Asset Allocation and Diversification
🗒️ Description
This is the mechanics note: how to build and maintain a multi-asset portfolio — the allocation decision, diversification math, asset-class roles, glide paths, rebalancing rules, position sizing, and the cushion that lets you hold through a bear. The philosophies (active vs passive, value vs growth, DCA, contrarianism) live in Investment Strategies; the behavioral side (why you sabotage the plan) lives in Investing Psychology and Cognitive Biases in Investing.
Written for a long-horizon (10+ yr), high-risk-tolerance builder who already holds equities/ETFs, crypto, real estate, and some bonds/cash/gold — someone who can stomach a −50% drawdown without forced selling. For that profile a high equity weight is defensible; the job of this note is to make that bet survivable rather than to dilute it.
Reference, not advice. Numbers below are sourced where possible; sample portfolios are illustrations, not recommendations. Verify any figure before acting on it.
🔗 Links
- 🧩 Why Allocation Dominates — the Brinson Figure and Its Honest Nuance
- 🧩 Diversification and Correlation — Spreading What Can Be Spread
- 🧩 Modern Portfolio Theory — Kept Practical
- 🧩 The Asset Classes and Their Roles
- 🧩 Strategic vs Tactical — and the Glide Path
- 🧩 Rebalancing — the Mechanical Sell-High / Buy-Low
- 🧩 Position Sizing the Volatile Satellites
- 🧩 The Cushion — What Lets You Hold
- 🧩 Sample Model Portfolios (Illustrations)
- 🧩 A Build Checklist
- Glossary
🧩 Why Allocation Dominates — the Brinson Figure and Its Honest Nuance
The most-quoted line in portfolio construction: asset allocation explains ~90% of performance. It comes from Brinson, Hood & Beebower, “Determinants of Portfolio Performance” (Financial Analysts Journal, 1986), which studied 91 large US pension plans 1974–83 and found that investment policy (the long-term asset-class mix) explained on average 93.6% of the variance of a plan’s return over time — far more than market timing or security selection. The follow-up (BHB II, 1991) put it at ~91.5%.
The honest nuance. That ~90% is constantly misquoted as “90% of your returns come from allocation” — which is wrong. Ibbotson & Kaplan settled it in “Does Asset Allocation Policy Explain 40, 90, or 100 Percent of Performance?” (FAJ, 2000) — the answer depends entirely on the question:
| Question asked | Answer | Meaning |
|---|---|---|
| What % of the variability of one fund’s return over time does policy explain? | ~90% | When the market swings, your portfolio swings with it — because of the mix. This is the BHB number. |
| What % of the difference in returns between funds does policy explain? | ~40% | Across funds, manager skill / timing / fees still account for ~60% of who beats whom. |
| What % of the level of the average fund’s return does policy explain? | ~100% | On average, active bets are a wash (or slightly negative after costs), so the policy mix delivers essentially all of the typical investor’s return. |
The practical reading for me: the single highest-leverage decision is the mix of asset classes, not which fund or stock. Get the weights roughly right and you have captured most of what you can control; agonizing over the “best” ETF inside a sleeve is rounding error by comparison. Time is far better spent on allocation, cost, and behavior (see Investing Psychology) than on selection.
🧩 Diversification and Correlation — Spreading What Can Be Spread
Total risk splits into two parts:
- Specific (idiosyncratic, diversifiable) risk — one company blows up, one country defaults, one protocol gets hacked. This is free to remove: hold enough uncorrelated positions and idiosyncratic shocks wash out. A single stock can go to zero; a 1,500-holding global index cannot.
- Systematic (market, undiversifiable) risk — recessions, rate shocks, liquidity crises. No amount of equity diversification removes this. Owning 50 stocks instead of 5 does almost nothing in a −35% bear; they all fall together.
The lever that controls portfolio risk is correlation, not the count of holdings. Combine two assets that don’t move together and portfolio volatility drops below the weighted average of their individual volatilities — the only “free lunch” in finance (Markowitz). The lower (or more negative) the correlation, the bigger the lunch.
