The first Weekend Reading every month can be read by anyone on the Monevator website. Subscribe for free to our email newsletter or become a member to ensure you see the rest.
What caught my eye this week.
Would you be happy handing over the reins of your portfolio to a robot? Given most of you will be regular Monevator readers and email subscribers, I can guess the answer – if not the specific gentle expletive added for colour…
Of course, the typical Monevator reader (rightly) invests passively in index tracker funds. And those funds are managed by software – albeit usually with some kind of human oversight to determine which companies go in and come out of a given index, as we saw with the recent controversy over SpaceX.
However it’s one thing to use software to follow a well-established and diversified benchmark via what’s now very mainstream index fund investing. It’s another to toss the keys to a novel AI agent with a cheery, “have it it, call me if you blow the kids’ inheritance!”
Okay, in practice any self-driving portfolio is going to have guardrails. But even so, you can easily imagine countless robot investing edge cases that are the financial equivalent of a self-driving car facing a hotdog cart trundling into the road, or the driver in front falling asleep at the wheel.
Or consider the market madness proxy of gridlock and traffic jams, when movement (liquidity) evaporates.
Think back to the crazy ride that was the Covid crash. How would a cheapo trading robot cope?
Investing under the AI influence
Naturally, just because we don’t need self-driving portfolios, that doesn’t mean we won’t get them.
Innovation in financial services is driven by what sells, not what is good for us.
Only this week CNBC reported that:
Larger brokerages are moving in [this] direction. Robinhood in May introduced tools allowing third-party AI agents to connect with customer accounts. Brokerage firm Public, meanwhile, is developing AI agents in-house that can automate investing workflows within its platform.
“What this era of agentic is doing … it goes away from just being able to research something by yourself and then make up your own ideas and then trade the way you’ve traded where it’s now becoming automated and where AI agents can actually execute investment strategies on your behalf,” said Leif Abraham, Public’s co-founder and co-CEO.
The article paints a breathless future of AI agents turning private investors into DIY hedge fund managers. There’s nary a mention of fees and costs, though – although to be fair the piece does conclude with caveats about the risks of letting Clippy 2026 trade stocks.
That latter sentiment is echoed by a blog from the CFA Institute, which reviewed the *cough* mixed results from research into trading via LLMs.
It concluded:
The evidence for multi-agent and LLM-augmented portfolio construction is promising. The failure literature does not invalidate this, but it does suggest that the gap between a research prototype and a production-grade institutional system is larger than the paper acknowledges.
The human overseer […] cannot yet take a purely passive safeguard role.
But who am I kidding? The reality is tens of thousands of retail investors are already experimenting with AI trading, whether through financial service scaffolding such as RobinHood or via the – hopefully judicious – interrogation of their nearest chatbot.
Top gear
As far as I can tell, this era’s Warren Buffett – part-man, part-machine, all alpha – has yet to reveal himself.
But if enough people do it then we’ll probably get an AI-enabled self-made trader billionaire someday, just thanks to the law of averages.
Famously, a few quant shops like Renaissance have smashed the market for years by force feeding gargantuan amounts of data into supercomputers. However that’s very different from Joe Day Trader setting a few rules in an AI-enabled investing account.
Yet even a few traditional stock picking active managers do beat the market, at least for a while, and no doubt so will some AI agents.
The odds have always been against it however – active investing is a zero-sum game – and AI cannot change that.
Have a great weekend.
