80 years of shmita, by the numbers

The Liquidity Map · Macro report

80 years of shmita, by the numbers

In June, I wrote about the 7-year cycle with only four examples.

Now I've laid out 12 complete cycles, from 1938 to 2022.

The result is clearer than I expected: the year before shmita has never ended lower.

October 4th, 2026 · 15 min read

Key takeaways

In the biblical cycle, shmita is the seventh year: the land rests, and at the end of the year debts are forgiven.
Across 12 complete S&P? cycles, from 1938 to 2022, the year before shmita was positive every single time, averaging +15.3%.
The shmita year averaged just +1.3%, and the year after it -1.4%. That is where the big declines clustered.
In 11 of 12 cycles, the peak came in the shmita year itself. The typical decline from there was 19%, and the two largest reached 50% and 58%.
Bitcoin repeated the pattern in both of its cycles so far: +245% and +382% in the years before shmita, then -46% and -64% in the shmita years.
On the calendar: 2027 is year 5 of the shmita cycle, and 2028 is year 6, the year before shmita. The next shmita year starts on September 21st, 2028.
The second big clock is 2035: that's where I see the secular cycle? ending, right before the 2035-36 shmita year.

In June, I wrote about why I watch a 7-year cycle, not just a 4-year one.

I ended that piece with an admission: four cycles don't make a law.

So I took everything I could get.

Daily S&P data going back to 1928, Hebrew calendar years calculated exactly, and 12 complete cycles, from 1938 to 2022.

Plus Bitcoin, for as much history as it has.

It's the most detailed piece I've written on shmita. Send it to anyone who asks why I keep talking about 2028.

Definition

shmita (the sabbatical year)

In Hebrew, shmita means “release.” It is the seventh year of a seven-year cycle in the Jewish calendar. During this year the land is not worked, and at its end debts are forgiven. The year begins at Rosh Hashanah, the Jewish New Year, in September or October, and runs until the next Rosh Hashanah.

The origin

A year for the land and a year for debts

The rule comes from the Torah.

The Book of Leviticus says that every seventh year the land must be left to rest: no sowing, and no harvesting for sale.

Deuteronomy adds the financial side: at the end of the year, debts between people are forgiven.

The day debts are wiped away is the last day of the shmita year, the 29th of Elul.

The rule was taken so seriously that it created a practical problem.

Toward the end of the cycle, people stopped lending money, afraid the debt would be forgiven before it was repaid.

Hillel, one of the great sages of Judaism, found the fix about two thousand years ago: the prozbul, a document that hands the loan over to the court so it is no longer forgiven.

After seven cycles of seven years comes the jubilee, the 50th year, when land went back to the families it had been bought from.

In today's Israel, the rule still applies to farming. There is even a legal workaround, heter mechira, under which the land is temporarily sold to a non-Jew for the duration of the year.

The link to the stock market only became popular in 2014, through a book by Jonathan Cahn.

He noticed that two of Wall Street's worst days fell exactly on the 29th of Elul: September 17th, 2001, with the S&P down 4.9%, and September 29th, 2008, with the S&P down 8.8%.

For 2015, though, he predicted a crash that never came.

I don't use shmita as a prophecy, or as a date to circle on the calendar.

I use it as a clock, and I check that clock with numbers.

29 ElulThe last day of the shmita year, the day debts are forgiven. In 2001 and 2008 it fell on a day of panic in the markets.

To see whether this clock works, I needed far more than four examples.

The method

How I ran the numbers

I counted every year the way the Jewish calendar counts it, from Rosh Hashanah to Rosh Hashanah.

The 2021-22 shmita year, for example, ran from September 7th, 2021, to September 25th, 2022.

For each year, I calculated the S&P's return and the largest decline within it.

12 complete cycles, 84 years of market history, with no cherry-picking of the years that suit me.

I started in 1938, with the first complete cycle after the 1937-38 shmita, and stopped in 2022, at the last completed shmita year.

The index is the S&P 500 without dividends?. Before 1957, the data come from its predecessor, a 90-company index.

Daily highs and lows only exist from 1962 onward. From then on, I measure declines with them; before that, with closing prices.

And an honest caveat: 12 cycles are far more than four, but still a small sample.

Read the numbers as a pattern, not as a law.

84 yearsThe S&P period analyzed: 12 complete 7-year cycles, from September 1938 to September 2022.

Here's what the cycle looks like, year by year.

The cycle

The 7 years, one at a time

I lined up the 12 cycles side by side and averaged each year of the cycle.

Years 5 and 6 come out best: +14.1% and +15.3% on average.