The cruel catch: correlations spike toward 1 in a crisis. In a true panic, leveraged and forced sellers dump everything that’s liquid to raise cash — so assets that look uncorrelated in calm markets crash together. In 2008, REITs, commodities, hedge funds, and most “diversifiers” saw correlations jump; only gold’s correlation reverted to its pre-crisis level afterward (State Street; Morningstar, “3 Assets That Might Not Diversify as Well as You Think”). This is the same mechanism as the “global lost decade” in Bear Markets — 100 Years of History: geographic diversification (US/EU/EM equities) protects against local problems, not against a synchronized worldwide collapse — MSCI World fell ~−34% in COVID 2020, almost identical to the S&P 500.
Conclusion: real diversification means owning assets with different return drivers, not just different tickers. Equities, long bonds, gold, and cash respond to different macro forces (growth, rates, fear, liquidity), which is why a multi-asset mix survives where “50 tech stocks” does not.
The two diversifiers that have historically actually helped during equity crises:
- Long government bonds — the classic ballast; in a growth-shock / deflationary bear (2008, 2020) high-quality treasuries rallied as equities fell (negative correlation). But the stock–bond correlation is not a law of nature — it went positive in 2022, when both stocks and bonds fell together under inflation + rate hikes (State Street, “Rethinking the Role of Bonds”). Bonds hedge growth/deflation shocks, not inflation shocks.
- Gold — the most reliable crisis hedge over long windows; low-to-negative equity correlation that, unlike most alternatives, holds up in panics. No yield, but that’s the point — it’s insurance, not an engine.
🧩 Modern Portfolio Theory — Kept Practical
Harry Markowitz (1952) turned “don’t put all your eggs in one basket” into math: portfolio risk is not the average of the parts — it’s a function of each asset’s variance and the covariances between them. The practical payoff is three ideas, no matrices required:
- The efficient frontier. For every level of risk there’s a mix that gives the most expected return (and vice-versa). Portfolios on that curve are “efficient”; everything below it is leaving return on the table for the risk taken. You don’t need to solve for it precisely — you need to not sit far below it (e.g., holding 100% cash for 30 years, or one concentrated stock).
- Risk/return is a trade-off you choose, not avoid. Higher expected return demands accepting higher volatility. The frontier just tells you the best available trade at each level. A high-risk long-horizon investor deliberately picks a high-volatility point — that’s rational given the horizon, not reckless.
- The Sharpe ratio = (return − risk-free rate) / volatility — return per unit of risk. Adding a low-correlation asset (gold, bonds, a tiny crypto sleeve) can raise the whole portfolio’s Sharpe ratio even if the asset itself is volatile or low-returning, because it cuts portfolio volatility more than it cuts return. That is the entire mechanical case for diversification in one number.
Where MPT breaks (know the limits): it assumes correlations and volatilities are stable and that returns are normally distributed. Markets have fat tails (crashes are more frequent and deeper than the bell curve predicts) and correlations move (see above). So treat MPT as intuition and direction, not a precision instrument — don’t over-optimize to three decimal places on historical data you can’t trust to repeat.
🧩 The Asset Classes and Their Roles
Pick assets by the job they do, not by recent performance. Each class is a different tool:
| Asset class | Role in the portfolio | Drives return via | Behavior in a crisis | Yield/income |
|---|---|---|---|---|
| Equities / ETFs | Growth engine — the long-run wealth driver | Earnings growth + reinvested dividends | Falls hardest (−20% to −57%); recovers (see bear note) | Dividends |
| Bonds (govt / IG) | Ballast — dampen drawdowns, dry powder | Coupons + duration | Hedge growth/deflation shocks; failed in 2022 inflation | Coupon |
| Cash / T-bills / MMF | Cushion & ammunition — survive + buy the dip | Short rate | Stable nominal; the asset you deploy at the bottom | Short rate |
| Real estate / REITs | Income + partial inflation hedge | Rents + appreciation | Listed REITs trade like equities in panics (correlation rises); direct property is illiquid but less mark-to-market | Rent / dividend |
| Gold | Crisis hedge / insurance | Scarcity, real-rate & fear premium | The one diversifier that historically holds in panics | None |
| Crypto | Convex, high-vol satellite — asymmetric upside | Adoption, liquidity, risk appetite | Highest beta; correlations to equities have risen as it institutionalizes | Mostly none (staking aside) |
Notes that matter for this investor:
- Equities are the engine — for a 10+ yr horizon, the historical case is to keep this weight high. The drawdowns are the price of admission, not a reason to under-allocate.