From Monevator
The Slow & Steady Passive Portfolio update: Q2 2026 – Monevator
A deep dive into FX hedging – Monevator [Moguls]
From the archive-ator: Compound interest can save our pensions – Monevator
News
Chancellor announces his first Budget will be on 28 October [Sigh] – BBC
Bank of England holds rates at 3.75% as inflation fears mount – Guardian
One million more Britons set to pay income tax – Which
Ban foreign stocks from Isa wrapper, says top pensions boss – City AM
BP puts its North Sea business up for sale – This Is Money
London only English region to see population fall – BBC
It now takes 216 days to move home – This Is Money
UK millionaires fall to 442,000, lowest since 2008 – Business Matters
House prices up just 0.1% in July, says Nationwide – Mortgage Strategy
Jim Leaviss, Bond Vigilante, 1971-2026 – FT
We’ve moved from income world to wealth world [Paywall] – FT
Kospi’s boom-bust-boom mini-special
South Korean bubble bursts, erasing $2.2trillion in value – Mugglehead
Minister apologises as leveraged ETF investors suffer deep losses… – CNBC
…while AI fund Situational Awareness dumps holdings to Citadel… – CNBC
…and then the Kospi closed up a record 18% in a day on Friday – Korea Times
It’s all because the AI boom creates a lot of uncertainty – Noahpinion
Products and services
Disclosure: Links to platforms may be affiliate links, where we may earn a commission. This article is not personal financial advice. When investing, your capital is at risk and you may get back less than invested. With commission-free brokers other fees may apply. See terms and fees. Past performance doesn’t guarantee future results.
Is your annual travel insurance still worth it? – Which
The fake Spotify emails that put you at risk of fraud – Guardian
Paragon Bank cuts five-year buy-to-let mortgage rates – Mortgage Strategy
Get £100 to £3,000 cashback when you open an Interactive Investor SIPP. Minimum £20,000 deposit. Terms and fees apply, affiliate link – Interactive Investor
Does Saga’s best buy interest rate for over-50s live up to the hype? – Which
A sea view could cost you up to £220,000 more – This Is Money
Get up to £1,500 cashback when you transfer your cash and/or investments to Charles Stanley Direct through this affiliate link. Terms apply – Charles Stanley
European wildfires and travel insurance – Which
Save up to 47% on your home by doing the postcode switch – What Mortgage
Charming homes for sale with family gardens, in pictures – Guardian
Comment and opinion
Scotland’s 48p tax rate may be losing money – Tax Policy Associates
Britain has tried War Bonds before, and savers paid the price – CNBC
How to think about the ‘full price’ – Best Interest
What 125 years of data tells us about investing – Behind the Balance Sheet
The wickedness of wealth management – The Net Worthwhile Weekly
Market indicators – Humble Dollar
Now show Japan – A Wealth of Common Sense
The problem with optionality – Of Dollars and Data
Retirement income security and more [Podcast] – Morningstar
Naughty corner: Active antics
IPOs have been a losing bet since 2019 – Apollo
SpaceX, PE, VC, and quacking ducks – The Falling Knife
Investment wisdom culled from old clip outs – Cove Street Capital
Cashing in on Japan’s cross-shareholdings – Verdad
Copart: from scrap to scale – Fiscal.AI
Picking stocks in a bloodbath – A Wealth of Common Sense
When size falls short – Novel Investor
Kindle book bargains
What They Don’t Teach You About Money by Claer Barrett – £0.99 on Kindle
Taxtopia by The Rebel Accountant – £0.99 on Kindle
The Savvy Spender by Megan Mickelwright – £0.99 on Kindle
The World for Sale by Javier Blas and Jack Farchy – £0.99 on Kindle
Or read one of the all-time great investing classics – Monevator shop
Environmental factors
New solar panels in Great Britain at 15-year high as fuel costs soar – Guardian
French climate lawsuit a window into next global legal fight – The Conversation
AI, authors, and writing mini-special
How AI books sneak their way into stores – New York Times
ChatGPT is blocking requests to copy an author’s style – Ars Technica
AI has made the ‘dead Internet’ theory come true – Futurism
Robot overlord roundup
Anthropic’s soaring AI revenues compared to some famous other brands – Axios
What will more intelligence actually do for us? – Noahpinion
How to lose AI in ten days – Spyglass
Not at the dinner table
The US economy is just a VIP list now – Your Brain on Money
In defence of gerontocracy – The Argument
The masculinity scam – The Atlantic [h/t Abnormal Returns]
Donald Trump keeps losing the Iran War – Drezner’s World
The Putinization of the American military – Paul Krugman
Off our beat
Could a single pathogen bring down civilisation? – Next Big Idea Club
How a near-death experience led to finding sea dragons in Wales – Guardian
Sell the company for $400m? He’s giving it away instead – N.Y.T.
