When the Price of Money Starts Rising: Reading This Market Through the 10-Year

No plain without a slope; nothing goes without returning.

—— I Ching, Hexagram Tai, Third Line (pre-Qin era; translated by the author)

What This Episode Is About

In the “What Did We Learn?” segment of The Compound and Friends on 2026-08-24, the host sat down with Nick Colas, co-founder of DataTrek Research, to talk about something everyone was staring at once earnings season wound down, and almost nobody was explaining clearly: why long-dated Treasury yields keep climbing, and how that seeps into equity valuations.

Normally the bond market and the stock market do their own thing, and most people are content to treat them as separate conversations. This summer broke that habit. The 30-year yield pushed out to fifteen- and even twenty-year highs, and the topic moved from the corner of the room to the middle of the table.

The spine of the whole episode is one sentence: the price of money is changing, and it is changing in the opposite direction from the one we spent forty years getting used to.

The Main Points

1. This is not an inflation story. It’s a real-rate story. A nominal yield splits into two pieces: how much inflation investors expect each year, and whatever is left over after that — the real yield. Nick pulled up a chart going back to 2010, and the inflation-expectations line is essentially flat across fifteen, sixteen years, sitting comfortably between 1.5% and 2.5%. The market always knows inflation is coming, and it prices it with remarkable stability. What moves — a ton — is the real rate, which has gone from roughly 2% to pushing 3%. This decomposition matters practically: if you think rising yields mean the market fears inflation, you go buy inflation protection, and you’ve mis-specified half the problem.

2. The hedge has been subtracting from portfolios for six straight years. From 2010 through 2019, holding a long-dated Treasury ETF (TLT) compounded at nearly 8% a year, coupons reinvested — that’s total return. From a credit standpoint the asset is effectively risk-free; what you’re taking on is duration risk. Earning 8% a year on that is why nearly every allocation model of that era carved out a permanent slice for it: the piece that saves you when equities blow up. In the 2020s, that same slice has compounded at negative 4.4%. It still worked in 2020, when everything was falling apart and TLT was the anchor. It has been destructive since. Nick’s framing is honest: a hedge is supposed to lose money when times are good. But plenty of allocators are now looking at that number and asking whether this is the hedge they actually want.

3. Four sources of higher real rates. Three are familiar; the fourth is new. First, the unwind. Real rates were artificially depressed — deliberately, by the Fed, through bond buying — and it worked spectacularly. Mission accomplished; now the balance sheet is stable and there’s no buying, so the compression releases. Second: since 2020 the economy has absorbed a remarkable run of shocks — the aggressive 2022 hiking cycle, a trade-policy shock, two separate oil shocks — with no recession. The inescapable conclusion is that the neutral rate is simply higher than we thought. Third, federal deficits are running above last year’s full-year total already, and credit quality is a live question. The fourth is the one people don’t connect: AI-related long-dated corporate borrowing is pulling demand away from Treasuries. Corporate issuance is running around $1.75 trillion this year, twenty to thirty percent ahead of the same point last year.

4. The step most listeners skip — why does Alphabet selling bonds move Treasury prices? The host asked it directly, and Nick answered with a thought experiment. If I offered you either a Google ten-year bond or a US ten-year Treasury, which one do you think is more secure, notionally? On one side you have a government’s power to tax. On the other you have cash flows that are profound, global, and very well managed. Bond investors are risk-averse by construction: the best outcome available to them is getting their coupons and their principal back — that’s it, there’s no upside beyond that. From that vantage point, Google is about as likely to deliver it as the government is. And every dollar into high-grade corporates is a dollar not going into Treasuries: less demand on the buy side, higher yields. One line from this stretch stuck with me — Google doesn’t have the Marines, the Coast Guard, or a navy, but it has other attributes that in times like these may be just as attractive.

5. Five percent on the ten-year is the trigger — but “how fast” and “how high” are two different questions. The reassuring line on television is that yields being higher doesn’t matter so long as the path is orderly. Nick agrees there’s something to it: when rates move slowly, equity investors have time to assess how the change flows into earnings, company by company. When it happens fast, there’s no assessment, only shock. Then he adds the correction: the move to 5% on tens back in 2023 wasn’t disorderly, and the market twitched anyway. So speed matters — and there is also a level at which the market simply says it isn’t comfortable paying twenty times for the index. Both are true. Don’t only keep the half that’s comforting.