- Bonds are ballast, not a profit center — long-term investors accept lower bond returns in exchange for an asset uncorrelated (usually) to equities (State Street). Hold them for what they do in a growth bear, and know they won’t save you in an inflation bear.
- Cash is a position, not laziness — it’s the thing that lets you buy from forced sellers at the bottom. Berkshire sat on a record ~$397B in cash in early 2026 (Bear Markets — 100 Years of History) for exactly this reason.
- REITs ≈ equities in a crisis — useful for income and inflation, but don’t count listed REITs as a separate diversifier when it matters most; since 2008 their correlation to a 60/40 portfolio has risen (Morningstar).
- Crypto is a satellite — sized so it can fall 80% without sinking the ship (next section). BlackRock frames a 1–2% bitcoin sleeve as a reasonable multi-asset allocation; academic work (Yale/Rochester) found 1–6% optimal. See Crypto Market State 2026 for the asset-specific picture.
🧩 Strategic vs Tactical — and the Glide Path
- Strategic asset allocation (SAA) — your long-term target weights, set from your horizon and risk tolerance, then held. This is the BHB-relevant decision; it does ~90% of the work. Boring and dominant.
- Tactical asset allocation (TAA) — deliberate short-term tilts away from target (overweight cheap, underweight expensive). Honest take: discretionary TAA usually destroys value because it’s market timing in disguise (recall the bear note: leaving the market to dodge declines reliably misses the best days). Only systematic, rules-based, small tilts have any defensible edge — and even those are a side bet, not the strategy. Keep tactical moves small and rare; let strategy and rebalancing do the heavy lifting.
The glide-path rule of thumb: “110 minus age” in equities. Age 35 → ~75% equities; age 60 → ~50%. The logic: more years to recover ⇒ more risk capacity. The newer “120 minus age” version pushes equity higher (longer lifespans, low bond yields).
Limits of the rule — read these before trusting it:
- It’s a heuristic keyed to a generic retiree, not to your risk tolerance, income stability, or other assets. A high-tolerance builder with stable income and a 10+ yr horizon is fully justified holding well above what the formula prescribes.
- Rising bond weight with age can increase failure risk, not lower it, given low yields and long retirements (Pfau–Kitces; Estrada). Some research finds a rising-equity glide path through retirement beats the traditional declining one.
- It ignores non-portfolio assets (a paid-off house, a pension, a business) that change your true risk capacity.
For this profile: a high equity weight (often 70–90%+ of the risk portfolio) is defensible provided the cushion and position-sizing rules below are in place — those are what convert “aggressive” into “survivable.”
🧩 Rebalancing — the Mechanical Sell-High / Buy-Low
Rebalancing = restoring your target weights after markets drift them. Its quiet genius: it forces you to trim whatever ran up (sell high) and add to whatever fell (buy low) — the exact opposite of the emotional default (chase winners, dump losers). It’s the disposition effect reversed, executed by rule instead of by nerve. In euphoria it takes chips off the most expensive asset without a forecast; in a bear it mechanically buys discounted equities — see the bear note’s phase playbook.
Two trigger styles:
| Method | Rule | Pros | Cons |
|---|---|---|---|
| Calendar | Rebalance on a schedule (quarterly / annually) | Dead simple; few decisions | Can drift far between dates; may trade when unnecessary |
| Threshold (bands) | Rebalance when a weight drifts past ±X pp (e.g. ±5pp, or relative ±25%) | Responds to actual drift; controls risk better | Needs monitoring; more potential trades |
Vanguard’s Dec-2024 research (“The Rebalancing Edge”) found threshold-based beats pure calendar on a risk-adjusted basis (~15–25 bp/yr), and recommends a hybrid: check on a schedule, act only when a band is breached. That’s the pragmatic default — quarterly look, rebalance only if something is off by more than your band.
The “rebalancing bonus.” When assets are volatile and mean-reverting, systematic rebalancing can add a small return premium on top of the risk control, by harvesting the swings. Don’t overstate it — the primary reason to rebalance is to stop your risk from silently drifting up (a 60/40 left alone for a decade becomes an 80/20 just before you need it not to be).
Costs and taxes — don’t rebalance blindly:
- Inside tax-advantaged accounts (IKE/IKZE, retirement): rebalance freely — no tax event.
