The Marks Temperature

Think of this as a thermometer for greed. It doesn't predict what the market will do next. It tells you two things: how much you're being paid to take risk right now, and how eager everyone else is to take it.

When the needle is high, people are excited, assets are expensive, and you're being paid very little for the risk you're carrying. When it's low, everyone is frightened, prices are beaten down, and you're being paid a lot.

Most of the time the needle should sit in the middle, and the honest answer is nothing to see here. That's deliberate. Howard Marks ran one of the world's biggest credit funds for decades and made only five big market calls in fifty years. A gauge that screamed every month would be lying to you.

The five bars underneath show which part of the picture is driving the number. Each one is scored 0–100 and they're blended by the weights shown.

What the colours mean. Red is stretched, amber is warm, plain white is unremarkable. Nothing here is a buy or sell signal — they're conditions, not instructions.

HY OAS · 272 bps. The extra interest a risky company has to pay to borrow, compared with the US government. "Basis points" just means hundredths of a percent, so 272 = 2.72% extra. It's the danger money lenders demand. Marks reckons 400–600 is normal, so lenders are currently charging unusually little for taking a real risk. The lowest it has ever been was 241 — in June 2007, weeks before the financial crisis.

Adequacy · 1.17×. Whether that danger money actually covers the damage. Some borrowers go bust each year; historically the cost works out at roughly 233 basis points a year on average. You're being paid 272. Divide one by the other and you get 1.17 — you're earning about 17% more than the losses should cost you. That's a thin cushion. Below 1.0 and you'd be taking the risk for nothing.

CCC OAS · 1023 bps. The same danger money, but only for the weakest borrowers. Above 1,000 is what Marks calls a "distress candidate" — the market is openly doubting it gets repaid. So while the overall market looks calm at 272, the bottom of it is already in trouble. That contradiction is the most interesting thing on this page.

Shiller CAPE · 42.3. How expensive shares are. It's the price divided by average profits over the past ten years — the ten-year average stops one freak year distorting things. The long-run typical reading is about 16. At 42.3, shares cost roughly two and a half times their historical norm. It has only ever been higher once: 44.2, in December 1999.

Forward P/E · 20.0. The same idea using next year's expected profits instead of the past decade's. You're paying $20 for every $1 of annual earnings. That matters because, historically, what you pay at the start has determined most of what you earn over the following ten years — and section 4 shows how strong that link has been.

Margin debt · $1.50 trillion. How much investors have borrowed against shares they already own in order to buy more shares. It's a good read on confidence — people only borrow to invest when they feel sure. The catch is that borrowed money works both ways: it magnifies gains going up, and it forces people to sell at the worst possible moment going down.

When a risky company borrows money, it has to pay a higher interest rate than the US government does. That extra bit is the "spread." It's the danger money lenders demand for the possibility of not being paid back.

When lenders feel relaxed, they accept a small spread. When they're frightened, they demand a big one. So the spread is a fear gauge — and it belongs to the people with the most at stake, the ones handing over the actual cash.

The numbers are in "basis points." 100 basis points = 1% of extra interest. Marks says a normal spread is 400 to 600 — that's 4% to 6% above government rates.

Right now it's 272. The lowest it has ever been was 241, in June 2007 — a few weeks before the financial crisis started. In plain terms: lenders are currently charging close to the least danger money they have ever charged.

This is the question almost everyone skips. People argue endlessly about whether the spread is "low." Marks says that's the wrong question. The right one is: does it cover the losses that are actually coming?

Think of it like insurance. Some borrowers go bust — historically about 3.5 out of every 100 each year on average (the median year is calmer, at 2.7, but recessions drag the average up, and the average is what a lender lives with). When one does, lenders typically recover only about a third of their money, losing the other two thirds. Multiply those together and you get the rough annual cost of things going wrong: about 233 basis points.

Now compare that to the 272 you're being paid. The ratio is about 1.17×, so you're getting only about 17% more than the damage should cost you, a thin cushion. On the calmer median-year assumption (2.7%) it would look like about 1.5×, which is why the slider is there.

Above 2× you're being paid well. Around 1.5× is fair. Below 1× you are effectively lending money for free and taking the risk as a hobby.

Drag the slider to see what happens if bankruptcies come in worse than a calm year. In a recession they can hit 8–10%, and the ratio collapses well below 1× — which is the whole reason thin spreads are dangerous rather than merely unexciting.