Year 7, the shmita year, drops to +1.3%. Year 1, the first year after a shmita, to -1.9%.

That covers 1938-39 through 2015-16. Counted after each shmita from 1944 to 2022, the year after averages -1.4%, as the next table shows.

The year before shmita was positive in all 12 cycles. Not a single exception in 84 years.

For a sense of how rare that is: on average, the S&P rose in 69% of all years.

Year 6 rose 100% of the time and had the smallest average decline within the year, -9.9%.

At the other extreme, years 7 and 1 had average declines of -16.5% and -23.6%.

Cycle year	Average return	Positive years	Average drop within year
Year 1	-1.9%	4 of 12	-23.6%
Year 2	+13.6%	10 of 12	-13.6%
Year 3	+11.5%	8 of 12	-14.1%
Year 4	+5.5%	7 of 12	-15.9%
Year 5	+14.1%	11 of 12	-12.2%
Year 6 (pre-shmita)	+15.3%	12 of 12	-9.9%
Year 7 (shmita)	+1.3%	6 of 12	-16.5%
All years	+8.5%	58 of 84	-15.1%

S&P excluding dividends, 12 complete cycles (1938-2022), years running from Rosh Hashanah to Rosh Hashanah. The average drop within the year is the average of each year's largest peak-to-trough decline.

12 of 12How many times the year before shmita was positive for the S&P, from 1938 to 2022.

The S&P's average return in each year of the 7-year cycle, across 12 complete cycles.

Averages hide the story behind each cycle, though. Let's go through them one by one.

The years before

The year before shmita, the shmita year, and the year after

In the table below, each row is one cycle.

The first column of figures is the year before shmita, the second is the shmita year, and the third is the year after.

The weakest year before shmita returned +3.7%, in 1943-44. The best, +36.6%, in 2020-21.

The last four shmita years, from 2000 onward, were all negative for the S&P.

Up to 1960 the cycle was gentler: the shmita years of that era even finished higher.

Since 1965, 6 of 9 shmita years have closed lower.

Even the exceptions had something hidden in them.

In 1986-87, the shmita year returned +37.4%, but the crash of October 19th, 1987, came just 26 days after the year ended.

Shmita window	Year before	Shmita year	Year after
1944-45	+3.7%	+26.3%	-4.0%
1951-52	+25.0%	+5.6%	-3.7%
1958-59	+12.9%	+17.9%	-4.6%
1965-66	+8.8%	-12.1%	+21.9%
1972-73	+10.2%	-1.2%	-39.1%
1979-80	+7.7%	+13.0%	-7.4%
1986-87	+27.8%	+37.4%	-16.9%
1993-94	+11.4%	+2.0%	+23.5%
2000-01	+6.3%	-27.7%	-13.9%
2007-08	+11.9%	-24.8%	-3.4%
2014-15	+20.9%	-1.9%	+10.6%
2021-22	+36.6%	-18.6%	+20.5%

S&P return in each Hebrew calendar year. Each row is named after the shmita year; the year before is the one preceding it (for example, 1943-44 for the 1944-45 row). The highlighted rows are the last four cycles.

+3.7%The weakest of the 12 years before shmita, in 1943-44. Even the worst of them finished higher.

The S&P 500 from 1938 to today, one cycle at a time. Gold is the year before shmita, red is the shmita year, and the numbers are each year's return.

The annual return is only half the story. Where the peak falls matters too.

The peak and the low

The peak comes in the shmita year, the low in the year after

I took the two-year stretch made up of the year before shmita and the shmita year, and looked for the highest point.

In 11 of 12 cycles, the peak came in the shmita year itself. The only exception is 2000, when the peak came in March, in the year before.

From the peak, I looked for the lowest point through the end of the year after.

The typical decline was 19%. The two largest reached 50% and 58%.

The low fell in the year after shmita in 11 of 12 cycles.

In 1974 and 2002, the market's actual low came a few weeks and a few months, respectively, after the period I measured had ended.

I also checked it the other way around, starting from the bear markets?.

Since 1956, the S&P has had 11 declines of more than 20%. Six of them began in the year before shmita or in the shmita year.

If it were just chance, I would have expected three.

The 1930s tell the same story, even though they're not in the average: in the 1930-31 shmita year the S&P lost 37.1%, and in 1937-38, 26.5%.