Why America’s super rich have embraced British football clubs – CNBC
The light narrows – Aeon
Poor countries are aging fast but can’t keep up with the cost – W.S.J.
Scientists rethink sun exposure risks and benefits – Scientific American
An uncomplicated man [On The Odyssey movie] – London Review of Books
How to exist – Raptitude
Who dares ridicule Gianni Infantino? – Guardian
And finally…
“Stop thinking about what your money can buy. Start thinking about what your money can earn. And then think about what the money it earns can earn.”
– J.L. Collins, The Simple Path to Wealth
Note this article includes affiliate links, such as from Amazon and Interactive Investor.
It’ll exist as a niche I think as a way to sell investing to people uncomfortable with complete passive (like wealthify does) – and therefore a good thing because at least it gets people to invest, it’ll be obvious that it can’t have an edge without human magic, it’ll be hard to justify it having high fees since competition can scale up, and it could be completely adamant about its strategy (unlike Terry Smith) as it has no job to keep, it might make markets behave that tiny bit more algorithmically
There could also be scam robots aiming to use your money for pump & dump scams, it’ll be hard to know what goes on under the hood of new ones
More trustworthy and predictable than humans
@matthew — I can uses for Monevator types who want slight tweaks. Sort of like direct indexing in the US.
For example, “replicate the US market without Elon Musk companies” or similar.
Obviously that’s active but it’s active lite and the endless emails we get asking about passive and ESG for example proves some demand…
What AI seems to do is crunch huge amounts data immediately -unlike previous methods that took time -possibly many days and used many employees
The human however still has to set the question-gets a rapid answer from AI-and then has to decide what to do with the answer
Two very human inputs therefore still needed reassuringly-at the beginning and the end of the process
Perhaps AI might restore the well educated polymath ie the best human archetype to the centre of decision making once again
xxd09
Having used AI (Claude) a lot recently on a range of more or less investing-related tasks, I’m a lot more positive about this.
For one thing, although there will definitely be some AI-enabled short term trading, which is unlikely to end well, the bigger value will be in well-thought out portfolio construction and management, and retirement and tax planning.
My own experience is that LLMs are pretty cautious and risk-focused, though of course everything depends on the prompts they’re given. They are great at explaining concepts and have a surprising (to me) ability to provide genuine insights from data.
Good use case could be periodically implementing predefined rebalancing rules for an otherwise passive portfolio
“Scientists rethink sun exposure risks and benefits” Aha, another medical reversal coming up!
Really, doctors and “scientists” should stick to tautologies. “Eat a healthy diet” is never wrong, whatever the current fashionable belief about what is or isn’t healthy. After all, they often base their views – or publicly expressed views – on evidence that is weak, non-existent, or downright crooked.
@dearieme (#6)
The article in Scientific American links to a book (which I’ve not read) – the blurb for which has two statements
‘For decades, a “zero-sun” policy has characterized our approach to sun protection’
and
‘In this incisive work, acclaimed journalist Rowan Jacobsen presents the growing case for the importance of modest sun exposure for our health and well-being’
I’ve never heard of a zero-sun policy (maybe it is a US thing, after all they are generally nearer the equator than the UK is). In the UK the NHS advice appears to be “Aim to strike a balance between protecting yourself from the sun and getting enough vitamin D from sunlight.” and “never burn”.
In other words, modest exposure is already suggested to obtain the benefits of at least one healthy outcome. So, not a reversal of the science (it rarely ever is except in the press!), but maybe a small shift in the balance between competing health outcomes (i.e., in this example underexposure or overexposure to the sun).
In the health context, your last sentence probably applies to the Daily Mail but not the vast majority of those undertaking the underlying science and the scientific literature.
I have been using ChatGPT , a less expensive paid for version, a lot !
I have a paid for Perflexity,( thank you Revolut Metal , it’s an included perk , with Nord VPN and the standard FT subscription, etc, all for £15 month, plug over ! )
I use it extensively for my part time consulting role. Good but don’t trust it too much with out checking …
This caught me out once, an authoritative tone can hide a missing technical procedural detail …
On the investing side I had it look over my Portfolio Handbook , ( what to do if I’m dead/incapacitated book for dependents)
That provided an interesting rewrite , it’s a work in progress ….