6. The get-out-of-jail-free card from 2022 is gone. There’s an under-discussed coincidence behind why the last violent hiking cycle didn’t kill equities: the largest companies in the index happened to be the ones with the least interest-rate risk. They weren’t big borrowers, and they’d refinanced what debt they had at near zero in 2020 and 2021, so roll risk wasn’t a thing. What they did instead was look at share prices down 30, 40, 50% and get religion on spending — Meta being the poster child, cutting the projects with no return. Prices recovered. You can’t tell that story now. These have gone from buyback companies to companies issuing secondary stock and selling debt, and some genuinely carry roll risk. Nick layers on the second change: it isn’t only leverage. The destination of the cash changed too. Where a third to half of cash flow used to come back to shareholders, now every dollar goes into the science project. You’re carrying reinvestment risk on top of roll risk, simultaneously.

7. The 1990s also averaged 5% — but it was the mirror image of this 5%. People reach for the nineties for comfort: the ten-year averaged around 5%, the economy grew, and it was one of the best decades the stock market ever had. Nick’s response is the sharpest passage in the episode: that 5% came down from 15%. Volcker squeezed the economy at seventeen percent to break inflation, and the release of that pressure — the cost of money falling fast — was itself an enormous tailwind. Our 5% is climbing up from zero. Same number, opposite gradient. And here’s the awkward property of bonds: yields probably don’t go negative the way Europe’s did, but there is no natural cap on the upside. He put it memorably: if you could guarantee a listener the ten-year stops at 5% for the rest of the decade, multiples would expand by two points tomorrow. What the market fears isn’t the number. It’s not knowing where the number stops.

Going Further

”I bought it as a hedge. Why is it the thing losing me money?”

This is the question a lot of people have been carrying around without quite saying out loud. You followed the textbook, you put a slice in bonds, the reason was “it goes up when equities fall,” and six years later it has compounded at negative 4.4% while equities never fell.

Set the frustration aside and look at the structure. A hedge can lose money in two ways, and they are not the same thing. One is losing money in good times — that’s normal, that’s the premium you pay for the insurance. The other is that the pricing premise underneath it changed, meaning the mechanism that made it a hedge has loosened. The first you tolerate. The second you re-underwrite.

The way to tell them apart is exactly the decomposition this episode opens with. Nominal equals inflation expectations plus real. If bonds are falling because inflation expectations jumped, that’s a cyclical phenomenon — inflation cools, the asset comes back, your insurance policy is intact. But the inflation line has barely moved for fifteen years. All the movement is in real rates, and real rates reflect the long-run price of capital in this economy: the neutral rate, the fiscal position, and who else is competing for the same money. None of that reverses because next quarter’s CPI prints well.

So the real question isn’t “will bonds come back.” It’s “is the reason I bought this still true?” That applies to anything you hold for protection — gold, cash, defensive sectors, an options overlay. Write down the one sentence that justified the position when you took it. Ask whether that sentence still parses today. If it doesn’t, the loss is just the result; the invalidated reason is the thing that needs handling.

The practical implication in the episode is stated plainly: this isn’t the moment to extend duration. Keep bond exposure short — Nick’s line to clients is under five years — and revisit extending when economic data genuinely weakens, because weakening data is precisely what erodes the “neutral rates have to be higher” argument and lets real rates stabilize at the long end. Note the ordering, it’s worth keeping: you don’t start by asking whether rates are high. You start by asking whether the reason rates are high has begun to loosen.

”Rates keep rising and stocks keep rising. What am I missing?”

The contradiction is uncomfortable enough that many people just pick a side and file the other as noise. This episode offers a clean formula that holds both at once.

The discounting math taught in the first week of business school is cash flow divided by the discount rate minus the growth rate. Most people retain the discount rate: rates up, denominator up, valuation down. Correct, as far as it goes. But the growth term lives in that same denominator, and it’s being subtracted — so a jump in growth can shrink the denominator harder than a rate rise expands it. Nick is blunt about it: if you can show me earnings compounding at, say, 18% for five years, I honestly don’t care whether the ten-year is at 5% or 6%, because moving from low-teens growth to high-teens growth more than offsets the higher risk-free rate.

That isn’t a rationalization for high valuations. It’s a statement about which variable to watch. When “rates up” and “growth up” run at the same time, the outcome depends on which runs faster. And that’s checkable — you don’t have to guess.

The episode hands you the check. The S&P is up roughly 12% year to date, while earnings revisions for this year and next are up 15% and 13%. Which means the entire move this year came from earnings revisions; the multiple didn’t expand at all, it contracted slightly. Nick, who was a sell-side analyst in the nineties and has watched this data for thirty years, points out that analysts essentially never raise numbers during a year — they start high and trim, start high and trim. This year inverted that. Very unusual.

You can run this decomposition yourself: split any period’s return into the earnings piece and the multiple piece. All earnings means the price the market pays didn’t get richer — you’re earning what the companies actually earned. All multiple means nothing new happened, people just decided to pay more, and that kind of advance has far less resistance to rising rates. Same gains, different sources, very different fragility. No model required, just two numbers.

