Hacker News

The AI sell-off compared to well-known stock market crashes(kadoa.com)

91 points by hubraumhugo 1 day ago | 75 comments

The NASDAQ dropped 5% out of nowhere.

It wasn't midterms. Everybody expected that the Democrats would take the house and senate, it should have been priced in. The results of the election played out on election night on the futures markets overnight and into the next day. Thursday was something else, perhaps some fund blew up.

[–] 2OEH8eoCRo0 12 hours ago

Everybody expected the House but not the Senate. The way the Senate flipped and by so much was unexpected.

[–] fakedang 2 hours ago

Senate and governorships, which now puts a lot of DC projects at risk.

[–] bensyverson 1 day ago

I guess we'll see if it's really a sell-off, or if it was normal market volatility around an election.

[–] mikeweiss 1 day ago

There's not much normal about the market last 5 years...

[–] metaphor 1 day ago

I wonder how much of that volatility is directly attributable to SPX 0DTE being a daily thing since circa 2022.

[–] spwa4 13 hours ago

Not much normal about the government in the past year either.

[–] sapphire4 5 hours ago

I think we're overdue for a big correction. There are just too many companies keeping the same "we’re growing so fast, so we’re not profitable yet" line for 10+ years now.

[–] charlieyu1 3 hours ago

If it takes 10 years then it becomes the new normal.

I definitely hadn’t expected a big correction, and would expect the market recover and recoup the losses soon.

The global events of last few years created a lot of inflation, and the new money has to go somewhere. A large part of stock market gains are the new money disguised as growth.

[–] paulpauper 1 day ago

It really is a sell off. that much is obvious. how long it lasts, is the next question. If history is any guide, not much longer. Rebounds tend to be very fast and sudden. See 2020 or 2025 for example.

[–] binoct 1 day ago

2020 isn’t exactly history in market terms

[–] lifeisloving 1 day ago

It doesn't fit any normal distribution actually.

[–] ViscountPenguin 1 day ago

Normal as in "typical", not as in "Central Limit Theorem".

Wiener processes aren't a completely terrible model for markets though.

[–] paulpauper 1 day ago

I think it's a fat tail or jump diffusion process. There is a ton of literature about this over the past 30 years.

[–] rmah 1 day ago

There are always bears who are predicting the next stock market crash. They always have various reasons that make sense on the surface. They're almost always wrong. But they're always eventually right.

[–] CoolestBeans 1 day ago

When it comes to finance and fashion there's no difference between being early and being wrong.

[–] ghtbircshotbe 16 hours ago

Maybe that's true if you sell too early, but if you buy too early then wait you'll generally do okay

[–] illiac786 3 hours ago

What is “buying too early”? Is that even a thing?

[–] howunfortunate 1 day ago

> But they're always eventually right.

Can you really call it "right" if the market goes up 40% before eventually going down 15%?

[–] chii 1 day ago

This is why you don't rely on broken clocks to tell you the time!

[–] 0xDEAFBEAD 19 hours ago

Down 15%... so far

[–] estetlinus 1 day ago

Are you implying the market will go up ad infinitum?

[–] bdamm 1 day ago

The "race to the bottom" (to devalue fiat currency and drive inflation) is a big reason why stocks, gold, real estate, and bitcoin are valuable to hold. The market will definitely go up ad infinitum, since that is the policy of all central banks.

[–] dd8601fn 1 day ago

It’s just a take on the old, “They predicted twelve of the last three crashes.”

If you can’t act on predictions they’re not meaningful.

[–] ngruhn 1 day ago

or picking the guy with the right prediction is not much easier than making correct predictions

[–] solyin 1 day ago

In the AI era, you just spin up 1024 bots making substack posts prophetizing the future. That's ten binary predictions. Delete the 1023 that fail. Keep the one account that perfectly "predicted" the future.

[–] xrisk 1 day ago

The derivatives market wouldn’t exist if there wasn’t a legitimate economic need to insure against unexpected crashes. Aka it’s not just bear speculation; crashes happen all the time in markets around the world.

[–] paulpauper 1 day ago

Even the Covid crash, as bad as it was, was over in a month. Stocks rocketed to new highs faster than anyone could have ever anticipated.

[–] leonidasrup 1 day ago

This stock recovery was so fast because of an external factor, governments injected money into economy, but the price was inflation.

" Fiscal stimulus during the pandemic contributed to an increase in inflation of about 2.6 percentage points. "

https://reason.com/2023/02/02/covid-stimulus-spending-played...

[–] charlieyu1 3 hours ago

It is also inflation this time. Inflation hasn’t slowed down for a few years now

[–] fauchletenerum 1 day ago

The crashes seem be getting smaller and more frequent

[–] shalmanese 1 day ago

I think that's just you noticing smaller blips in the market as you pay more attention. Markets are random walks so they exhibit a fairly fixed distribution of different sized crashes. If they deviated, someone could notice that pattern and trade on it, returning the market back to a random walk.

