Today's 30-year Treasury yield and the 2000 dot-com top: US Treasury yields and stock valuations since 1962

How this study happened

On the morning of 5 October a popular post comparing today's bond market with the 2000 top caught my eye, so I set out to check it. By early evening the first version of this page and a thread on X were out. An hour later, going back over the method, I saw the mistake: I had treated yields falling to 6% and yields rising to 6% as the same thing. I took the posts down and rebuilt the study around direction and valuation.

A post by @Dan_Kostecki on the forward P/E then took me one step further, hence the section on earnings forecasts.

Wojciech, 5 October 2026

The findings at a glance

4 of 5

past returns of the 30-year to 6% came with yields falling; today they're rising

68% to 85%

How often the S&P 500 was higher a year after a first new high in the 10-year yield, depending on how the high is defined; after any day: 75%

CAPE 41

higher in only 16 months since 1871, all from April 1999 to September 2000

−1.4 pp

S&P 500 earnings yield minus the 10-year yield; −2.7 pp in 1999 and 2000

Claims and verdicts

Six claims and what the data says

Every claim from every study, in one list →

Claim tested	Finding	Verdict
The 30-year Treasury yield at 6% marked the 2000 stock market top	The S&P 500 peaked on 24 March 2000, the day the 30-year closed at 6.00%	Correct, to the day
Today's bond market looks like 2000	In 2000 the 30-year was falling to 6% (from 6.75% in January); today it is rising (from 4.73% a year ago). Only June 1999 looked like today	Wrong direction
A new high in the 10-year Treasury yield hurts stocks over the next year	After a first new high in the 10-year yield, the S&P 500 was higher a year later 68% to 85% of the time, depending on which of the four definitions in the study's Table 2 is used. After any day: 75%. None of these differences is bigger than chance alone would produce. Check it yourself →	No consistent effect
…over the next 3 months	Higher in 46% to 62% of cases against 66%; weaker after 1 and 3-year highs, not after the others	Mixed
Stocks are as expensive today as in 2000	CAPE 41 today, 42 in June 1999, 43 in March 2000; higher in only 16 months since 1871. Forward P/E (FactSet 19.0) sits near its last-10-year average	Close, on CAPE
A low forward P/E means stocks are cheap	The forward P/E fell from 20.4 to 19.0 while the S&P 500 rose 2.0%: the fall came from higher earnings forecasts (+9.3%), which assume +32% growth in 2026. A year like that from record earnings happened since 1962 only in 1994 and 2021	No: higher forecasts

1 · The 2000 comparison

In 2000 yields fell to 6%; today they're climbing to it.

The S&P 500 peaked on 24 March 2000, and the 30-year closed at 6.00% that day, so the date is right. On 30 September 2026 it closed at 5.64%, the highest since 2002.

Since 1993 the 30-year has come back to 6% five times. In four of them, March 2000 included, it was falling at the time. Only once, in June 1999, was it rising, like now.

The 30-year yield since 1990: five returns to 6% between 1993 and 2000, four with yields falling, one in June 1999 with yields rising; today 5.64% and rising. — WB Trading Labs · Fig. 1Weekly close. Triangle down: the yield came to 6% falling; up: rising.

The five returns to 6%, and today

Moment	30-year a year before	Yield arrived	CAPE	Earnings yield minus 10-year	S&P 500 a year later
Sep 1993	7.33%	▼ falling	21	−0.9 pp	+1.6%
Jan 1996	7.85%	▼ falling	25	−0.1 pp	+19.3%
Mar 1998	6.83%	▼ falling	36	−2.0 pp	+17.0%
Jun 1999	5.79%	▲ rising	42	−2.7 pp	+10.7%
Mar 2000	5.54%	▼ falling	43	−2.7 pp	−24.5%
Today	4.73%	▲ rising	41	−1.4 pp	?

