.

But as I read through the rest of the publication, I was struck by how much it made the US policy of increasing fossil fuel production looks like a short term win, but long term loss. This review starts strong by highlighting how renewable energy became the biggest source of new energy in 2025 - the first time outside of a recession. Now you may have, like President Trump, a natural born hatred of renewable energy. But last year it proved it can add capacity at a fast enough rate to not slow GDP growth.

And I find it stunning that Pakistan gets 20% of its energy from solar panels. This does hold out the possibility of decarbonised future. If developing markets can adopt renewable energy early, then the outlook for fossil fuels looks negative to me. For places like Pakistan, building out a decentralised system seems like a good option.

We can see that the relationship between GDP and fossil fuel supply is already breaking. This is probably the most negative chart you could possible produce on fossil fuel in my view.

China, where EV usage is higher, is already slowing oil consumption at a much lower per capita level. This is almost a reverse of the emerging market investing model. When I started, it was assumed emerging markets would converge on western consumption levels. Not it looks like Western consumption levels will converge to Chinese levels. Potentially you could see western demand for road fuel drop by half.

The world is moving towards an electrified future, with renewables meeting much of the demand growth.

It follows it up with this very arresting graph of US exceptionalism. When I look at this, I start to see a potential problem for the US. A decarbonised world does seem inevitable to me - despite the protestations of corporates and older voters. Eventually the Greta Thumbergs of the world will be politicians and not activists. North America is alone in making its energy supply more carbon intensive.

Here is where the politics gets interesting. European carbon prices have remained at a high level for a few years already. Which comes back to the original argument that European electricity costs too much. But does it really cost too much, when it incentivising the growth of renewable energy and true energy independence? Does it make sense to switch reliance on Russian energy exports to US energy exports?

Which is a much higher level than markets elsewhere. California Carbon Allowance is one third of European levels.

But this has had the effect of leading to a much more thoroughly decarbonised electricity generation system than either the US or China. Europe is much closer to fossil fuel independence than either China or the US.

In this pro-labour world we live in, with tariffs and industrial policy. It would be very easy to see Europe seeing US renewable energy policy as “carbon dumping”, or race to the bottom deregulation. Why not tariff US exports to price the externality of the carbon heavy energy production? What I am saying is that the data already shows that fossil fuel consumption has already broken away from GDP growth, something that markets have already shown when Russia turned off the energy supply in 2022, and again this year with the closing of the Straits of Hormuz. In fact, sometime the best way to think about US foreign policy changes, is going from a free trade era to ensure maximum oil supply, to a competitive era, where it is looking to reduce competition for its exports. But long term, do you want to be investing in an industry that looks to be heading for decline? North America already lags badly behind China and Europe in Wind capacity.

Likewise in solar capacity.

For the first time, I wonder if we could see US LNG exports actually decline at some point? While the Straits of Hormuz are closed, this should be fine. But in my mind, the pieces for globally falling fossil fuel consumption are coming together.

I do wonder if US shale producers see this same world as me. Permian drillers have been running down their stock of wells now for a few years. Even with the supply disruptions in Russia and the Middle East.

If there is one thing I have learnt in my life, is not to bet against technological change. And it is hard to not see technology making fossil fuel yesterdays energy. The interesting part for me, is that we are already beginning to see signs of the benefits of European investment into renewables. Negative pricing is becoming more common.

And this negative pricing in driving a build out in battery storage.

Putting it all together, it looks more and more like fossil fuels are a dead industry.

Subscribe now

From a peak in June, SK Hynix, a DRAM and NAND manufacturer has seen its share price fall 37%. Given the huge run up over the last year, this still leaves it well above its 100MDA.

Generally speaking, SK Hynix follows the DRAM price. So it is odd that it has been weak as DRAM prices have hit new highs.

You could perhaps argue that we have hit the end of the AI trade - and capex is going to be cut, as we saw all the way back in 2000. This is of course possible, but the JGB market gave a pretty good heads up that we were heading to a deflationary environment back in 2000. These days it keeps sending an inflationary signal

Much more likely is the announcement that SK Hynix will double capex to help reduce memory short fall has been the negative priced into markets.

The market does not like capex. An increase in capex makes it more likely it becomes cashflow negative in the future. This is both as natural consequence of spending more, and the second order effect of increasing supply leading to lower prices.

A similar logic has been holding back the hyperscalers, who have been lagging the market for a year now.

Even more intriguing has been the break out in that decidedly non-capex spending company, Apple.

Apple, has not raised capex in years, despite surging cashflow.

This is where this market gets very interesting to me. You have Warren Buffett style capital discipline in Apple, up against Elon Musk style capex heavy investing. Both have become rich with very different investing styles. Steve Jobs probably would have Apple investing as well, as he would no doubt fear that Apple could become obsolete, while the money managers are happy to see capital returned. And this I think goes to the heart of the problem for the markets. Not investing keeps cashflow high, but exposes you to competitors, something that is being priced into other software companies. My view, which remains unchanged, is that with SpaceX entering the AI race, everyone has to invest. If you see capex spend as negative - then being bearish is correct. But if you bearish think that capex is going to suddenly stop, and prices are going to collapse, then I think that is far less likely. And this seems to be what sovereign bonds are saying too.

Subscribe now

I have been thinking about markets in a political way for a few years. I have this theory that we were moving to a “pro-labour” era, which would be good for wages, but higher interest rates would be bad for asset markets. In some ways we have seen this, with market doing very well while 2 year yields were falling, and now we are getting all sorts of market weirdness now that 2 year yields are rising.

What has been most interesting is that rising yields HAVE been hugely negative for SPECULATIVE assets. Silver, which does have central bank supporting it, has gotten smacked this year, down 55% from its peak.

Crypto has been given the same treatment - with Ethereum at 2020 prices.

I have mentioned private credit and private equity have also been at the mercy of this speculative asset smackdown. But the trillion dollar question is whether AI is a speculative asset or not? Lets take a closer look at Korea. The Kospi, which has broken out of a long term ranges, has become a memory/AI trade.

But weirdly, the Kosdaq, another speculative Korean index has been very poor this year. What I am trying to say here is that generally speaking Korean stocks have been difficult, but the memory/AI plays have been good.

Now here is the rub. I think many people think AI is entirely speculative. I think this is not correct. But I do think speculative structures have been erected around AI trades. Why is AI not speculative? Well memory prices have moved, and remain at highs.

If the AI boom is done, then why does TSMC remain at all time highs?

If you are bearish, then you could point at Korean memory stocks or Micron, and say, look these stocks are down 30% - its the beginning of the end.

But the thing is, the memory stocks sit at the intersection of AI and speculative trading. When I look at MUU - the 2x levered Micron ETF - and see shares outstanding up 8 times from a year ago (market cap USD 4bn), I see speculation has run wild.

What I am saying is that it is right to be bearish on speculation, but I am not sure it is right to be bearish on AI. I am also going to follow that up with a view that perhaps we are near the end of the “speculation” bear trade. As pointed out, memory prices and TSMC local shares remain strong. But demand for TSMC ADRS (the US listed stock) has historically been so strong, it has traded at a premium to locals for many years. As of today, the premium has fallen to 6%, down form 25% not so long ago.

Typically the collapse in premium happens at lows. That is speculative unwind may have run its course.

VKOSPI, which is VIX for the Kospi200, is running at about 80. This is a level I normally associate with a washout in markets, and potential buying opportunity.

So what am I try to say? The pro-labour trade should be good for growth, but higher yields should be and has been bad for speculative assets. From where I sit, AI trade does not look speculative, but the market has built up a range of “speculative” assets around AI. For my money, I think the chat about AI crash or not is missing the bigger picture. If short term rates are bad for speculative assets, their remains one speculative asset that really moves markets - the US 30 year treasury. Very quietly, yields have passed through 5% again.

As my substack colleagues over at Grizzle Research and Quant point out, the US government really does not have any money anymore. Everything that comes in the door goes out in interest payments and entitlements.

I do think it is right to be bearish speculative assets here - but I am not sure AI is speculative. On the contrary, the “safe asset” that underpins all finance looks to be more problematic to me.

Subscribe now

Memory stocks, namely Micron, SK Hynix, Samsung and Sandisk have been wild of late. These stocks have been whipsawed by the increasing use of leveraged ETFs, and the capital raisings from SK Hynix and (potentially) Samsung. A good example would be KORU US, which is a 3 times leveraged South Korean bullish ETF. South Korean markets is dominated by SK Hynix and Samsung. KORU US has been volatile.

The volatility of KORU US does not match up with the still soaring DRAM prices.

Same can be said of leverage Micron ETF - MUU US.

The presence of such leverage makes the memory stocks tough holds here. Wafer stocks on the other hand still seem to lack the leverage ETF presence.

Read more

Subscribe now

Below is a note from 2013 - In Italics, with added thoughts after

Gold buying as a leading indicator of financial distress

I wrote a note on gold late last year, pointing out that the gold price and the Indian rupee seemed to be correlated to me. With that view I suspected that if the Indian rupee fell, gold would then also fall. However market action has been the reverse, with gold falling before the Indian rupee fell, and the fall in gold was greeted with a record surge in gold buying in India. These Indian gold buyers have been rewarded with gold once again moving to new all time highs in rupee terms as India has entered a period of currency and possible financial crisis.