Imagine a school where the average grade looks perfectly healthy — but only because the top students got brilliant marks while a chunk of the class is quietly failing. The average hides the problem.

That's what's happening here. The headline number (272) looks calm. But split the borrowers into tiers and the picture changes completely: the safest ones pay almost nothing extra (160), while the weakest pay a fortune (1,023). Anything above 1,000 is what Marks calls a "distress candidate" — the market is openly doubting it gets paid back.

There's a second effect. The make-up of the class has changed. In 1999 only a third of these borrowers sat in the top tier. Today more than half do, so part of the calm in the average comes from the group looking safer on paper.

How much? Take the tier spreads in the table below and weight them by the 1999 shares: 0.327 × 160 + 0.546 × 330 + 0.127 × 1,023 comes to about 362 basis points. On the 2024 shares the same arithmetic gives about 336. So the shift in mix explains roughly 25 basis points of the tightness, and even the 1999 mix leaves the index below Marks's 400 to 600 band. The mix matters, but it is a small part of the story. (The tier weights and spreads are rounded, and they do not reconcile exactly to the 272 index, so read these as rough.)

The price you pay decides most of what you'll earn. Buy a flat at a silly price and your rental yield will be poor no matter how nice the flat is. Shares work the same way.

The J.P. Morgan chart plots the forward P/E at the start of each period on one axis and the annualised return over the following ten years on the other. The dots slope clearly downward: the more investors paid, the less they went on to make, and the highest-multiple starting points are the ones followed by the weakest ten-year returns.

Marks reads the high-multiple end as a range of roughly plus or minus 2% a year, at a multiple of about 22. Today's forward P/E is 20.0, a little lower, so the same logic points to a modest decade rather than a spectacular one. This page does not turn that into a point forecast, because the chart is a relationship with a lot of scatter and not a formula.

Two honest caveats. Each dot needs ten full years of follow-through, so the most recent starting points that can be plotted today are from around mid-2016. And the long stretch behind the chart had interest rates falling, which flattered returns. Treat it as a strong hint, not a prophecy.

See the chart in the J.P. Morgan Guide to the Markets.

When shares are expensive, you are paying for a company to stay on top for decades. So it's only fair to ask how often that actually happens.

Take the twenty biggest companies in America in the year 2000. Twenty-four years later, only six were still in the top twenty. General Electric, Cisco, Intel, Citigroup, IBM, AT&T — untouchable giants at the time, all gone from the list.

And of today's "Magnificent Seven" — the handful of tech names driving the whole market — only Microsoft was even in that 2000 list.

So being the biggest company today tells you surprisingly little about being the biggest in twenty years' time. Roughly a 30% survival rate. Yet today's prices quietly assume near-certain survival. That gap between what's priced in and what history delivers is the entire argument.

Marks published this as a bit of a joke and it turned out to be one of his most useful tools. It's just a list of opposites. For each row, pick whichever side sounds more like the world today.

There's no maths involved, and that's the point. Some of the most reliable warning signs can't be measured, only noticed — are people excited? Are silly deals getting done? Is everyone at the dinner party suddenly an investor?

If most of your ticks land on the left, the market is running hot and, in Marks's words, you should "hold on to your wallet." If most land on the right, everyone is miserable and prices are depressed — which, historically, is exactly when the best bargains have been available.

Rows marked ⚡ auto are pre-filled from current data. The rest are your own judgement, and that's not a weakness — Marks says plainly that he can't quantify these, he just listens for them.

In late 2022 Marks argued the investing world had fundamentally changed — that the era of nearly-free money was over and wouldn't come back. He called it a sea change, and said it was only the third sea change in his 53-year career (CNBC, 13 December 2022).

He laid it out as a simple before-and-after list: how things looked during the easy-money years (2009–2021), versus how they looked once rates rose.

The last column is where things stand today. Read down it and you'll notice something uncomfortable. Mood: optimistic. Buyers: eager. Credit window: wide open. Spreads: near record tight. On the mood and credit rows we have drifted back to the left-hand column.

The rate rows have not gone back. The Fed hiked again on 16 September and the 10-year yield is 5.17%, so the higher-rates half of Marks's thesis has held. The risk-appetite half, the idea that investors would demand more for taking risk, has reversed.