The 12% Lie
What 99% of Investors Are Told

What actually happens to the S&P 500's average return once it meets your money.

I.

January 2000. 

The S&P 500 sat near an all-time high, tech stocks priced for a future that hadn't arrived yet, and somewhere in the country a worker retired with a million-dollar portfolio and a plan to draw it down at 4% a year — the withdrawal rate an obscure California planner named William Bengen had spent the 1990s defending as safe, based on decades of historical data. Within three years the dot-com collapse cut the market roughly in half. Before that portfolio had time to recover, 2008 cut it again. This retiree hadn't done anything wrong. He'd followed the rule, held the index, diversified the way every book told him to. He'd simply started at the wrong moment, and the wrong moment turned out to matter more than any thirty-year average he could have quoted on his way out the door.

Now compare him to a colleague who kept working through both crashes, still funding a 401(k) every two weeks. That colleague watched the same headlines and probably felt just as sick about them — the balance on the statement fell just as hard, in percentage terms. But the paycheck kept buying shares, and it bought more of them when prices were cheap. The market didn't treat these two people identically, even though both were, in the most literal sense, invested in the same index over the same twenty years.

The arithmetic explains why. Drop a portfolio 50% and it needs a 100% gain just to get back to even — not another 50%. Rally 50% off the bottom instead, and the portfolio is still 25% short of where it started, because that second move lands on a pile of money that's already half its old size. Someone still buying barely feels this distortion, since new contributions keep landing on both sides of the crash. Withdrawing through it is a different story: every dollar pulled out during the drawdown is a dollar that never gets to participate in the recovery, because the shares sold at the bottom are gone. They don't come back when the market does.

This is what the financial planning world calls sequence-of-returns risk, and it's the reason two people can hold the exact same fund, experience the exact same twenty-year average return, and end up in entirely different financial positions. Researchers who've run the numbers on historical retirement dates keep landing on the same handful of bad years — 1966, 1973, and 2000 show up again and again as the worst times in the last century to have retired, not because the following decades were poor on average, but because the first five to ten years were.

None of this is an argument against buy and hold. The strongest data point in its favor: no rolling twenty-year period in S&P 500 history has ever produced a negative total return — good news for anyone with a long enough runway and no near-term withdrawals to make. For most people still building wealth, low-cost exposure to American corporate earnings remains more sensible than trying to time the next turn. The problem isn't the strategy. It's what gets bolted onto the strategy — a single return figure treated as a promise instead of a historical result.

Long-run S&P 500 total returns do run close to 10% nominal, depending on which dates you use. Fidelity puts the thirty years through 2025 at 10.4%. Since the modern 500-stock index began in 1957, the inflation-adjusted figure lands somewhere between 6.5% and 6.8% across independent calculations — a tighter, less flattering number than the one usually printed at the top of a retirement projection.

Neither number arrived smoothly. Both are the sum of decades that included 2000, 2008, and 2022, not a steady climb that happened to average out. Treating either one as a coupon the market owes you, rather than a historical outcome that depended heavily on timing, is where the trouble starts.

A familiar ticker does not make an aggressive assumption conservative.

That trouble is getting harder to see because a lot of current return assumptions quietly extend the best features of the last fifteen years — fat corporate margins, mega-cap valuations that keep expanding, discount rates that kept falling, and a US market that kept outrunning the rest of the world. Perhaps all of that persists. But a thirty-year plan that requires it to persist is a fundamentally different plan than one built to survive if it doesn't, and most retirement calculators don't ask you which one you're building.

A familiar ticker does not make an aggressive assumption conservative.

II.

Use a range, not a promise

I would never anchor a retirement portfolio, a fund target, or a company treasury to a single return figure. The useful question isn't “what does this become at 12%?” It's “what survives if the first decade looks like 2000 instead of 1982?” Someone who retired in 1982 walked straight into one of the longest bull markets in US history. Eighteen years later, someone else retired with a nearly identical résumé and an identical withdrawal rate — and walked straight into two of the worst stretches the market had to offer instead. The plan on paper was the same plan. The outcome wasn't, and no amount of averaging the two eras together would have warned either retiree which one they were about to live through.

Start by running three cases instead of one — optimistic, base, and stressed — and let the stress case actually hurt. Push the starting valuation higher, since expensive markets have historically delivered lower forward returns than cheap ones, and put the large drawdown in the first five years instead of the twentieth, since that's where the damage compounds. Raise the inflation assumption above the last decade's unusually calm numbers, and layer in the fees, taxes, and withdrawals that retail calculators tend to strip out because a bigger number sells better than an honest one. None of this is about predicting the exact sequence the next thirty years will hand you — nobody can do that — it's about finding the sequence that breaks the plan, so you know it's there before it happens instead of after.

Keep four numbers separate while you're doing it, because they get blurred into one constantly and the blurring is where the false confidence comes from.

Figures illustrative and rounded · nominal includes reinvested dividends · ZMACRO / ZTrader Research

Metric	What it measures	30-Year 1996–2025	Since 1957	Since 1928
Arithmetic average	Simple mean of yearly returns	≈1–2 pts above CAGR	≈1–2 pts above CAGR	≈1–2 pts above CAGR
CAGR (nominal)	What the money actually compounded at	10.4%	≈10.4–10.7%	≈10.0%
Real return	CAGR minus inflation	≈7.5%	≈6.5–6.8%	≈6.9%
After costs & withdrawals	What an investor actually keeps	Fund fee alone (~0.03%) leaves the real return nearly intact. A 4% annual withdrawal is sequence-dependent and can equal 0% in a bad first decade, regardless of the average.

Start with the arithmetic average, which is also the most misleading of the four — it's just the mean of the yearly returns, and it ignores that losses and gains don't offset symmetrically once you're compounding. 

CAGR fixes that: it's what your money actually did, asymmetry and all. Strip out inflation next and you get the real return, which matters enormously over thirty years even at a mild 3% annual pace. 

What's left after that — the return after costs and withdrawals — is the only one of the four that describes what you personally get to keep. It's usually the smallest number on the page, and the one every projection quietly leaves out.

Bengen's original 4% rule already built in some cushion for sequence risk — it was designed using the worst historical starting years, not the average ones, which is why it held up reasonably well even for the 2000 retiree. But it assumed a fixed withdrawal pattern, adjusted only for inflation, with no flexibility. More recent research on dynamic guardrails — cutting withdrawals modestly in bad years, letting them rise in good ones — has shown meaningfully better survival odds for a portfolio than sticking rigidly to a number set on day one. The lesson generalizes past retirement withdrawals: a plan that can bend under a bad sequence tends to outlast one that can only succeed if the sequence behaves.

If a plan only works at 12%, that plan is leveraged to optimism, whether or not anyone involved would describe it that way.

The S&P 500 can print a 12% average over some selected window, and that statistic can be completely accurate and still tell you nothing useful about your own compounding path. Your capital has to travel through every drawdown, every valuation reset, and every inflation regime between the first dollar you put in and the last one you take out. The index doesn't skip any of them on your behalf, and it doesn't care what year you happened to start.

The average lives in the spreadsheet. Your wealth lives in the sequence.

Sources: