Violent Ping Pong

If you ever watch ping pong pros play a long point, you see how much effort it takes to hit a small ball back and forth.  The ball just goes from one side to the other, but there is a lot of energy expended.  This stock market is violently going up and down.  But it ends up in the same place.  Its a ping pong match.  Sectors are making big moves, but the index is relatively calm.  Correlation is low. Intraday volatilty has gone up, even day to day volatility has gone up, but week to week volatilty remains low.  

It reveals supply and demand equlibrium at the index level, but disequilibrium at the individual stock and sector level.  Tech stocks, in particular software has been weak.  Consumer staples, industrials, energy, and materials has been strong.  Combining -8 with +8 is zero.  Sectors are ignoring the index, moving to their own beat.  The index is also ignoring the sectors.  

Passive inflows are keeping the beach ball afloat. There is still risk seeking capital entering the system via retail investors, but it is being offset by reduced stock buybacks due to AI capex.  Before the AI capex boom, stock buybacks were the axe.  Now its passive investing flows via retail investors who have been conditioned to believe that equities are the best long term investment, no matter what.  FYI, equity inflows are the greatest from January to April, and then tail off into the fall, where they are the lowest.  

Hyperscaler capex has gone from $200B in 2024 to an estimated $600B in 2026.  Corporate buyback trends have made a big down move since 2024.  They are reflection of big cap tech doing less buybacks in order to fund capex on the AI buildout.  

Some counterintuitive moves are happening among retail favorite stocks.  While retail continues to pile into individual stocks (see below), its not having the expected effect of higher prices.  Instead, Reddit Wall Street bets type investors are experiencing tough times.  Even with the SPX just 1% below all time highs.  

Bitcoin/ether are trading like they have the plague.  TSLA and PLTR continue to lag badly.  Same goes for nuclear, quantum, drone, and AI data center plays.  Even the beloved space theme has been weak lately.  Its a world of pain for retail, even at SPX near all time highs.  This is an ominous sign.  Their disregard for valuations is hitting back, as the market now is less friendly to bets on obviously speculative and overvalued tech stocks.  

Once again, this reminds me of spring and summer of 2000.  After a huge influx of retail money into stocks the previous several years, the index entered a stall period, unable to go higher.  Semiconductors were the hottest sector.  With the SPX near all time highs, lots of retail favorite stocks were lagging, except a few select internet related favorites like CSCO and MU.  Consumer staples and energy were outperforming tech.  And in 2000, you had lots of dispersion, with many Hindenburg Omens signals firing off.  Lots of new highs and lows as the index was near the highs.  Like we’ve seen the past several days. 

On that list of signal dates, there are quite a few that are around significant intermediate to long term tops.  July 2007, January 2018, September 2018, January 2020, December 2021. 

There are many bearish long term signals in the market, which I have noted earlier this year.  Last week you had a mini flush out of risky assets, in particular precious metals, bitcoin, and SaaS stocks.  There a couple of bullish short term signals coming from hedge fund shorting activity and the options market.  

GS Prime book saw the biggest increase in short trading flow among US single stocks over the past 5 years.  

You also have seen elevated put activity, similar to November 2025 levels, which saw lots of short term volatility but eventually led to a V bottom.  ISEE index is back down towards the levels seen 3 months ago.  

Investors are getting well hedged for further downside, which will mitigate any reflexive effects from investor selling.  The ISEE index 20 day moving average is still not close to the November lows, so it may take a few more days of a choppy down move before you get a tradeable bottom.  

Big picture, the bearish long term signs continue to be confirmed by market price action, with activity very similar to early 2000.  For the short term, you have bullish fiscal flows from the OBBA bill which will counteract the bearish long term signals.  Income tax refunds in 2026 are going to be about $1000 more than the past few years, which will be a short term stimulus.  Some of that is priced in, as OBBA is one of the reasons investors are so bullish for the start of 2026.

 

The talk of the Street has been Anthropic's Claude taking over the software world.  Now everyone is extrapolating an LLM wrapper as a cheap replacement for coding and software development from engineers.  The SaaS stocks are getting obliterated.   LLMs are probabilistic engines, and they are only as good as the data that is fed into the model.  And there is a lot of junk data out there.  No matter how fast and expensive the LLM is, it doesn't change the quality of the data.  

Investors are extrapolating the current trends and improvements into infinity.  But AI doesn't get better like semiconductors did following Moore's law.  There is an S curve in technological advancement, and AI seems to be much further towards the flattening out stage than people believe.  ChatGPT is a perfect example of diminishing improvements with each edition.  The jump in performance was much bigger from GPT2 to GPT3, than from GPT4 to GPT5.  Many complained that GPT5 was almost like GPT4, and were massively disappointed.  You didn't see that in CPUs until the 2010s, when Intel CPUs couldn't make anything much faster, so they just overclocked their previous CPUs to improve performance.  It took Intel over 30 years to reach that performance wall.  Its taken OpenAI less than 3 years to see that wall.  And while  some people may think Anthropic is now the top dog and OpenAI is a runner-up, these LLMs are very similar to each other.  You can tweak OpenAI's model to make it look like Anthropics.  What Claude can do now will be similar to what it can do in 5 years.  Those extrapolating the AI improvements over the past 3 years due to more intensive compute are trying to fit Moore's law into a technology that doesn't apply.   

Software companies were loved due to the SaaS revenue model, but they got tremendously overvalued.  And they issue tons of stock based compensation, which inflates their earnings.  These big down moves in growth stocks happen because of overvaluation.  And people say that valuation is not a timing tool.  It is not a good short term timing tool, but it is a very good long term timing tool.  

The COT data didn't reveal much for the indices.  It did reveal that the blowoff top and subsequent violent pullback in gold caused a big reduction in long positioning.  Open interest is down huge over the past 2 weeks.  Lower open interest means that you have taken out a lot of weak hands/scared money.  It also means lower volatility because you will see a lot less forced buying/selling going forward.  I expect a range bound market for gold and silver for the next several weeks.  

Last week, you saw violent price action ending with a strong Friday rally in SPX.  CNBC Fast Money experts seem more bearish than usual.  That usually signals a short term bottom in bull markets.  But it is an aging bull market, and it has shown its age in the past few months.  In bear markets, CNBC "experts" being bearish doesn't mean much.  The best buy the dip assets out there seems to be gold/silver, followed by international equities, and then SPX.  The worst is bitcoin/ether.  Bitcoin is trading like a classic post bubble asset.  I expect a milder version of what is happening to bitcoin to happen to precious metals in a few months.  Still remain long gold, but with plans to exit this week.