This led me to thinking about whether it was common to see elevated gold buying prior to financial crises. Certainly, the most motivated buyers of gold will be those depositors who take a view that their banks could go bust, either through their own reckless behaviour or government’s reckless behaviour. I went to the World Gold Council website (www.gold.org) to collect data on consumer (not official) buying of gold. Alas there was no single spreadsheet which I could consult, so the data had to be complied from a number of news releases. Below I present gold buying over time with what I believe is the best visual representation of financial distress in each market.

Thailand

We can see an increase in Thai gold buying before its major devaluation 1997.

Indonesia

Likewise Indonesian gold buying spiked and peaked before its currency devaluation.

Japan

Japan has never faced a meaningful currency crisis, but an ongoing financial crisis. Japanese gold buying peaked as Japanese banks started to underperform the markets. That is Japanese investors perceived the weakness of their own financial system and started buying gold.

USA

Much like Japan US gold buying seems to peak at the same time as banks peaked versus the market, and seem to reflect depositor’s views on US banks that they were undertaking reckless lending behaviour.

Conclusion

We have recently seen big increases in gold buying by emerging market countries with the World Gold Council reporting a 71% increase in Indian buying, Chinese buying was up by 87%, Turkish buying was up by 73% and Russian buying up by 184% in volume terms in the second quarter of this year. For gold bulls they take this as a sign of ever increasing emerging market demand for gold potential driving gold prices higher. For me I take it as a sign of domestic emerging market investors voting with their feet against their own currencies and financial systems. Investors should be aware of the rising financial risks in emerging markets.

In a recent note, I showed how Chinese consumers had gone all in on gold.

And since this buying orgy, GLD/TLT has fallen back below its 200MDA.

Looking at this note from 2013 (which I reposted because the data is chore to find, and putting on Substack means I will be access it for ever), the implication would be that China is about to enter a financial crisis of some sort. China devaluing would be very negative for GLD/TLT - so it stuck me as a good note to repost. All the crises above were in part property/financial crisis. China has already seen a slow down in construction.

And the Asian High Yield Market was very poor in 2022, mainly driven by Chinese debt problems.

The HSCEI Index has been very poor since 2007, reflecting these problems.

But the Chinese Yuan has been strong this year in contrast to Japanese Yen or Korean Won. Could Chinese buying be signalling a potential devaluation?

Here is where I reflect on how much the world has changed from when I first wrote this gold note. Back then I would be convinced that Chinese gold buying was a sign of trouble. But now, in the modern world we live in, I think it reflects the political world we live in. In the 1990s and 2000s, central banks were sellers of gold, and the lack of tariffs and industrial policy make devaluation a popular policy choice. But today, what would a Chinese devaluation solve? Nothing, as almost certainly tariffs would follow. And with trust in the US ebbing, central bank act at buyers of last resort for gold - just as they used to do with treasuries. A changed world indeed.

Subscribe now

Peak Oil Price

From $10 per barrel in 1999, crude oil has been in bull market with the brief exception of the financial crisis in 2008/9. So much so, that the recent rise in Brent oil to $119 per barrel barely generates a mention in the financial press. However, the reality is that the current spend on energy is similar to the level of spending seen during the second oil shock in the late 70s (see chart below). At that time, in a decade, oil went from $2 per barrel in 1970 to $35 in 1980 before falling gradually back to $10 over the next twenty years.

To try and adjust for the emergence of China, I have looked at its energy spend as percentage of GDP and then compared it to both Japan and South Korea. As can be seen in the chart below, China is currently tremendously inefficient in its usage of energy. Japan has always been far more efficient in its use of energy to produce GDP than either of its neighbours. While China has spent more energy as percentage of GDP, this was in the 1970s when the Chinese economy was imploding.

Of course one of the big differences is that China has huge domestic resources of coal which allows it to be inefficient with its energy. However, even then it still compares badly with the US. Recent US numbers are also overstated as I use the Brent price rather than WTI, and ignore lower cost of natural gas in the US.

China already consumes more energy that the US, even though its GDP is half that of the US. I like to look at energy usage in terms of per capita and also in terms of how much energy is needed to produce a dollar of GDP. As seen below, China is around half the per capita level of Japan, UK and Germany. Given that Chinese energy consumption has more than doubled over the last ten years, this implies a slowdown in consumption growth going forward in my mind. Or in other terms, the base level of consumption has increased so much that the Chinese growth rate must begin to slow. Regarding GDP produced per unit of energy, China is around 25% less efficient as Germany, UK, and Japan, and half as efficient as the US. Again, efficiency gains should reduce the growth rate of Chinese energy consumption going forward.