[–] eru 1 day ago

> Markets are random walks so they exhibit a fairly fixed distribution of different sized crashes.

Market prices being unpredictable random walks doesn't mean that the volatility or shape needs to be constant. To make the notation more precise, you need to talk about (sub-) martingales.

> If they deviated, someone could notice that pattern and trade on it, returning the market back to a random walk.

This is true, but "fairly fixed distribution of different sized crashes." doesn't follow at all.

[–] fauchletenerum 1 day ago

Or it could be a sign of a greater amount of financial intervention to preserve the value of the stock market after crashes, particularly since most people are largely counting on their 401k/IRAs to maintain their value in their retirement years

[–] therein 1 day ago

When does it give birth to a new market?

[–] eru 1 day ago

Which would be progress.

A small sell-off is just business as usual.

[–] lifeisloving 1 day ago

Who called this small? Datacenter stocks are largely down 50% this month.

[–] fizwidget 1 day ago

The further you narrow down to specific industries/companies the more volatile things get. That’s hardly surprising.

The discussion here is around broader stock market indices.

[–] lifeisloving 1 day ago

Who said this conversation was about broader stock market indices?

The title literally has AI in the title, the conversation is centered around an "AI selloff", as its been labeled, that happened this week.

Roughly 40% of stock market gains since 2023 are tied to less than 10 companies with massive AI investments.

The entire conversation is about this sector, and most of the gains in the "broader indices" that nobody was discussing, are largely attributed to AI invesment and speculation.

[–] fizwidget 1 day ago

The article is all about movements in the S&P 500 though. Yes it’s an “AI selloff”, but looked at from the perspective of the broader stock market.

[–] zardo 1 day ago

The cheap debt is drying up for neoclouds. New bonds are going to be at payday loan rates.

[–] ion_trapper 1 day ago

The beauty of increasingly efficient markets.

[–] adrianN 1 day ago

I don't think the efficient market hypothesis says anything about the convergence time or the path to convergence, but I'm not an economist.

[–] pedro_caetano 1 day ago

As someone not in finance, what actually leads to this pattern (of smaller blips followed by faster recovery)? Is this a consequence of central banks knowing more than before? Algorithmic trading?

My intuition is that financial markets seem increasing dissociated from any correlation with street level economy or the financials of traded companies?

[–] ctmlr 1 day ago

If it's so efficient how come we can't seem to build enough houses... or build anything anymore for that matter?

[–] lelanthran 1 day ago

It's a bit early to be calling it, I think.

A sell-off is required to start a panic-sell and bubble-pop, but not all sell-offs start a panic-sell and bubble-pop.

It's hard to tell in advance what this is.

[–] Animats 1 day ago

Yes. This is just the first day.

It's going to take a year or two more. Will the humanoid robots being built be useful? Will Tesla's robotaxi ever work without a huge number of remote monitors? Will the free models start to displace the expensive ones? Will someone figure out a way to do all this at lower cost? We don't know yet. But there are too many huge companies with rather inflated market caps. Some of that has to be air.

[–] gizajob 1 day ago

Will we still need ever-larger GPUs and Memory? Yes.

There’s no version of the immediate future where this isn’t the case.

[–] rolisz 1 day ago

Really? I find local models such as Qwen 3.8 and Muse Glimmer to be pretty good. Not Opus or Sol, but.... Close to Sonnet maybe? So if I had to only use the local models from now on, it would not be a huge loss. Granted, I can't work on parallel on multiple tasks locally. But it's still 48 GB of RAM vs 750 GB or what is required for GLM5.3

[–] gizajob 1 day ago

You’d happily take a TB of RAM if offered though. The fact you can’t get it is evidence that hardware manufacturers have a huge amount of life left in them and the current correction is a buying opportunity.

[–] lelanthran 1 day ago

> There’s no version of the immediate future where this isn’t the case.

Pity you qualified your prediction with "immediate"; a simimlar sentiment was around just before the dot-com crash: "There's no version of the future where we won't be able to use all that extra network infra."

Turned out, in the immediate future, just post-crash, we actually didn't need them. They only got used a decade later.

Will we need every larger compute and RAM?

Sure?

Will we need it in a post-crash scenario?

Nope!

[–] gizajob 1 day ago

So we have enough RAM kicking around in the world today that will suffice for the next 5-10 years?

And an RTX5090 is going to drop down to a few hundred dollars is it? Just because the price of stonks is temporarily correcting downwards?

[–] lelanthran 1 day ago

> So we have enough RAM kicking around in the world today that will suffice for the next 5-10 years?