Related: the upside scenarios in this episode are worth studying for their structure more than their conclusions. Base case — earnings keep growing, multiple holds near twenty — gives 6% to 16% over the next twelve months. Better case adds a resolution to the geopolitical conflict and lower oil, pushing the multiple to twenty-two. Best case additionally requires AI capex to start showing up in tech earnings, taking the multiple to twenty-four. The point is that the third case needs three things to go right at once. When a conclusion depends on three independent conditions, its probability is those three multiplied, not those three reasons added together. A lot of people treat “I have three bullish reasons” as tripled confidence. The arithmetic runs the other way.

”Everyone says there’s no recession coming. Should I just go all in?”

There’s a passage here I’d copy down verbatim, because it drags an invisible assumption out from under the floorboards.

Sector and stock correlations within the S&P are at ten-year-plus lows, and aggregate volatility is low too. On the surface that reads as good news: low correlation means you get to pick, you can be long healthcare and financials and tech simultaneously and have them do their own thing. Nick points out that the phenomenon rests on a single shared premise: nobody believes a recession is coming. If one were actually in the wings, investors wouldn’t be picking — they’d all head for the door together, and correlations would go to one overnight.

Then the harder observation: the companies levering up their balance sheets to build AI infrastructure are making that exact same bet with their behavior. If you were worried about a recession, you would not plow every dollar of cash flow into something that takes years to pay off. So the implicit assumption of this market and this investment cycle is no recession for five years. Full stop. If you want to be maximally long, you have to believe that one hundred percent.

The useful part isn’t the conclusion, it’s the move: translate “what everyone is doing” into “what everyone is assuming.” Low correlation, low volatility, enormous capex, equity issuance — four apparently unrelated facts sharing one foundation. Move the foundation and all four move together, which is precisely why four positions diversified underneath a single assumption aren’t diversified at all.

The episode offers one concrete handle. Around 40% of the global index by weight sits in tech and big tech; the equal-weight S&P is 14% tech. Nick and his co-founder have been long-time skeptics of equal weight, and he now says he sees a point to it — you still own these companies, just with less leverage to that one bet. That’s not a recommendation to buy anything. It’s a reminder that what you think is diversification may just be several phrasings of the same assumption. The check is mechanical: list your holdings, and next to each one write the precondition it needs in order to do well. If that column reads nearly identically all the way down, your diversification is cosmetic.

One more thing worth keeping: the episode notes that AI demand exploded this year for essentially one reason — coding. Nothing wrong with that, but the non-coding side is the vast majority of the economy, and it hasn’t started working yet. Meanwhile the money going in isn’t being spent on what AI does today; it’s being spent on the race to artificial general intelligence — on getting there first and compounding an advantage nobody can close. That also explains something counterintuitive: China’s open-source models are very good and very fast, and investor sentiment barely reacted, because within the framing of that race, the finish line isn’t there.

Worth a Look

• The Compound and Friends, “What Did We Learn?” segment, 2026-08-24 (the source for this piece)
• The DataTrek Research daily briefing, run by Nick Colas and Jessica Rae — where most of the charts in this episode come from
• artificialanalysis.ai: public tracking of model speed and capability over time, used in this episode to show AI progress running well ahead of the Moore’s Law cadence
• Ben Carlson’s work on the 1990s averaging 5% on the ten-year during a golden decade — read it alongside this episode’s “where you’re coming from” rebuttal

The One Thing to Take With You

One idea only: your situation is defined by where you came from, not just where you are.

Five percent in the 1990s and five percent in 2026 are the same number and two opposite experiences — one is the tailwind of unwinding from 15%, the other is the climb up from zero. The same body weight, the same salary, the same quality of a relationship, sitting on different trajectories, feel different and lead to different places. Human brains read levels well and gradients badly, so we pass judgment on an absolute number and then feel confused that someone else lives entirely differently at that same number.

Do this tonight. Pick one thing you’ve been quietly complaining about lately — your salary, your weight, hours of sleep, how many times a week you actually talk to a particular person. Just one. Take a piece of paper and write down exactly two numbers: what it was three years ago, and what it is today. No commentary, no explanation. Two numbers.

Then ask yourself one question: is what bothers me the level, or the direction?

If it’s the direction, what you need to work on is the slope rather than the number — and slopes are usually movable. If it’s the level — the number has barely budged in three years and you’ve been uncomfortable the whole time — then you’re facing something else entirely: you’ve already endured it for three years, and it isn’t going to fix itself.

Those two answers lead to completely different actions. And until you write down those two numbers, you genuinely can’t tell which one you’re in.

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.