Skip to main content

Compounding · A reproducible essay

The Long Hill

Warren Buffett once said that building wealth is like a snowball — all you need is wet snow and a really long hill. This essay is about the hill: how long yours is, and what a few forgettable dollars a day would have done rolling down it. Two honest questions, real market history since the 1870s, and no way to hide the losses.

The invitation

What wouldn't you notice leaving your account?

Not a budget. Not a sacrifice. Just the small, forgettable amount that could slip out the day after payday without changing a single thing about your week — the coffee you'd have bought anyway, the subscription you forgot you were paying for.

Name that number honestly and something strange happens to it over a few decades. Not because the number is large. Because the hill is long, and the snow, most years, is wet.

Life is like a snowball. The important thing is finding wet snow and a really long hill.— Warren Buffett

Three honest inputs

Two numbers, answered honestly.

This should feel less like a form and more like answering a friend. Set the two numbers below and every figure in the rest of this essay recomputes from them — the history behind you, and the hill ahead.

The amount · a day
$3
$90 a month — the amount you'd genuinely not miss.
Your age · today
26
39 years of hill left before 65.
Read every figure in
Dollars of the day — not adjusted for inflation.

Where does it go? Three honest destinations, in rising order of risk: something safe (short Treasuries), the broad market index, or a single story basket. We'll open on the index — the honest default, five hundred companies at once, dividends reinvested — then weigh it plainly against the safe road below.

Figure 1 · What could have been

If you had started at 18, you would have put in $8,640.

Suppose that on the day you turned 18 you began setting aside $3 a day into the broad index, and never skipped — through 8 years of booms, crashes, and forgettable Tuesdays. Here is what the market did with it. The dashed line is your own money; everything above it is what compounding added — and where the line dips below, the months you were briefly underwater. The calm slate line is the very same money in something safe — short-term Treasury bills; the sage gap between it and the hill is the premium you are paid for bearing the ride.

What you put in
Something safe · T-bills
The market's premium
The market hill
$4.4k$8.8k$13k$18k20182020202220242026
Something safe would give
$9,856
short Treasuries · the near-sure baseline
Your reward for the ride
+$5,745
the market's premium over safe — this gap is the whole lesson
Contributed
$8,640
what you put in
The market added
+$6,961
compounded on top
Ending balance
$15,601
Mar 2026
Your return
+14.7%
money-weighted / yr

Real S&P total-return history, dividends reinvested, 2018–2026. "Your return" is the money-weighted return of your actual monthly stream — not the index’s headline number, which ignores when each dollar arrived. Figures are before tax and inflation. Every number here recomputes from the two inputs above.

Figure 2 · The road ahead

From here to 2065, the honest range.

From today, $42,120 would go in over the next 39 years. Where it lands is not one number — it depends entirely on which stretch of history you happen to live through. Rather than draw a smooth fan of probabilities, which would imply a precision nobody has, here is every real stretch at once.

One real history
Median outcome
Worst start on record
What you put in
$303k$606k$909k$1212ktypicalworst20262036204620552065

Each faint line is one real 39-year window of market history, from 1871 to 1987 — the same habit run through all 1,396 of them, each ending on your 65th birthday. Every dot on the right is where one of those histories actually landed.

A typical outcome
$525,852
12.5× what you put in · the median history
The worst start on record
$102,824
2.4× what you put in · the honest low end
The best start on record
$1,154,547
the luck you cannot count on

Discrete real histories, not a probability fan. Overlapping 39-year windows, US market, dividends reinvested, before tax and inflation. The past is the only data there is — it is not a promise. Not a forecast, and not advice.

Before you do anything

First, the unglamorous part.

If you are carrying high-interest debt — a credit card at 22% — paying it down beats any market return on this page, guaranteed and tax-free. Keep a small cushion for the month the car breaks and the roof leaks in the same week.

Only then does the hill make sense. This is general education, not advice for your particular situation. The hill will wait for you.

Figure 3 · The price of safety

What if you took the sure thing?