WB Trading Labs · Table 1"Rising": the yield above its level 3 and 12 months earlier and at a 12-month high. CAPE: S&P 500 price over 10 years of inflation-adjusted earnings (Shiller). Earnings yield: the inverse of the P/E. S&P 500 as a price index, no dividends.

2 · New highs in yields

After a new high in the 10-year Treasury yield, the S&P 500's next year looked like any other.

So I looked at what stocks did after the 10-year yield hit a new high. Since 1962 that's happened 42 times, counting the first new 52-week high after at least three months without one. A year later the S&P 500 was up 73% of the time, against 75% after any day.

A different definition of a high doesn't change much. 3 and 5-year highs, and highs after a rise of at least 1 percentage point in a year, give 68% to 85%. September's high was both a 3 and a 5-year one, and the yield is 1.14 points above a year earlier. None of these is further from the usual 75% than chance would explain.

The S&P 500 after a first new high in the 10-year yield: 1, 3 and 5-year highs, and 1-year highs after a rise of at least 1 percentage point

First new high in the 10-year yield	New highs with a full year after them	Higher after 3 months	Higher after 12 months
1-year high	41	50%	73%
3-year high	22	46%	68%
5-year high	17	59%	82%
1-year high, up 1 pp in a year	13	62%	85%
every day at a 1-year high	804 days	·	63%
any day since 1962	·	66%	75%

WB Trading Labs · Table 2I count only the first new high after at least 3 months without one; grouping by 1 or 6 months gives the same conclusions. The "every day" row counts all days at a high, so long climbs weigh more in it.

Check it yourselfChoose the definition and the horizon, and see every case since 1962 →

Two things look worse. In the first three months after 1 and 3-year highs, the S&P 500 was up only about half the time (50% and 46%, against 66% normally), though not after 5-year highs or fast rises. And if you count every day at a high, not just the first, the share a year later drops to 63%. Almost half of those days fall in 1966-1981.

S&P 500 returns after 42 first new highs in the 10-year yield: after 3 months a median of −0.1% against +2.6% for any day, after 12 months +10.9% against +10.5%. — WB Trading Labs · Fig. 2Each dot is one new high (circle: higher, triangle: lower), the white bar is their median. Grey band: the middle half of all windows since 1962, dashed line: their median.

3 · Valuation

On CAPE (the S&P 500's price over 10 years of inflation-adjusted earnings), today does look like 2000.

CAPE, the S&P 500's price over ten years of inflation-adjusted earnings, is 41.4 today. Since 1871 it's been higher in only 16 months, all between April 1999 and September 2000. In June 1999, the one past case of rising yields at 6%, it was 42; in March 2000, 43.

Against bonds, today is further from 2000: the S&P 500's earnings yield, 3.8%, is 1.4 percentage points below the 10-year yield, against 2.7 in 1999 and 2000.

June 1999 is the only case that matches on both counts, rising yields and CAPE above 40. A year later the S&P 500 was up 10.7%, though the top had already come in March 2000, nine months in. It's one case, so it proves nothing - I show it because it's the only one there is.

FactSet's forward P/E tells a calmer story: 19.0, below its 5-year (19.8) and 10-year (19.1) averages. Both numbers can be right. The forward P/E compares the price with analysts' forecasts and with the last ten years, which ran high by long-run standards; CAPE compares it with ten years of actual earnings and with history back to 1871. A high CAPE partly reflects fast earnings growth lately, and analysts' forecasts tend to run optimistic. I don't have forward P/E history for 2000, so I haven't checked that comparison.

CAPE since 1962: a peak of 44 in December 1999, June 1999 at 42.2, today 41.4; triangles at the five 30-year returns to 6%. — WB Trading Labs · Fig. 3Monthly. Triangles: the five 30-year returns to 6% (down: yields falling, up: rising) and today. Dashed line: today's level. Data: Robert Shiller via multpl.com; the latest 4 months use estimated earnings.