- In a taxable account, selling to rebalance can trigger capital-gains tax (19% Belka tax in Poland). Prefer to rebalance with new contributions (direct fresh cash to the underweight sleeve) and, if you must sell, favor the most-overweight asset and higher-cost-basis lots. Widening bands a little in taxable accounts is rational — the tax drag can exceed the rebalancing benefit.
🧩 Position Sizing the Volatile Satellites
The rule for any high-volatility satellite (crypto, single stocks, a thematic bet): size it so the worst plausible outcome is survivable and doesn’t force a sale.
Crypto routinely draws down 70–80%+ peak-to-trough (it has done so repeatedly — see Crypto Market State 2026). So size the sleeve against that, not against the hope:
- A 5% crypto sleeve that falls 80% costs the total portfolio −4% — annoying, fully survivable, doesn’t touch the plan.
- A 25% crypto sleeve that falls 80% costs −20% of the whole portfolio on top of an equity bear — that’s the kind of hit that triggers panic-selling at the bottom (the worst possible moment).
The sizing test (BlackRock / practitioner framing): if this sleeve dropped 80% overnight, would I still sleep and stick to the plan? If no, it’s too big. A high-tolerance investor might run a larger satellite (say up to 5–10%) than a cautious one — but the principle is identical: the position is sized by your tolerance for its drawdown, not by your conviction in its upside. Convexity (small bet, asymmetric upside) only works if the small bet is genuinely small. Treat the sleeve as a separate line, rebalance it back to target when it balloons (banking gains) and — within plan — top it up when it craters.
🧩 The Cushion — What Lets You Hold
The thing that quietly determines whether your whole strategy works isn’t an asset class — it’s the emergency fund / cash cushion sitting outside the risk portfolio.
- What: typically 3–12 months of living expenses in cash / T-bills / money-market — instantly accessible, not marked to market, never invested.
- Why it’s the keystone: it removes the one thing that destroys long-term investors — forced selling at the bottom. In a bear, those who must sell (margin calls, redemptions, job loss with no buffer) dominate those who want to, which is what overshoots prices to the downside (Bear Markets — 100 Years of History, capitulation phase). If you never have to sell, a −50% drawdown is a paper number you wait out; without a cushion it can become a permanent, realized loss. The cushion is what converts “high risk tolerance” from a feeling into an actual capacity.
- Dual purpose: it’s also ammunition — the dry powder to buy from forced sellers when assets are cheapest.
For this investor the cushion is prerequisite to the high equity weight, not an alternative to it: the bigger and more stable the cushion (plus stable income, no leverage), the more aggressive the risk portfolio can responsibly be. Leverage does the opposite — it manufactures forced selling, so the long-term portfolio should ideally carry none.
🧩 Sample Model Portfolios (Illustrations)
Illustrations only — to show how the pieces fit, not recommendations. Each reflects a different risk appetite and worldview. The cushion (above) sits outside all of these.
| Portfolio | Equities | Bonds | Gold/Commodities | Cash | Crypto | Character |
|---|---|---|---|---|---|---|
| 60/40 (classic balanced) | 60% | 40% | — | — | — | The default benchmark; simple, struggled in 2022 |
| Three-fund (Bogleheads) | ~80% (US + ex-US) | ~20% | — | — | — | Low-cost, total-market, set-and-forget |
| All-Weather (Dalio, risk-parity) | ~30% | ~55% (long+intermediate Treasuries) | ~15% | — | — | Smooth ride across regimes; lower expected return, higher bond reliance |
| Aggressive long-horizon | ~85% | ~5% | ~5% | ~5% | — | High-tolerance builder; equities do the work |
| Core + crypto satellite | ~75% | ~10% | ~5% | ~5% | ~5% | Core indexed; small convex crypto sleeve sized to survive −80% |
Reading them: the All-Weather trades return for a smoother ride (heavy bonds + gold balance growth and inflation shocks); the three-fund maximizes simplicity and cost-efficiency; the bottom two fit this profile — a high equity core, a thin gold hedge, a cash cushion, and a deliberately small crypto satellite. The right one is whichever you can actually hold through a −50% drawdown without selling — that constraint, not back-tested return, is what makes a model portfolio “yours.” The deeper philosophical trade-offs behind these (indexing, factor tilts, contrarianism) are in Investment Strategies.