Not everyone belongs on the hill. If you'll need this money within a few years, or the swings would cost you sleep you can't spare, the honest answer is the calm one: short-term U.S. Treasury bills — lending to the government a few months at a time. It is about as safe as money gets.

Here is the same $3 a day poured into that instead. It roughly keeps pace with inflation and little more — no crashes, no drama, and no long hill either. The faint line above is the market you didn't ride; the muted band between them is the price of safety.

What you put in
Something safe · T-bills
The price of safety
The market hill, not taken
$4.4k$8.8k$13k$18k20182020202220242026
Contributed
$8,640
what you put in
Safe ending
$9,856
+$1,216 over what you put in
Safe return
+3.3%
money-weighted / yr
The hill would have added
+$5,745
the price of safety, in full

Same window as Figure 1 (20182026), same monthly habit — only the destination changed. Real 3-month Treasury-bill history, before tax and inflation. The safe line almost never falls; it also almost never runs.

The honest risk ladder
01
Pay off high-interest debt
Guaranteed and tax-free. A 22% card beats every hill on this page. First, always.
guaranteed
02
Short Treasuries
About as safe as money gets. Keeps pace with inflation, little more — the calm baseline you just saw.
near-sure · modest
03
The broad index
The risk premium. Decades of ups and downs — and, historically, the reward for sitting through them.
higher · volatile
04
A single story basket
Concentration risk. Could soar, could sink below what you put in. Play money only — never the whole hill.
highest · unsure

Safe is not the mistake.

For money you'll need soon, or nerves a downturn would wreck, the calm line is the right line — full stop. The lesson here isn't "safe is bad." It's the tradeoff, seen plainly over a long hill: the safe road asks nothing of your stomach and pays you little; the market asks for patience through every crash and, historically, has paid you for it. Which price you'd rather pay is a personal question, not a math one.

Four stories

You didn't need to pick the winner.

Here is the tempting version. Instead of the whole market, what if the same $3 a day had ridden five companies you would have felt certain about? Four baskets below, each a story you already know. The names are sealed.

Pick one, watch the same habit play out — then turn it over. At least one of these lost money. The reveal tells you which, as plainly as it tells you the wins.

The hard break

Notice what does not change: the road ahead in Figure 2 stays on the broad index, even while you ride a story here. Projecting a hand-picked basket of five names decades forward is exactly the false precision this essay refuses. You would never stake a retirement on five names — so we never do.

The graveyard · after the stories

Now the names that didn't make it.

The stories above share one quiet thing: every name in them survived long enough to chart. That is survivorship — the failures fall off the list, and what is left starts to look like a plan. So here is the honest rest: the names two earlier eras were just as sure of, and what a patient buyer was actually left holding. The survivors below did fine — steady monthly buying even rescued the ones that crashed. But two came back worth less than what went in, and two went to zero.

The Dot-Com Believer

2000

The five most-hyped names of the 2000 tech peak. Three are still here — and monthly buying rescued even the ones that crashed for years. Two are not here at all, and those are the ones survivorship bias hides.

MSFTMicrosoft
$349,725
12.2× what you put in · +15.8%/yr money-weighted

The one name of the 2000 hype that never needed rescuing — bought at the peak, it still compounded at double digits for 26 years.

CSCOCisco
$183,052
6.4× what you put in · +12.0%/yr money-weighted

Bought at the very top of the bubble, it fell hard and stayed down for years — yet buying a little every month through the wreckage still ended well ahead.

INTCIntel
$207,064
7.2× what you put in · +12.8%/yr money-weighted

A decade-plus in the wilderness after 2000, and a brutal 2024-25 — but averaging in every single month turned even this laggard into a multi-bagger.

Almost all of this came from a sharp 2026 rally — the same monthly buying was only about 2x as recently as early 2026. A real move (verified against a second source), but a very recent and volatile one.

AOLAmerica Online

There is no clean series to buy monthly, but a 2000 believer came back worth roughly 70 cents on the dollar — a slow, quiet loss.

Merged into Time Warner in 2001 in what is still called the worst deal in corporate history, then was spun off and sold for parts.

Approximate — no reliable free price series survives to compute a multiple.

