The Compound: The Pain in Your Account and the Median Stock in the Index Are Two Different Things

Contents

1. What the episode covers
2. ”The whole market is awful” is a sampling problem in your own holdings
3. Alphabet lost 500 billion dollars in a day, and three days later had not bounced
4. Apple became the hedge against AI, and the bill arrives as memory
5. Down nine points after hours, flat by the close
6. Rates get into the cracks, and the cracks have names
7. An opinion will not turn into a position
8. My holdings are bleeding — has the market broken?
9. The report was good, so why is it down nine points?
10. I was right but I sold early — does that count?
11. Worth a look
12. The one thing to take with you

The thing I fear most is fear itself. It is harder to bear than the accidents it warns of.

—— Montaigne, “Essays”, “Of Fear” (1580; my own translation)

On the What Are Your Thoughts livestream from The Compound and Friends on 4 November 2026, Josh Brown and Michael Batnick put up a chart that stopped me: two thirds of S&P 500 members have had a 20% drawdown from their highs this year, and that share sits right around the historical average. The tech column reads 93% — the only names that avoided a bear market this year were Apple, Nvidia, Cisco, Teledyne and NetApp. In the same week, the equal-weight S&P beat the top 20 by market cap by the widest margin since 2009. Alphabet, which they spend a long segment on, shed roughly 500 billion dollars of market cap in a day and sits 30% off its high, while Apple closed higher on 65% of the days the Philadelphia semiconductor index fell 2% or more. Every figure here is a reading from that day’s session and after-hours; AMD and Toast had reported minutes earlier, and the reading could look different a day later.

What the episode covers

What Are Your Thoughts is the Tuesday evening livestream Josh Brown and Michael Batnick do together. Both manage money at Ritholtz Wealth Management; Josh handles trading and single stocks, Michael brings the charts that contradict him. This one was recorded on US midterm election night, and they push the politics aside in the opening minutes — their argument being that the election outcome does not change what the quarter is actually pricing, which is AI capex and AI financing.

The backdrop was the tail of earnings season: Alphabet and Meta had already been hit, Amazon had dropped after hours and recovered, Apple and Palantir reported the night before, AMD and Toast reported during the stream. The semiconductor ETF (SMH) closed below its 200-day moving average that day, ending the fourth-longest streak above it in more than a decade; the memory-linked basket was briefly 40% off its high. And then they spend half the episode on something unrelated to prices: how much ordinary people have come to hate the subject of AI.

”The whole market is awful” is a sampling problem in your own holdings

Michael asked chart kid Matt one question: how many index members have had a 20% decline this year? The answer, two thirds, overlaps with the historical average. Broken out by sector, tech reads 93%. So the feeling that single stocks have been brutal this year has a real source — the source is concentrated in one sector.

Another set of charts from the same week runs the other way. The 21-day average of advancing issues in the S&P 500 just printed its highest reading of 2026, and the equal-weight index beat the top 20 by the widest margin since 2009. Josh’s explanation is the most useful line in the episode: nobody owns the names that are going up, and nobody in the media talks about them, so nobody feels it. The P&L in front of you is computed from the dozen or so names you picked. It is a sample of your choices.

Alphabet lost 500 billion dollars in a day, and three days later had not bounced

Alphabet’s drawdown reached 30%, which puts it alongside the tariff selloff and the Covid selloff. The cause was a miss in both search and YouTube. Josh’s mechanism is clean: the monetisation rate on AI mode is lower than on traditional search, because a user who types into the box and wants the answer gives you no links to click, and the ad revenue goes with the links. Alphabet is doing this to itself on purpose.

What makes the segment sting is another number: Google search is having its best first nine months since 2021. Michael’s rebuttal is one sentence — the market cares about this quarter and the next one. He also brought a figure he had picked up, that OpenAI now handles something like 20% of Google’s search volume, and said that if there is a real crack in the search business he does not want to own the stock down 30%. The control group: Apple, Palantir and Amazon all gave back most of their post-earnings losses the next day, and Alphabet had not bounced once in three sessions.