One of the reasons I scrutinise Chinese energy demand, is that oil demand has become increasingly dependent on China. If you look at the chart below, you can see that in the five years to 2007, oil demand grew in most regions. However, since 2007 China has continued to grow while the US, Europe and developed Asia have reduced consumption. Furthermore, other areas that are growing demand for oil also tend to be oil exporters, hence their demand for oil tends to move with the oil price rather than increase at lower oil prices. Hence the drivers for oil are becoming increasingly narrow.

While demand growth is increasingly reliant on Chinese demand, supply is also beginning to respond to higher prices. As can be seen below US oil production is surging.

We are seeing signs that increasing oil production and slower demand growth from the western world is starting to cause spare capacity at OPEC producing nations to rise. Historically this has occasionally preceded a fall in the oil price.

What is also surprising about the oil market is that speculative investors no longer seem willing to bet on lower oil prices. In the late 90s speculative investors were often willing to take a net short position in the oil market – but particularly since 2007 the market seems to have become structurally long. This may reflect the rise of passive investors in the commodity market via ETFs, or possibly the effect of zero interest rates making speculation cost free. I suspect a combination of these two reasons explains the long bias of investors in oil.

The sustained rise in oil prices has also led to a huge surge in government spending in oil producing nations, with Russia in particular needing $100 per barrel to balance its budgets. Some analysts take this to imply that oil now has a floor at 100 USD a barrel. I think this is unlikely. Far more likely is that if and when oil prices begin to fall, either through slower Chinese growth or higher interest rates, we are likely to see large currency devaluations from oil producing nations in order to balance their budgets. The most recent example of this is Venezuela, which just devalued by 47% in a single day. When I see record inflows in to emerging market bonds funds, which in my view, are the most exposed assets to this type of devaluation, I feel compelled to be short areas that I believe will fall in value when the oil price begins to fall.

Subscribe now

For various reasons I have been having discussions about gates. A gate for an investment fund is pretty common these days - as it seen as a way to protect investors from themselves. It is so common that is just usually added into many investment fund articles. Personally, I hate gates. I have on occasions had opportunities to impose gates, and have turned them down. When I took over the management of the Horseman Global fund, we had redemptions of well over 80% of assets - but chose not to gate. The only time I can think a gate might make sense maybe in time of total crisis, like GFC or Covid, when there is no liquidity anywhere - but even then your investors will likely have other concerns so a gate makes little sense. But gates tend to be standard parts of most funds these days - but my advice is for most fund managers to be extremely cautious in using them.

In the midst of the on/off war in Iran, and an epic AI boom, and all sorts of other shenanigans, you may have forgotten that private credit funds have been gating their clients.

And this gating has been why the Invesco Listed Private Equity ETF (PSP US) (many private equity firms do private credit as well) has been poor on both an absolute and relative basis.

For me, once you gate a fund, it pretty much a dead product. Fund management is a trust game - and if people want their money back you should give it to them. And once you gate, you create terrible incentives for other investors. Gating means or implies you have a bunch of assets that cannot be priced or sold. And for investors who have not asked to redeem, they start asking themselves some difficult questions. Will the fund manager have to sell the liquid positions to meet redemptions, leaving the fund only with illiquid assets? The answer is almost certainly yes in my experience. And does anyone want to then invest into this fund, with only illiquid assets that cannot be priced? Not really. Cliffwater is a good example of this phenomenon.

And with more money going out that coming in - you would wonder if that NAV is correct.

Cliffwater has capped redemption at 5% of assets, despite have 17% redemptions. Following the logic laid out above, this looks to be a dead product. Who would put money in this now? The one thing I find very odd about this is that Cliffwater is getting into trouble when high yield spreads are at all time lows.

For reference - Blackstone listed in 2007, just before credit spreads blew out - and fell 90%.

One wonders what private credit would actually do in a recession?!?

Anyway, the point of this notes are gates are bad, and there is almost never a good reason to use one, and using one will likely make a bad situation worse.

Subscribe now

I was cleaning up some files from my laptop, and I found a stash of my oldest research I first published back in the early 2010, when I was trying to convince everyone China was going wrong (tough sell back then). I am just going to put it here for posterity, so I can easily reference going forward. Enjoy. Three notes, first is on iron ore, second on mining, and third on Brazil - all from 2011 and 2012.

A follow up note on mining.

And a note on a potential Brazilian debt crisis.

And in conclusion, yes, I have always been this boring.

Subscribe now