🧩 A Build Checklist
A one-page version of the mechanics:
- Cushion first. 3–12 months of expenses in cash, outside the risk portfolio. This is what lets you hold.
- Set strategic weights from your horizon (10+ yr ⇒ high equity) and your honest drawdown tolerance — not the 110-minus-age formula in isolation.
- Diversify by return driver, not ticker count: a growth engine (equities/ETFs), ballast (bonds), a crisis hedge (gold), and a cushion (cash). Add income/inflation (REITs) and a convex satellite (crypto) to taste.
- Size satellites by their drawdown, not their upside: a crypto sleeve small enough that −80% is a flesh wound (e.g. ≤5–10%).
- No leverage in the long-term portfolio — it manufactures forced selling.
- Rebalance by rule: check quarterly, act on ±5pp bands; rebalance with new cash first; mind taxes in taxable accounts.
- Stress-test: compute the portfolio after −22% / −35% / −50% (the bear-note medians). If any result would force or panic you into selling, your risk weight is too high today.
- Write the plan down (Bear Markets — 100 Years of History) so in a panic you execute, never decide. The biases that wreck step 6 in real time are catalogued in Cognitive Biases in Investing and Investing Psychology.
Glossary
- Asset allocation — the split of capital across asset classes (equities, bonds, cash, gold, real estate, crypto). The dominant driver of return variability.
- Strategic (SAA) vs tactical (TAA) allocation — long-term target weights, held vs short-term deliberate tilts away from them.
- Diversification — combining assets of different character to cancel specific risk and lower portfolio volatility below the weighted average.
- Specific vs systematic risk — diversifiable (one holding) vs undiversifiable (the whole market / macro).
- Correlation — co-movement of two assets, −1 to +1. Low/negative = good diversifier. Spikes toward +1 in crises.
- Covariance — the unscaled co-movement term that, with variances, sets portfolio risk in MPT.
- Modern Portfolio Theory (MPT) — Markowitz’s mean–variance framework: optimize expected return for a given risk via the covariance structure.
- Efficient frontier — the set of portfolios giving the most return per unit of risk.
- Sharpe ratio — excess return ÷ volatility; risk-adjusted return in one number.
- Glide path — a rule that shifts allocation (usually equity↓) as you age; “110/120 minus age.”
- Rebalancing — restoring target weights; mechanically sells high / buys low. Calendar vs threshold (band) triggers.
- Rebalancing bonus — small extra return from systematically harvesting volatility via rebalancing.
- Ballast — bonds’ role of dampening drawdowns (works for growth shocks, not inflation shocks).
- Cushion / ammunition — emergency cash outside the portfolio: prevents forced selling and funds buying the dip.
- Satellite (core-satellite) — small high-conviction / high-vol position around a broad indexed core.
- Position sizing — choosing a position’s size so its worst plausible drawdown is survivable.
- Risk parity — allocating by risk contribution rather than dollars (the All-Weather idea).
- Fat tails — crashes are more frequent/severe than a normal distribution predicts; the main reason MPT under-states real risk.
📖 Further reading/watching
- Brinson, Hood & Beebower, “Determinants of Portfolio Performance”, Financial Analysts Journal, 1986 (and BHB II, 1991)
- Roger Ibbotson & Paul Kaplan, “Does Asset Allocation Policy Explain 40, 90, or 100 Percent of Performance?”, FAJ, 2000
- Harry Markowitz, “Portfolio Selection”, Journal of Finance, 1952; William Sharpe (Sharpe ratio, CAPM)
- Vanguard, “The Rebalancing Edge: Optimizing Through Threshold-Based Strategies”, Dec 2024
- State Street Global Advisors, “Gold as a Strategic Asset Class” & “Rethinking the Role of Bonds in Multi-Asset Portfolios”
- Morningstar, “3 Assets That Might Not Diversify as Well as You Think”
- BlackRock Investment Institute, “Sizing Bitcoin in Portfolios”
- Pfau & Kitces, “Reducing Retirement Risk with a Rising Equity Glide Path” (2013)
- Bogleheads wiki — “Three-fund portfolio” & “Rebalancing”; Ray Dalio, Principles / All-Weather (Bridgewater)
- Related: Investment Strategies · Investing Psychology · Cognitive Biases in Investing · Bear Markets — 100 Years of History · Crypto Market State 2026
Template: knowledge_note_info