YHOOYahoo

Even with the Alibaba windfall, a peak-2000 buyer still ended down about 40% — and only because of a side bet, not the business.

An early bet on Alibaba quietly saved it from zero; the company sold its core to Verizon, renamed itself Altaba, and slowly liquidated over years.

Approximate — no reliable free price series survives to compute a multiple.

The 1980 Blue Chips

1980

The safe, boring blue chips of 1980 — the names everyone owned and nobody worried about. Three compounded for four decades. Two took their shareholders to zero, and the companies trading under their names today are strangers.

IBMIBM
$791,460
15.8× what you put in · +9.5%/yr money-weighted

A boring, real compounder — dividends reinvested, held quietly from 1980 straight through to today.

GEGeneral Electric
$1,490,002
29.7× what you put in · +11.4%/yr money-weighted

The bluest chip of 1980 — it nearly died in 2008 and later broke itself into three companies, yet the monthly habit rode all of it.

Excludes the value of the 2023-24 GE HealthCare and GE Vernova spin-offs, so this understates a 1980 holder's true return.

XOMExxon Mobil
$651,813
14.5× what you put in · +10.4%/yr money-weighted

A century of oil that never stopped paying a dividend — and every one of them reinvested, month after month.

Series starts in 1985, the vendor's clean start for the modern ticker (five years after the other 1980 names).

GMGeneral Motors
$0

Filed for bankruptcy in June 2009 — the old shares were cancelled, worth exactly zero.

The 'GM' that IPO'd in 2010 is a legally different company; the 1980 shareholder recovered nothing.

EKEastman Kodak
$0

Filed for bankruptcy in January 2012 — the old stock went to zero as digital cameras erased the film business it dominated.

The Kodak (KODK) trading today is a new company built from the wreckage; the 1980 shares never recovered.

The broad index in Figure 1 held every name on this page — the survivors, the mergers, and both zeros — and still climbed the hill. It never needed to know which would be which. That is the case for the whole basket in one sentence: you are not paid to dodge the losers, only to hold the market long enough for the winners to more than cover them.

What to do Monday

One small transfer.

If any of this moved you, the action is far smaller than the essay. Set up one automatic transfer — $3 a day, or $90 once a month — into a broad, low-cost index fund, dated the day after payday so you never watch it leave.

Then do the single hardest thing in investing, which is nothing at all. Don't check it. Don't tune it. Let the snow be wet and the hill be long.

Method & disclosure

Every number on this page recomputes from your two inputs; none is a headline figure lifted from elsewhere. The intent, in the house style, is that every numeric claim cites a reproducible run.

Market history
Real S&P Composite total-return history — the S&P 500 since 1957, reconstructed predecessor indexes back to 1871 — dividends reinvested, monthly from 1871-01 to 2026-03. Source: Robert Shiller's public long-run dataset, checksum-pinned in the committed feed.
The safe baseline
3-month U.S. Treasury bills (FRED TB3MS, secondary-market rate), monthly from 1934-01 — the near-sure baseline drawn in slate throughout. Roughly inflation-tracking; about as safe as a nominal return gets.
Your return
Money-weighted (internal rate of return) of your actual monthly stream — it accounts for when each dollar arrived, unlike an index CAGR.
The road ahead
Discrete, overlapping historical windows drawn at your exact horizon to age 65 — real paths, not a smoothed probability fan. Every terminal dot is one real window's outcome. Always the broad index.
Story portfolios
Fixed real windows for five named companies each; outcomes shown as the blended basket, names sealed until you reveal them, losses shown as losses.
Inflation
The Nominal / Today's-dollars toggle up top deflates every cashflow by U.S. consumer prices (CPI, Shiller cpi_nsa) to 2026-03 purchasing power. Real mode adjusts the cashflows, not the price series — the money-weighted way to read a real return.
Not included
Taxes, fees, inflation, and currency. USD only. This is general education, not personal financial advice.
The data feed
Committed alongside this page: engine long_hill/1.0.0, data through 2026-03, Shiller source pinned by SHA-256 (ded04739860a…).