No, but we may have enough GPU (with their RAM) sitting around in datacentres that our AI needs will be taken care off for the foreseeable future.

I mean, RAM is not really fungible - these valuations are not tied to RTX5090s, they are tied to the server GPUs.

The investment was not used to buy RAM sticks for desktop and server m/boards. ANthropic did not use their VC money to buy RTX5090s. The VC money went to power generation, datacentres, and server GPUs with 768GB of RAM.

So, yeah, if the bubble pops, we're well taken care of in the immediate future by the stuff the VC money already bought.

[–] gizajob 1 day ago

And they’re gonna let you access that super cheap at fire sale prices yeah? Because of the oversupply. Right.

And what about local AI and agentic stuff? We can all stick with 32Gb of ram at home and piddly lil GPUs?

In my opinion this turmoil is hedge funds and market makers flushing out liquidity, same as they always do. Paying Bernie Sanders to tweet “no more datacentres” is pretty cheap to do.

Extreme fear = always a buying opportunity.

[–] eli_gottlieb 6 hours ago

> Will Tesla's robotaxi ever work without a huge number of remote monitors?

Tesla? Waymo is operating in American cities right now. I've taken them.

[–] 0xDEAFBEAD 19 hours ago

Shiller P/E was approaching dot-com levels:

https://www.multpl.com/shiller-pe

I think a big bubble pop is plausible.

[–] sumeno 1 day ago

That seems to be the entire point of this page.

Everyone either misinterpreting what it says or only reading the headline, because it's pretty clear that the author is arguing that only 1 in 5 sell-offs equivalent to the current one end up being significant.

> The recent AI sell-off brought a lot of fearmongering headlines. Put next to past stock market crashes, it's only a mild correction for the S&P 500, 6% below its record.

[–] usernomdeguerre 1 day ago

Or this is a correction that has already happened in other markets, and is now unwinding here:

https://www.reuters.com/world/china/china-ai-bubble-has-alre...

[–] yipinwong 1 day ago

both the media and the post is making a deal out of nothing.

When the author is right, he will be like, "i told you so", if wrong, "It's hard to predict" and shrug off.

Don't buy into crap post like this.

[–] yxhuvud 1 day ago

The posts entire point is that it is too early to tell.

[–] pamcake 1 day ago

Isn't it overly bullish plus quite early to refer to this as "The AI sell-off"?

[–] devin 1 day ago

Do you mean bearish?

Maybe. Nobody can see the future, so I agree it's kind of arbitrary to call this the sell-off, but plotting it against these other events over the coming months may be interesting, and you have to start somewhere.

[–] DonsDiscountGas 1 day ago

The bullish part is assuming it's over. Which the post kinda implies.

[–] cgio 1 day ago

Maybe it’s because a behemoth ipo is coming and investment bankers are at work.

[–] jasonkhanlar 1 day ago

Reminder NVDA market capitalization skyrocketed around the time Brett Harrison and Leopold Aschenbrenner exited from FTX.

[–] gizajob 1 day ago

Correlation /= causation

[–] estetlinus 1 day ago

Burry recommended me to buy Build-A-Bear, and it’s actually up! Where are your gods now, bulls of HN?

[–] Lerc 1 day ago

I have seen a few YouTube Thumbnails declaring the bursting of the AI bubble. It feels like they are so desperate for it to happen that they are not prepared to wait a week or two to actually have some real data to work with.

It's not like people will forget that they were predicting it. If it is inevitable it will happen, if it isn't inevitable it still might happen. If it bounces back it still might happen six months down the track. Declaring it to have already happened seems reckless if your intention was to inform. We will know soon enough. I guess at least they have staked their credibility on something falsifiable. It won't make them retroactively less reckless if they turn out to be right but it'll make them seem particularly foolish if they are wrong.

[–] lconnell962 1 day ago

Between it and the "next housing bubble" getting predicted I almost wonder if there's some kind of stock market psychology to bullheaded resistance in the face of "it will fail" predictions.

That on something big if you have enough people constantly predicting it's upcoming failure people just keep investing anyway keeping it going. Never really sure if something is on life support or not.

[–] nirui 1 day ago

Can someone explain how to read this? Since "past performance is not indicative of future results" and "technical analysis is astrology for men"? I'm assuming these data is useful in some way, just don't know what.

The tech oligarchs are certainly panicking, given how they are acting cute now, like "space laser didn't work lol" and "Ram Altman" etc.

But I'm guessing, the market is panicking because it don't know what exactly the democrat would do to AI developments, so maybe some clarity from the party could fix it. More clear than "lowering living costs" that is.

[–] relaxing 23 hours ago

My “AI sell off is nothing compared to the 2008 financial crisis” website has a lot of people asking questions already answered by my site.