Window	Peak	Low	Decline
1944-45	Sep 1945	Sep 1946	-9.0%
1951-52	Aug 1952	Aug 1953	-8.7%
1958-59	Aug 1959	Mar 1960	-11.9%
1965-66	Feb 1966	Oct 1966	-23.7%
1972-73	Jan 1973	Sep 1974	-47.3%
1979-80	Aug 1980	Sep 1981	-13.8%
1986-87	Aug 1987	Oct 1987	-35.9%
1993-94	Jan 1994	Apr 1994	-9.7%
2000-01	Mar 2000	Jul 2002	-50.0%
2007-08	Oct 2007	Mar 2009	-57.7%
2014-15	May 2015	Feb 2016	-15.2%
2021-22	Jan 2022	Oct 2022	-27.5%

The peak is taken from the year before shmita and the shmita year; the low, from the peak through the end of the year after. Daily highs and lows from 1962, closes before that.

-19%The S&P's median decline from peak to low, across 12 cycles. The average is -26%, pulled down by 1973-74, 2000-02 and 2007-09.

The S&P 500 since 1938, on a log scale. Gold marks the year before shmita, red the shmita year. Labels show the five largest peak-to-trough declines.

The level to watch isn't just the peak. It's also the low that follows it.

The low rule

After the window, the low held

In August 2023, I wrote on Twitter: “Last two SHMITA years: '15 & '22. After SHMITA, no new lows.”

The idea was simple: once the shmita year is over, the market doesn't go below its low.

I've now tested it on all 12 cycles, with one adjustment: I also counted the year after, because that's usually where the lows come.

In 9 of 12 cycles, the low of the shmita year and the year after was not broken again until the next cycle.

The three exceptions tell us something.

In 1974 and 2002, the market dipped slightly below the window's low shortly after it closed. The decline was still underway.

1970 was different: a new bear market, four years after the 1966 window.

If the rule holds this time too, the 2029-2030 low will be the base for the final leg of the secular cycle.

9 of 12Cycles in which the low of the shmita window, including the year after, was not broken again until the next cycle.

In stocks, the cycle has 84 years of history. In Bitcoin it has just two chapters, but both are clear.

Bitcoin

Two cycles, the same rhythm

Bitcoin has been through two complete shmita windows: 2014-15 and 2021-22.

In both, the year before shmita was an explosion: +245% in 2013-14 and +382% in 2020-21.

In both, the shmita year was a collapse: -46% and -64%.

The cycle low came inside the window or right after it both times: in January 2015, at $152, and in November 2022, at $15,479.

The 2022 low came eight weeks after the shmita year ended, in the year after, just like in stocks.

Two cycles prove nothing on their own. But they fit what 84 years of S&P data show.

The details on the 2022 low and on the halving? are in my earlier piece on shmita, so I won't repeat them here.

+382% / -64%Bitcoin in the year before shmita, 2020-21, and in the shmita year, 2021-22.

Bitcoin since 2013, on a log scale, with the years before shmita (gold) and the shmita years (red). On the right, the years 2027-2029.

With this clock in hand, the question becomes: where are we now?

2028

Where we are now and what comes next, through 2030

The current cycle began in September 2022, after the last shmita.

The first four years ran above average: the S&P rose in every one of them, by 14% to 28% a year.

Year 5 began on September 12th. Historically, it's the second-best year of the cycle, positive in 11 of 12 cases.

One detail matters here.

The years of the cycle don't start in January.

They start in the fall, on Rosh Hashanah.

On the regular calendar, year 5 covers almost all of 2027, and year 6, the year before shmita, almost all of 2028.

In short: 2027 is year 5 of the shmita cycle, 2028 is year 6, and the shmita year starts in September 2028 and covers most of 2029.

The year before shmita starts on October 2nd, 2027. The shmita year starts on September 21st, 2028, and runs until September 9th, 2029.

If this cycle matched the historical average for years 5 and 6 exactly, the S&P would reach about 10,100 points by September 2028.

The level I'm watching is higher, around 11,000. That would be roughly +74% from the March 2026 low and +43% from today.

I explained why 2027 could be explosive when I wrote that 2027 is THE year.

Then comes the window.

Historically, the decline from the peak was 19% at the median? and 26% on average. I expect something steeper, more than 30%.

I'd look for the low in the year after, between fall 2029 and early 2030.

But 2029 won't be 1929: a year of decline, then a return to growth, not a lost decade. I explained why in a video on September 21st.

