Surging stockmarkets are powered by artificial intelligence
economist.com
economist.com
"On its own, the price of Nvidia is responsible for an enormous slice of the stockmarket recovery. Since the end of November the firm’s market capitalisation has soared from under $400bn to $925bn—accounting for a fifth of the rally. Add Nvidia’s surge to the growing market capitalisations of the 13 other firms with ai exposure and a remarkable 73% of the broader rally is explained."
"In November the average price to current earnings multiple of an s&p 500 firm, excluding the 14 most exposed to ai, was around 27. As we went to press, the multiple had dipped to 26. Meanwhile, the average multiple of firms in our ai bucket had leapt from 43 to 77."
Not sure what you would expect - they just point to the fact that most of the rally is achieved through the 'AI' companies - the rest of the index is actually down, or mostly down. That's fact. What more do you want to be more persuasive ?
A more reasonable explanation is that there was an overcorrection in big tech valuations during the downturn a couple years ago (while they are all continuing to make record profits quarter after quarter), and now after layoffs and cost cutting the market is slowly normalizing in their favor.
Nvidia's rise is much more directly tied to this.
> [...], but it's not at all clear that their lead is going to stick long enough to turn it into anything
People are saying that since January, but here we are in July and GPT-4 is still the best model. While everybody is catching up, they have (1) the processes in place to come up with new ideas and get them working (2) a head-start of many months (note that GPT-4 was basically done in Sep 2022).
Six months is... not a long time. Nobody knows where this is going to be in a year, or two, or five.
Certainly true, but their stock price is goosed by their hype, which is "AI AI AI!!!"
By historical standards, current P/E ratios are extremely high - especially for companies of this size.
And Google's rally corresponded to their very successful set of announcements in the AI space at Google IO (and also a profit expectation beat to be fair).
Yep. Like a neural network powered search engine, right?
> In zero interest rates you can instead make investor hype your product so they increase the share price, but it doesn't work forever.
As this article points out, investors believe.
They've already got that - it's those text snippets. They're quite unreliable, so a reliable one would be different of course.
But Google doesn't get paid to provide search results, they get paid to show ads you click on. It doesn't seem to me this is the same product as AI that shows you text summaries of search results.
Worse, if you mine all the content out of search results instead of making users click on them, why will people even make new webpages for you to show in the results?
I actually meant the neural ranking of results, and the selection of ads for result sets.
At least that would be somehow entertaining.
None of these recent surges feel sustainable but who knows.
Additionally, gas prices are to pre-war levels (https://www.gasbuddy.com/charts), natural gas prices are back below pre-war, actually below pre-Covid levels (https://www.eia.gov/dnav/ng/hist/rngwhhdm.htm). Lumber is back to pre-covid levels (https://tradingeconomics.com/commodity/lumber). Milk is back to pre-covid levels (https://tradingeconomics.com/commodity/milk).
Meanwhile unemployment is at the pre-COVID baseline already, historically low 3-4%, below the traditional 5% "full employment." Real wages (i.e. adjusted for all that inflation) is ALSO above pre-covid levels. https://ycharts.com/indicators/us_real_average_weekly_earnin...
House prices aren't increasing like they were, and the interest rate hike has made it expensive to buy a new house, but overall, it's really hard to claim people are being drained right now. Almost any other time has had worse employment, etc.
That doesn’t seem significant and if you look at the shape of the graph it’s trending down from the real wage gains during the height of Covid. December 2020 shows $398 per week while today shows $378 per week, so while it’s still above a pre Covid level of $376 I don’t think this is a good signal of recovery.
It's impossible to do a real counter-factual comparing what would have happened, but:
Dec 2018: $376.20
Dec 2019: $376.87
COVID
Dec 2022: $378.76
Apr 2023: $378.50
So real (inflation adjusted) wages have been roughly static compared to before COVID. That seems a reasonable point to make compared to the OP's negative sentiment, which seems not to be backed up by the numbers.
Consumer Price Index for All Urban Consumers: Food in U.S. City Average https://fred.stlouisfed.org/graph/?g=166b0
Consumer Price Index for All Urban Consumers: Rent of Primary Residence in U.S. City Average https://fred.stlouisfed.org/graph/?g=166b6
Have nominal wages really kept up with what we see in these charts? Is there a report using selective data to confirm that assertion?
https://ycharts.com/indicators/us_real_average_weekly_earnin... Overall I’d still say that between low unemployment and wage growth keeping up with inflation, we’re doing better than nearly any other time in history.
>> Over the year, food prices rose 7.4 percent. Prices for food at home increased 6.2 percent since a year ago, with higher prices in all six grocery categories. Prices for food away from home increased 9.5 percent.
https://www.bls.gov/regions/west/news-release/consumerpricei...
Not sure where you live but where I live it’s double the price.
Real wages ran flat from Q4 '21 through Q1 '23 [1]. Real disposable personal income also doing fine [2].
I don't mean to minimize the real struggles folks in our cities, or in tech, are facing. But it's not the whole picture.
I feel like you could make the argument "wouldn't the stock market not be going up then if this is the case?" Wouldn't people have less free money to spend and therefore corporations would receive fewer of their dollars and post less profits?
