Time: The Fourth Variable

I

The silent variable

The first three parts of this series have treated investing as a question of what to buy, how much, and at what price. But there is a fourth dimension that quietly shapes every outcome: time. How long you hold an asset changes its expected return, its volatility profile, and — most importantly — whether variance helps you or destroys you.

Most investing content treats time as a background assumption: “invest for the long term.” But the long term is not monolithic. A 5-year horizon and a 30-year horizon are not just different versions of the same bet — they are mathematically different games played with the same pieces. And the closer you get to needing the money, the more the rules flip.

This article covers three things the prior parts implied but didn’t address. First, why long horizons collapse variance into something close to certainty — and why short horizons do the opposite. Second, why a bad year at age 62 can end a retirement that a bad year at age 32 wouldn’t even register. Third, when a rational investor should actually sell.

II

Time is variance's enemy

From Part I we know that variance is the enemy of the arithmetic-geometric gap. What Part I didn’t emphasize is that this relationship has a time component. Over a single year, the S&P 500 can return anywhere from −40% to +50%. Over 30 years, the annualized return has never historically fallen below roughly 7% and has never exceeded about 14%. The extremes don’t disappear — they average out.

Over a 1-year horizon, the S&P 500 has ranged from about −43% to +54%. Over a 5-year horizon, from about −12% to +29% annualized. Over 20 years, roughly +3% to +18%. Over 30 years — every single 30-year rolling period in modern US history has been positive, ranging from roughly +8% to +14% annualized.

A caveat worth burning into your forehead: this is US data. Dimson, Marsh & Staunton (Triumph of the Optimists) document that Italian, Japanese, and Belgian investors saw 30-year real returns near zero. The US is the survivor. Do not assume your country, or the next century, will be.

Over one year, the stock market is gambling. Over thirty years, it is arithmetic.

The underlying mechanism is simple. Each year’s return is roughly independent of the previous year’s. When you average many independent random variables, the standard deviation of the mean shrinks proportionally to 1/√n. The variance tax from Part I still exists — but over long horizons, the law of large numbers brings the geometric return closer and closer to its expected value.

Caveat — and this is the one most retail content omits: Samuelson (1969) and Bodie (1995) argue that “time diversification” is a fallacy for a constant-utility investor. Annualized variance shrinks with horizon, but the dispersion of terminal dollar wealthgrows. Both views are partially true: annualized risk falls, terminal-wealth uncertainty rises. Which one matters depends on whether you’re still compounding or already spending.

Horizon math is not the same across asset classes

Equities are the canonical case for variance-collapse-with-horizon because they have a positive long-run drift and roughly independent year-to-year returns. The other asset classes from Part II have their own horizon dynamics, and three of them break the equity intuition.

Treasury bondshave an explicit, deterministic horizon: duration. A 10-year Treasury held to maturity returns approximately its starting yield, full stop — variance-collapse is built into the asset, not statistical. Held shorter than duration, you eat realized rate moves; held longer, you reinvest at uncertain forward rates. Match the bond’s duration to your horizon and you neutralize most rate risk. T-billsare the limit case: zero duration, zero horizon-mismatch, and the explicit risk-free anchor for every other asset’s expected return.

Gold and broad commodities mean-revert on decade-scale cycles, not annual ones. Gold can spend 5–10 years underperforming (2011–2018) and then 5 years compounding faster than equities (2019–2024). Commodity supercycles have spanned 15–20 years historically. The relevant horizon for these assets is not 1, 5, or 30 years — it is at least one full cycle, often two. Holding gold or commodities for under 5 years is essentially a bet on a single regime; holding them across cycles smooths the path but does little for terminal wealth.

Bitcoin has too short a history to claim horizon-collapse with confidence. The 2015–2025 window contains three completed cycles, each with progressively smaller drawdowns and progressively smaller cycle-bottom premiums (Part III). On a 15-year forward horizon the central question — whether Bitcoin continues to compound at equity-plus rates or reverts to gold-like behaviour — is unresolved by the data. Treat Bitcoin as an asset whose long-horizon variance has not yet collapsed.