Following the house upsizing last year, ( went up in square footage by a factor of 4…. There was a good reason!) Subsequent property sales are rebuilding the pot but losing an almost 7 figure sum of ISA allowances was painful….
I have some very interesting ‘discussion’s’ with ChatGPT over portfolio construction, bonds , factor tilts.
Useful, but it likes to ingratiate itself with you , if it was an investment advisor you’d fire them for expressing a view and folding without sufficient pushback.
Throw in some mistakes….
What is also interesting is when you give a ChatGPT answer or policy to Perflexity or Claude…
Overall very useful , it’s investment advice was pretty sensible , conventional, would be ok.
@dearime #6
““Scientists rethink sun exposure risks and benefits” Aha, another medical reversal coming up!”
I say! Steady on!
This is an interview by a journalist of another author and journalist, in a magazine which now has a reputation for somewhat sensationalised science stories. It used to be better.
The actual article does not reflect the headline.
The summary of the article is: Being out in nature is good for you, some sun is good for you (vitamin D), getting sunburned is very bad, sun impact depends on your complexion, you need to balance risks and benefits.
Sounded pretty much like the advice I have been given over the years. So I checked the NHS website. That’s exactly what they say, except that they give specific advice on when the risks are highest and how to protect yourself.
Maybe it is a US thing? Well, the FDA says that skin cancer is the most common cancer in the US and 1.4 million people are living with melanoma. Having just lost a friend, a keen sailor, to melanoma, it is not a nice disease or a pleasant way to go. So perhaps they are right to warn as people are apparently not taking the advice. FDA recommend avoiding midday sun and sun protection. The NIH acknowledges the value of exposure to sunlight for vitamin D, mood, and biological clock regulation. It also stresses the importance of avoiding sunburn and gives practical advice on reducing risks.
I have not read the book, so I don’t know exactly what the author is implying about changes in view, but what he advises is exactly what the NHS, the FDA and NIH are saying, and in the case of the NHS, have been saying for years.
the reigns of your portfolio
reins
@Hariseldon
I can’t decide between downsizing again or possibly finding something suitable for 2-4 families with a decent amount of land. Can you elaborate on your 4x situation, and how did it work out….
Giving up tax benefits on £1m ISA couldn’t have been easy. I’d even consider renting but my wife is having none of it
@BBB — Oh dear, thanks. I actually had it right this time and then edited it to the mistake in proofing! I just have a blind spot to that word, among many others. (You should see me struggle to spell The Odyssey. Or algorithm!)
I hear the sound of the most massive mis-selling scandal in the distance…
AI lies to you, with aplomb and at random. It’s pretty hot, 80% of the time, but if you can’t qualify the 20% that’ll kill you.
Unlike your own cockups, you can’t really learn from AI’s mistakes, even if you catch ’em. You just have to go AI, you got that wrong. It’ll say good catch and start over.
Whereas if you screw up yourself, you’re in with a chance. I thought this would happen. It didn’t. Where is my model of the world or execution of the plan wrong? Or is this something that is fundamentally unknowable to me…
@ermine — I read this morning that Vlad Tenev of Robin Hood said a few days ago that their app now has 100,000 users deploying AI trading agents. As I understand it this is just the US so far, but the demand seems to be there.
I’m not sure I’d get anywhere near trusting AI (as it stands) with any of my cash, but I can see some use in the brokers opening up their systems to the API calls needed to make those automated calls work.
That would make “automating” something like a glidepath in early retirement much easier than doing it all manually (potentially) every month for a few years.
One or two are open enough already, but this might encourage the rest.
Thinking about this a bit harder… I read an interesting article yesterday about how most LLM AI models only use pre-trained (and therefore, old) information unless they are explicitly connected to the internet and brokerage prices – You would hope that this has been worked into the models offered by your provider of choice, but it might be worth reading the small print 🙂