Did high valuations change what followed a new high in yields? Not consistently, and the samples are tiny. Valuation on its own looks like the bigger factor: on days with CAPE at 35 or more, the S&P 500 was up a year later 58% of the time, against 75% overall. But that rests on three independent periods (1998-2001, 2021-2022 and 2024-2025), which is too little to build on.

S&P 500 higher a year after a new high in the 10-year yield, by valuation

First new high in the 10-year yield	Stocks expensive (CAPE 25+)	Stocks cheaper (CAPE below 25)
1-year high	79% (n 14)	70% (n 27)
3-year high	57% (n 7)	73% (n 15)
5-year high	83% (n 6)	82% (n 11)
1-year high, up 1 pp in a year	100% (n 2)	82% (n 11)

WB Trading Labs · Table 3CAPE in the month of the new high. The 25 threshold is roughly the median since 1962. With this few cases, no gap is decisive.

Is the low forward P/E a product of high forecasts?

Mostly, yes. From 30 June to 30 September the S&P 500 rose 2.0% while FactSet's forward P/E fell from 20.4 to 19.0. The price didn't fall - the earnings forecast went up, by 9.3% in one quarter. About two-thirds of that is analysts raising their numbers: FactSet's 2026 estimate rose 6.2%. The rest is the calendar: the "next 12 months" window moved a quarter further into 2027, when analysts expect earnings to grow about 16%.

Why the forward P/E fell in Q3 2026

	Period or level	Change
S&P 500	30 Jun → 30 Sep	+2.0%
Forward P/E (FactSet)	20.4 → 19.0	−6.9%
Forecast earnings, next 12 months	per FactSet	+9.3%
of which: higher 2026 estimates	$340.49 → $361.66	+6.2%
of which: the window moving a quarter	into 2027, at +15.8% growth	+3.7%

WB Trading Labs · Table 4Forward P/E and growth estimates: FactSet Earnings Insight, 2 October 2026. Changes in the 12-month and 2026 estimates are FactSet's own. The window effect is approximate: it assumes the 12-month estimate blends 2026 and 2027 by time. The parts don't add up exactly, because 2027 estimates moved too.

Those forecasts assume a rare year. FactSet's analysts expect S&P 500 earnings to grow 32.4% in 2026. Since 1962, growth of 30% or more over 12 months, starting from earnings near a record, happened in 8 of 383 such months, in 1994 and 2021. The other years like that started from depressed earnings, after a slump.

It isn't all hope, though. Actual earnings grew 28.2% in the 12 months to June 2026, so results back much of the forecast. FactSet does point out that the biggest tech companies' growth was helped by investment gains, which are lower-quality earnings.

I've seen a similar pattern before, in my study of analyst ratings: there, price targets followed the share price; here, earnings forecasts follow results up. That study tested price targets and ratings, not earnings forecasts, so it's a resemblance, not proof.

4 · Credit

Baa-rated companies pay little more than the US government to borrow, as in 2007.

On 24 March 2000 a Baa-rated company paid 2.13 percentage points more than the US Treasury to borrow for 10 years, right at the median since 1986. On 1 October 2026 it paid 1.49, less than on 96% of days since 1986. The same spread was 1.60 in June 2007, at the 10-year's last new high before the financial crisis, so a low reading doesn't rule anything out.

The Baa spread over the 10-year since 1986: 2.13 in March 2000, 1.60 in June 2007, 1.49 in October 2026; median 2.13. — WB Trading Labs · Fig. 4Weekly close. Dashed line: the median since 1986 (2.13). Data: Moody's via FRED.