Josh then offers something I value more than any of the charts. In the summer of 2025 both of them worried that Alphabet would be the first major casualty of AI, and he sold at 180 and watched it run to 400. He asked Michael on air whether he was right in the end. Michael’s answer: stop.

Apple became the hedge against AI, and the bill arrives as memory

Jared Blikre’s chart for Yahoo Finance: on days the Philadelphia semiconductor index fell 2% or more, Apple closed higher 65% of the time this year. The next chart shows the change in capex over four quarters — Apple up 10 billion dollars, Amazon up 215 billion. In an XLK where almost everything lives or dies on data centre spending, Apple is the one not playing, which makes it where a large-growth manager goes when they want out of the others.

The bill shows up in components. The Wall Street Journal reported Apple expects to spend about 25 billion dollars on flash storage this year, against roughly 6 billion for the same thing a year ago. Tim Cook called rising memory prices a hundred-year flood. Michael’s follow-up is the part I keep thinking about: iPhones had a record quarter, so the consumer absorbed the price increases — how much further can they be pushed? Josh thinks the comps get easier in the first half of next year once those costs are in the base. He also points at the new touchscreen MacBook, around 900 dollars above the prior model, and says he does not know who is buying that.

Down nine points after hours, flat by the close

Amazon fell 9% after hours, opened down 9%, and closed down 1%. Palantir fell 9% after hours and closed down 1.5%. Apple gapped down 5% and closed down 2%. Josh called the Apple reaction an adult reaction — the people buying Apple this quarter were not buying this quarter’s growth, they were buying the stability of the earnings.

One Palantir number deserves to be written down on its own: commercial revenue went from 397 million dollars a year ago to 984 million, up 148% year over year, against 953 million in government. Inside one year the commercial business passed the government business, which retires the description of this as a tech defence contractor. Michael pre-empted the bear case himself, said “it can’t be sustained,” and then added that the Atlantic Ocean will eventually run dry and become a dust bowl, and between now and then they are doing the thing.

Rates get into the cracks, and the cracks have names

Kevin Warsh told the Wall Street Journal on 28 October that he had been sceptical of the narrative that tighter monetary policy would not touch AI, because interest rates get into the cracks and affect almost everything directly or indirectly. Michael’s first reaction was to disagree — he does not think today’s rates are driving hyperscaler prices. Then he reversed himself live: the real impact is a cost of capital high enough that a company says it will stop issuing debt and raise equity instead, and equity dilution is the thing the stock market likes least.

The cracks have names. Two weeks ago the CEO of Nebius said certain funding is simply closed, and the whole neocloud basket went down with it. CoreWeave’s drawdown reached 60%. Against the Wall Street Journal weekend piece on everyone racing to insure their AI bets, Josh’s reading is that 800,000 put contracts traded over two weeks is behaviour you do not see at the top of an actual bubble.

An opinion will not turn into a position

The back half of the episode has a segment I replayed. Josh went on Scott Galloway’s show last week and the YouTube comments ran to a thousand, and he says he has never seen that level of revulsion — aimed less at him than at the letters A and I. His read is that AI is now absorbing everything people feel about their own lives: the job that gets taken, the data centre that pushes up their power bill, the company that profits by laying people off. Both of them grant that the anxiety has real content, and that the labour market will be affected.

Then he goes after the loudest bears. If you know in your bones that this is 1999 about to turn over, he says, you are on your way to becoming a billionaire — what are you so angry about? He names a neocloud ETF, NCLD, 44% off its high with 84 million dollars under management, holding Treasuries used notionally against swaps on private neoclouds, bitcoin data centres, and penny-stock names. His line: an opinion earns reputation capital, not dollars, so show me the position.

I did not enjoy hearing it, because I hold several views I have talked about for years and never bet on. The force of the segment is that it applies far beyond AI.

My holdings are bleeding — has the market broken?