Yet profits continue to grow.
aka, yet another reductionist dramatic wrong outlook online that's out of line with reality
yay internet! keep pushing that narrative though
But a human brain fart is probably just as likely.
It doesn't seem to continue to sustain the same explosive growth as TikTok, but to be fair, it wasn't intended to be a full consumer product to begin with.
This is why it's so hard to beat the market and why conventional wisdom keep failing: people were expecting high inflation to hurt stocks, or expecting rate hikes to hurt stocks, or bought meme stocks which crashed and burned, or bought crypto instead of stocks. Staying invested in index funds regardless of what the macro situation is tends to produce the best outcome.
Is any of those ideas truly conventional wisdom, or just popular culture among some social media sites?
> Staying invested in index funds regardless of what the macro situation is tends to produce the best outcome.
Are the majority of people _not_ doing this?
> Is any of those ideas truly conventional wisdom, or just popular culture among some social media sites?
These are absolutely conventional wisdom. It's as conventional as you can get, and the opposite of crazy ideas you see on social media.
However there are two better strategies than index funds. They are 1. get married and 2. buy a house.
It's hard to beat the market, because people have way too short horizon. Take your example here - it's absolutely meaningless what's up or down or flat over a 2 weeks period...
The reality in the stock market is, that you can make money investing long term (let's say >5 years horizon) in good or great companies, running well a very short term strategy (market making, hft, arbitrage), but usually you cannot do any money when your horizon is somewhere in between. And the issue is, for most people, that's the case....
But somebody's on the other side of those trades.
But it doesn't really matter if the percent who lose on net is 50% or 51% if you're at the 75th percentile. And a large plurality of people are. You don't have to be in the top 0.1% to come out ahead.
The reason for the usual advice is that 50% of people are below average. And then they get mad when the market proves it to them, so the advice that keeps them satisfied is to buy an index fund.
The average of 1,1,1,1,1,37 is 7. If these are % yoy gains of traders who try to beat the market.
The average of 37, 8, 8, 8, 8, -27 is also 7.
1.43988^(1/6) = 1.06264
So, really 6.26% CAGR.
Not a huge difference here. But it is lower. The difference really shows up when you start seeing losses. The extreme case of one year with 100% loss demonstrates things.
So most people loose money, and a rather small group (say 10%) is raking up most of the gains.
I think it's pretty common knowledge that most funds and stock pickers underperform the benchmarks. I think for retail investors the ballpark figure is like 90% underperformers.
Warren Buffett took a bet in 2008 that over the next 10 years, a low cost S&P 500 index fund will outperform active hedge funds - and it did.
Granted, it's not a proof or anything, but it's interesting: https://www.investopedia.com/articles/investing/030916/buffe...
But people often get ahead by understanding an industry. For example, when Stable Diffusion became available last year, someone who was paying attention might have noticed that it was likely to become popular and spike demand for GPUs to run it on. If that person went out and bought some shares of NVDA or AMD at that point, they'd be happy with where it is today compared to an index fund.
But a major index fund is unlikely to temporarily change the ratio of those companies in the index in response to that kind of information, so the higher returns are only available to people who are paying attention and willing to take the risk with their own money.
I completely agree that it is possible to beat the market - it's just that it's very difficult, mostly due to misaligned time horizons.
Take your insight here about NVidia - it's false. So far, GPUs sales DID NOT increase, at least not significantly. Go check NVidia sales for past 4 quarters. They said they expect them to increase significantly in the next quarter. Even if it does, and even if it doubles what they say, the valuation of the company is still in the bubble territory.
Another great example is Tesla - stock surged on anouncement that GM will use their chargers. When you look at the numbers, they will make an insignificant money on that (optimistic scenarios are between 3-10 billion $, by 2030 !) - yet the stock added more capitalization than ... the entire captialization of GM !
Now, if you say that NVidia has a huge business moat that is untouchable for the next 5 years - that's a different story.
I’m imagining a kind of “inbreeding” where AIs are doing fast trades based on signals other AIs are giving, but it’s not my expertice.
"Regular traders" are gamblers whose job is to give market-makers their money in trade fees.
for the retail people who are bad at actually discovering prices, their buying of index funds instead actually makes the market more efficient.
But they'll make similar recommendations to many people, so those stocks will go up, and people will think it's working. Then some new AI will be released that makes different recommendations and the stocks artificially inflated by the obsolete AI will crash. People will say that you need the latest AI or you'll be left behind.
This is not a new thing. It's how every commodified stock-picking mechanism works. By the time retail investors have it, you're already inside a bubble.
I think at the moment a lot of old-er financial people are quite sceptical of complicated black-box models.
AI could be a huge win in the sense that you could say "When has this bond future been in [conditions] like now and what happened". The potential for organising information is fantastic.
No paywall…
Any firm worth their salt has risk systems in place to prevent runaway trading too, and those aren’t black boxes like AI/ML. They’re limits set for specific strategies or teams etc, and typically require approval from firm partners to increase past certain levels.
Collapses of firms are far more often caused by lapses in proper risk management than bugs in code. The KCG collapse looks like a bug on the surface but was ultimately a misuse of accounts that sidestepped these risk systems, so they couldn’t stop the runaway caused by the bug.