III

Sequence risk — when the order of returns destroys you

Everything in Section II assumes you are either accumulating money or just holding it. The math changes completely the moment you start withdrawing. When you pull money out of a portfolio, the order in which returns arrive suddenly matters enormously. This is called sequence-of-returns risk, and it is the single most underappreciated concept in retirement planning.

Same asset. Same average return. Same withdrawal rate (the canonical 4% rule from Bengen 1994 and the Trinity Study, 1998). The only difference is whenthe bad years happened. Retiree A died poor. Retiree C died with more money than they started with. More recent valuation-aware work by Pfau suggests SAFEMAX may be closer to 3.3% in today’s high-CAPE environment.

The practical takeaway from sequence risk is brutal: the first 5–10 years of retirement matter disproportionately. A strong market in that window nearly guarantees a comfortable retirement. A weak market in that window can irreversibly compromise it, even if the long-run average is fine.

IV

Why the glidepath exists

The traditional “glidepath” advice — hold more stocks when young, more bonds as you age — isn’t arbitrary. It is a direct response to the interaction between time horizon and sequence risk. Young investors benefit from variance because they have decades to let the law of large numbers work. Near-retirees face the opposite: a major drawdown in their final accumulation years can delay retirement by a decade.

A 25-year-old with 90% stocks is making the mathematically correct bet. A 65-year-old with 90% stocks is gambling with their retirement security. The asset has not changed. The person’s relationship with time has.

Research by Wade Pfau and Michael Kitces (Journal of Financial Planning, 2014: Reducing Retirement Risk with a Rising Equity Glide-Path) showed that retirees who follow a rising equity glidepath — starting retirement with lower stock allocations and gradually increasing them — often outperform those who start high and decrease. They reduce exposure to sequence risk exactly when it matters most (the early years), and re-embrace equities after the danger zone has passed. Counterintuitive but mathematically sound.

Practical implementation — the “bond tent” or cash buffer: hold 2–3 years of expected withdrawals in cash or short-duration bonds entering retirement. This lets you avoid selling equities during a drawdown — you spend the buffer until the market recovers, then refill it from rebalancing once equities are back near target.

V

When should you actually sell?

The series has so far avoided this question. Here is a framework, grounded in the prior parts.

Your need for the money is approaching.

Money needed within 2–3 years should not be in volatile assets, period. Not because the assets are bad, but because your window is too short for variance to average out. The asset is fine; the horizon is wrong.

Valuations are extreme and you have a rebalancing rule.

Part III showed that starting valuations have enormous predictive power. Having a disciplined rebalancing rule — trimming positions back to target when they exceed it — is the right response. You’re not trying to catch the top. You’re refusing to let a single position become catastrophic concentration.

The thesis is broken.

For individual positions, sell when the reason you originally bought is no longer valid. Not when the price goes down — when the storychanges. “Would I buy this position today at this price, knowing what I now know?” If no, you should not continue holding it.

You need to fund your life.

An investment portfolio exists to serve your life, not the other way around. Selling to fund your retirement is not failure — it is the whole point.

You do not owe your portfolio eternal loyalty. You owe yourself a life that uses it.

Almost never sell because of a crash.

Selling during a drawdown is the single most destructive act a long-horizon investor can perform. It locks in paper losses, disconnects you from the recovery, and moves money into whatever felt safe at peak fear. If your horizon is long and your withdrawal need is not immediate, a crash is not a sell signal — it is the exact opposite.

VI

The time-based Kelly adjustment

Part I introduced Kelly sizing: bet in proportion to your edge. But Kelly assumes you are playing the game indefinitely. When your time horizon is finite and shortening, the optimal bet size shrinks even if your edge remains unchanged. A bad draw late in the game leaves no time to recover.

Time is part of your ruin tolerance. A young investor’s ruin tolerance is effectively infinite on most single bets, because they have time to rebuild. An older investor’s ruin tolerance is what’s left in their portfolio when they stop earning. These are not the same quantity, and Kelly sizing should reflect the difference.

VII

Common questions