Cycle year	Exact dates	On the calendar	S&P history
Year 4	Sep 23, 2025 - Sep 11, 2026	2026	+5.5% · 7 of 12
Year 5	Sep 12, 2026 - Oct 1, 2027	2027	+14.1% · 11 of 12
Year 6 (pre-shmita)	Oct 2, 2027 - Sep 20, 2028	2028	+15.3% · 12 of 12
Year 7 (shmita)	Sep 21, 2028 - Sep 9, 2029	2029	+1.3% · 6 of 12
Year 1 (year after)	Sep 10, 2029 - Sep 27, 2030	2030	-1.9% · 4 of 12

“On the calendar” is the calendar year that holds most of that cycle year. “S&P history” is the S&P's average return in that year of the cycle and how often it was positive, across 12 cycles (1938-2022). The highlighted row is the year we're in now.

Sep 21st, 2028The start of the next shmita year. It ends on September 9th, 2029.

Actual closes through September 28th, 2026, then a simulated path through the levels in the text: around 11,000 at the start of the shmita year, a decline of more than 30%, and the low in the year after. The path is illustrative, not a forecast.

2028 is the first clock. The second one strikes in 2035.

2035

The second clock: the end of the secular cycle

I think we're in a secular bull? market that began in 2009 and will end in 2034-2035.

The next shmita year after 2028 starts on October 4th, 2035, and runs until September 21st, 2036.

In January, I tied the two together in an article on X: the secular cycle ends right before that shmita year.

The last three major secular peaks in stocks came in 1929, 1966 and 2000. Two of them fell in year 6 or 7 of the cycle.

The February 1966 peak came in the shmita year itself. What followed was a sideways market? that lasted 16 years.

The March 2000 peak came in the year before shmita. What followed was a decade with no gains for the S&P.

The 1929 peak doesn't fit: it came in year 5.

Around 2035, two other clocks I watch line up as well.

The inflation cycle, measured in gold, has an estimated peak between 2033 and 2036.

And that's also when 99% of all Bitcoin will have been mined.

In January, I wrote that this threshold arrives in 2033. I've redone the math on the actual issuance schedule: 99% is reached around early 2035.

For 2035, I'm not setting a price level. The map below shows only time, and here time is what matters.

2035-36The next shmita year after 2028, close to the end of the secular cycle I expect in 2034-2035.

The clocks I watch, from 2020 to 2037. Time markers, not price targets.

I'm not the only one looking at 2028.

Michael Burry

Burry sees the same year, but he's speeding up the clock

Michael Burry, the investor who bet against subprime mortgages before 2008, named 2028 in August as his base case for the AI bubble bursting.

On September 28th, he wrote that he's moving his timeline earlier and wants more leverage? on his short positions?.

His posts are for paying subscribers, so I'm quoting them from the financial press.

Burry and the shmita cycle are looking at the same window. The difference is that, historically, the year before it has never ended lower.

2028 sits right on the border between the year before shmita and the shmita year itself. On that, we're on the same page.

Where we part ways is the rush.

Historically, the year before shmita was positive in all 12 cycles. The next one runs from October 2027 to September 2028.

Anyone betting on a collapse during that year is betting against an 84-year streak.

And the 2008 crisis Burry saw coming hit hardest in the shmita year too, with Lehman Brothers collapsing on September 15th, 2008.

Timing matters as much as the idea.

2028The year Michael Burry named in August as his base case for the AI bubble bursting. He now says it may come sooner.

From here, our plan is tied to the clock, not to a fixed date.

How we're positioned

Right now, in year 5, we're almost fully invested, in crypto and in stocks.

We've been bringing cash down gradually and want to be at 5-10% by January 2027.

In the year before shmita, we stay in the market. Historically, it has never been the right time to get out early.

As we approach fall 2028, we sell in stages. On altcoins? we begin earlier, in January 2028.

In the shmita year, we want to be mostly on the sidelines, with cash ready.

We rebuild positions in the year after, between fall 2029 and 2030, also in stages. The average, not the bottom.

Invalidation?: if year 5 or the year before shmita closes lower, the clock is no longer working the way it used to, and we adapt.

Conclusions

The shmita cycle isn't a prophecy. It's a clock, and over 84 years of data it has worked surprisingly well.

The year before shmita was positive 12 times out of 12.

The peak usually came in the shmita year, and the low in the year after.

On the calendar, 2027 is year 5 and 2028 is year 6, the year before shmita. The shmita year starts in September 2028.

And in 2035, the second, bigger clock strikes.

What to watch from here

Does year 5, which runs until October 1st, 2027, close higher, as it did in 11 of 12 cycles?
Does the year before shmita, from October 2027 to September 2028, extend the 12-of-12 streak?
Does the peak come in the first months of the shmita year, starting in fall 2028?

80 years of shmita, by the numbers

Keep reading

New highs are in. Now comes the best part of the year

The Olympics, the election and shmita all land in 2028

The weak window is over. Is the best quarter next?