Every new high

The 42 first new highs in the 10-year yield

Show the table

First new high	10-year yield	S&P 500, 3 months	S&P 500, 12 months
Jul 1963	4.05%	+4.5%	+19.2%
Aug 1965	4.24%	+7.3%	−2.4%
Jul 1966	5.03%	−16.3%	+5.8%
Sep 1967	5.35%	−1.6%	+7.7%
Mar 1968	5.95%	+13.4%	+14.7%
Dec 1968	6.07%	−7.0%	−15.7%
May 1970	8.15%	+12.8%	+43.8%
Aug 1972	6.30%	+5.7%	−5.8%
Feb 1973	6.64%	−2.9%	−16.6%
Apr 1974	7.58%	−5.9%	−4.6%
Apr 1975	8.22%	+15.1%	+23.8%
Aug 1975	8.37%	+2.2%	+18.4%
Oct 1977	7.54%	−5.5%	+10.0%
Oct 1978	8.72%	+6.1%	+4.1%
Sep 1979	9.44%	−2.0%	+16.8%
Jan 1980	11.10%	−7.4%	+14.3%
Feb 1981	13.65%	+4.1%	−9.9%
Mar 1984	12.22%	+0.3%	+16.9%
Apr 1987	8.39%	+7.3%	−10.6%
Aug 1987	8.92%	−26.3%	−21.8%
Mar 1989	9.49%	+9.8%	+16.8%
Feb 1994	6.22%	−1.6%	+5.1%
Sep 1994	7.52%	−3.4%	+23.8%
May 1996	6.85%	+1.0%	+24.1%
May 1999	5.66%	−0.7%	+8.4%
Jul 2003	4.49%	+5.8%	+11.7%
May 2004	4.61%	−3.6%	+4.6%
Nov 2005	4.65%	+3.7%	+11.8%
Mar 2006	4.68%	+0.1%	+8.4%
Jun 2007	5.26%	−1.4%	−10.6%
Apr 2010	4.01%	−13.9%	+12.2%
May 2013	2.15%	−0.2%	+15.0%
Dec 2013	2.99%	+0.9%	+13.6%
Nov 2016	2.34%	+8.3%	+18.3%
Jan 2018	2.62%	−3.8%	−4.5%
Oct 2018	3.15%	−13.4%	+0.9%
Feb 2021	1.37%	+8.1%	+10.9%
Jan 2022	1.76%	−4.0%	−16.2%
Sep 2022	3.49%	−1.2%	+12.9%
Aug 2023	4.28%	+2.1%	+26.1%
Jan 2025	4.77%	−8.0%	+19.5%
May 2026	4.59%	+4.5%	not yet

Method and caveats

What this study measures, and what it doesn't

Correction: the first version of this page (the morning of 5 October 2026) counted every return of the 30-year to 6% without asking which way yields were moving, and read "once in five" as reassurance. That was the wrong comparison: four of the five cases do not resemble today. This version replaces it.
Correction (10 October 2026): this page said most of the third-quarter rise in forecast earnings was the window moving and about 2 points was forecasts being raised. It was the other way round: by FactSet's own figures the 2026 estimate rose 6.2% and the window adds about 3.7%. I had rolled the window at 2026 growth instead of 2027.
Data: daily closes. 10 and 30-year US Treasury yields and the Baa spread from FRED (Federal Reserve Bank of St. Louis); the S&P 500 from Yahoo Finance; CAPE, P/E and earnings yield monthly from Robert Shiller's data via multpl.com (within 1-3% of Shiller's own file; the latest 4 months use estimated earnings).
No dividends: the price index, after the new highs and after any day alike, so the comparison is like for like, but the levels sit below total return by the dividend.
Definitions: I count only the first new high after at least 3 months without one. Direction ("rising"): the yield above its level 3 and 12 months earlier and at a 12-month high. The four definitions of a high and the direction test were added after reviewing the first version, before their results were computed.
"Chance": the share of random draws of the same number of days with an equally weak or weaker result (5,000 to 20,000 draws). The windows overlap, so it is a rough gauge, not a formal test.
Few cases: the five returns to 6%, 13 to 42 new highs and three periods of high valuation are few. The results describe the past and say nothing about what happens this time.
Not tested: markets other than the US, why yields were rising (inflation, deficits, Fed policy), or individual stocks.

I post new studies on X first.

Follow @WojciechBernPL

The thread on X All studies

The 30-year Treasury yield is near 6%, as in 2000 - but heading the other way