I ask myself this every few months. The episode’s way of taking it apart starts by separating two things: the experience of the median index member, and the experience of your list. The median experience this year is close to historical normal, and the tech column reads 93%. If 80% of your holdings are tech, your felt experience is accurate — it is measuring your choices rather than the state of the market.

One layer down, the gap has a cost. When you believe the whole market is collapsing, you take actions designed for a collapsing market: cut exposure hard, stop contributions, move to cash. And the equal-weight index beating the top 20 by the widest margin since 2009 means the money went somewhere else during exactly that stretch. Eight of my own holdings are tech and one is communication services, so this episode works on me as a mirror rather than as a conclusion I can carry off.

The report was good, so why is it down nine points?

Josh gave a decomposition I plan to keep. The price you see in your brokerage app is an equation with three variables: today’s fundamentals, sentiment toward the company, and what people are guessing about unknowable future fundamentals. Layer on leverage and you get a fourth, equally unknowable variable — who has to sell, and who is forced to buy.

That decomposition explains the memory basket. Michael’s judgement is that the move was positioning and leverage rather than fundamentals: double-levered ETFs, options, crowding. His example is Korea this summer, a whole country’s stock market tangled up in two or three tickers until the regulator had to write new rules on trading and leverage. So when a stock drops 9% after hours and closes flat, the better question than “what did I miss in the filing” is how much of those nine points came from the fourth variable. It also moves the moment of reckoning later: the price on day one after earnings is not the market’s view, only a record of who happened to be holding.

I was right but I sold early — does that count?

Josh selling Alphabet is the part I sat with longest. His 2025 worry became this quarter’s reality: search really does have a monetisation problem now. He sold at 180, the stock went to 400, and it is back near 280. When he asked whether he was right in the end, Michael’s answer was that as a trader, no.

My own approach is to keep two separate ledgers for the thesis and the trade. The thesis is judged on whether the mechanism happened the way you said and on the right time scale; the trade is judged on the return over your holding period against the volatility you carried. When the ledgers get mixed, the usual result is using later events to vindicate an early exit, and then exiting earlier next time. What Alphabet taught this quarter is about time scale: a structural worry can be correct, and the moment it pays off can arrive late enough to make you wrong twice over first.

Worth a look

• The What Are Your Thoughts livestream from The Compound and Friends, 4 November 2026, the source for this piece
• The Barclays equity tactical strategy note from election night on the midterms and the AI capex narrative
• Tony Pascarella’s trading desk notes at Goldman Sachs, including the S&P ex-AI basket and the observation that earnings growth has peaked
• Adam Parker’s list of index names down more than 10% in a single week
• Steve Deppe’s sample of the ten times the Nasdaq Composite printed a bearish outside reversal month from an all-time high
• Jared Blikre’s Yahoo Finance data on Apple behaving as a hedge against the semiconductor index
• Kevin Warsh’s Wall Street Journal remarks of 28 October on rates getting into the cracks
• Aswath Damodaran’s valuation of Anthropic’s equity, and the Hyperliquid perpetual futures prices on Anthropic and OpenAI
• Matt Levine’s column on prediction markets and perpetual futures

The one thing to take with you

A strong opinion that has changed none of your concrete actions is an emotion wearing an analytical mask.

That is where Josh’s segment earns its keep. He does not try to prove the AI bears wrong; he hands over a self-check. If your view holds, something in the world should look different because of it — your position, how you spend your time, what you said to whom. If you cannot find that thing, the view is still at the feeling stage. Several of mine have lived at that stage for years.

Here is something I have tried that works outside investing. Write down one strong view you have stated to someone at least three times in the past month — house prices, which company goes under, which institution fails, whether a friend’s marriage lasts. Underneath it, write one line: assuming I am right, what did I do differently this month because of it. If you can fill the line, keep the view. If you cannot, write one word beside it: emotion. Then glance at that word the next time you are about to say the sentence.

Raw Markdown

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.