Most investors argue about multiples. Edgar Allen argues about the clock.
The founder of High Ground Investment Management, $10m at launch in 2019, $3bn today, walked me through his process, and much of it reduces to a single sentence he delivered with disarming cheerfulness:
“One thing that we know for sure about all the companies that we invest in is that they’re all going to get wiped out. They’re all trending to zero. It’s just a matter of time.”
That isn’t nihilism; it’s entropy, and it’s the foundation of his investment strategy.
Here’s what falls out of it.
For your capital markets training needs, visit my friends at Finance Talking.
The maths of duration is brutal, and nobody runs it
Try Allen’s example. A business on 15x earnings, throwing off a 5% free cash flow yield. It grows that cash organically at 5% a year, compounded, for 27 years. Then it declines gently over the following decade or so. That’s a forty-year life with 27 years of uninterrupted growth, a wildly optimistic case for almost any company you can name. Your IRR? It’s about 8%.
Fine. But not thrilling.
Now shorten the runway.
If your business has ten or fifteen good years instead of twenty-seven and then hits trouble, the return isn’t merely worse; it’s poor, more or less regardless of what you paid. His conclusion? The entry multiple you spent weeks agonising over matters less than the duration assumption you probably never explicitly considered, let alone factored into a model.
Terminal value is the biggest lie in the model
Conventional analysis builds discounted cash flow, or DCF models. But almost nobody contemplates fully what the terminal value assumes. Will it still be generating cash in 50, 100, or 300 years?
The average lifespan of a company in the real world is about six years.
Allen calls this the most jarring gap between theory and reality in the market. The single largest inefficiency in equity investing. Why? Well, most investors are only holding this for six months, so who cares about decades to come. But the valuation cares. Even over six months, you’d rather own the undervalued asset than the overvalued one.
Own the future that is boring and knowable
If duration is the prize, hunt where duration is structural. Allen looks for long-cycle industries, places where the supply response is slow and the competitive set two decades out is already visible.
Airbus. The A320 launched in 1988 and gets replaced around 2040. Across sixty years, that’s one product cycle. And you already know the 2040 competition field: Airbus, Boeing, Comac, nobody else.
Knorr-Bremse. Train braking systems. Its main competitor in 1890 was Wabtec. Its main competitor today is Wabtec. Siemens tried to bundle its own brakes at a discount. Every train operator asked the same question. How many million kilometres have these done? And passed. Nobody wants to be first on a safety-critical system.
Cities, and the property in them. London has absorbed wars, fires, plagues and floods for a thousand years and hasn’t moved.
Death care. The lowest bankruptcy rate of any US industry. Consumer preferences here change slowly.
Contrast that with most businesses. When you ask who their biggest competitor will be in twenty years, the answer is a company that hasn’t been founded yet. Good luck modelling that.
Moats rot & the brand moat already has
Fifteen years ago, Allen would have called big consumer brands long-duration. He’d have been wrong. Consumers stopped trusting multinational food companies back in the 1980s. The economics held up because challengers needed supermarket shelf space, a national TV campaign, their own logistics, their own factories. Kellogg’s cut Walmart a better deal and your cornflakes died.
Now you advertise on social media, contract out manufacturing, sell direct, and let someone else ship it. The old barriers are gone. Brand value has followed.
Shorts are more of an accounting exercise than a valuation one
Allen isn’t shorting the most expensive stock in the market. He’s looking for businesses that will be worse in three years, where the profit number the market is valuing it on is materially higher than the real one.
The tells include:
A widening gap between net income and genuine free cash flow
Migration from net income → EBITDA → adjusted EBITDA (when a company beats adjusted EBITDA by two cents and it takes you a day to work out whether they beat or missed, that’s the signal).
Falling gross margins propped up by cuts to R&D and marketing.
Factoring, reverse factoring.
Management that changes the goalposts in their own LTIP
These characteristics cluster where businesses are already under pressure. Healthy companies don’t need these adjustments.
What should worry AI investors?
Edgar is suspicious of the assumption that the companies driving a technological change are the ones who capture its value.
The dot-com boom may have misled an entire generation. Yes, created enormous value, eventually. But it destroyed massive value first. That’s not always the case.
The safety bicycle transformed Victorian society and social mobility. Every optimistic forecast about bicycle adoption came true. There were no winners. Just destruction of shareholder capital. Railways, much the same. Airlines offer another case. Show a Pan Am shareholder in 1950 the next seventy years of passenger-kilometre growth, and they’d have remortgaged the house to own airline stocks. The demand forecast was right. The returns were wrong. Airlines revolutionised travel and never sustainably earned a high return on capital.
Warren Buffett famously said that,
If a farsighted capitalist had been present at Kitty Hawk, he would have done his successors a huge favour by shooting Orville down.
For Edgar, the AI stack looks commoditised. Sustainable excess returns? Allen says, “almost certainly not.” LLMs, one’s ahead, then another’s ahead, and switching costs look close to zero.
His summary is the line to keep:
“A lot of AI is about destroying stuff — which is great for society, great for economic growth, great for the consumer. It’s just not great for investors.”
Amazon created real value in retail. It also vaporised shopping malls and retail chains, and most of the surplus went to consumers rather than shareholders.
AI may rhyme. There might yet be an AI 2.0, he conceded, but that shouldn’t be a base-case assumption.
Allen’s 2025 was excellent. None of it came from AI or DRAMs. It came from the same portfolio that did very little in 2024, which is the real takeaway. There was no clever pivot. Just a durable question, asked patiently, about assets most people are valuing as though they’ll live forever. They won’t. It’s just a matter of time. But investors who can gather on the higher ground of long-cycle industries by investing in longer-duration companies should outperform over the long haul.
As some of you know, everything good in the world started from the music of the 1970s. As I was editing the chat with Edgar, the Stevie Wonder classic "Higher Ground" kept spinning through my mind. I considered many of the great recordings of the man himself playing this classic song. However, I discovered this fantastic Music For Change version, which made me stop and listen several times. Please enjoy.
Going to keep on trying
'Til I reach my highest ground
If you enjoyed this and know anyone else who might enjoy it too, please share here. It is the only way this thing grows:
Please consider subscribing; it means so much and takes no time at all:
If you have any comments or thoughts to share, you can now buy me a coffee and leave a message. It would be great to hear from you.
Jeremy
