# Portfolio of portfolios: how nested portfolios work

Canonical URL: https://marketheist.io/portfolio/learn/portfolio-of-portfolios
By Jeffrey Lin · Published 2026-11-12 · Updated 2026-11-12 · Data through Sep 2026
Category: Construction

## Key takeaways

- A portfolio of portfolios has two layers: the parent splits the money between its sleeves (the portfolios it holds), and each sleeve splits its share among its own funds. A fund held by several sleeves adds up its share from each.
- Each layer rebalances on its own schedule: a sleeve resets to its own mix, the parent resets the sleeves to their shares of the whole. Between resets, both the sleeves and the funds inside them drift.
- Risk is measured per sleeve: a sleeve with a fifth of the money can carry well over a fifth of the portfolio's risk.
- Combining two well-known diversified portfolios can add much less diversification than their names suggest, because they often hold the same building blocks; the correlation between the sleeves is the number to check.
- MarketHeist builds each sleeve as its own portfolio first, then runs the parent over the sleeves' monthly returns with the same simulation as any portfolio, so a nested portfolio gets the same metrics, risk split and checks as a plain one.

## Definition

A portfolio of portfolios, or nested portfolio, is a portfolio whose holdings are themselves portfolios, often called sleeves, each with its own target weights and rebalancing rule, combined at weights set by a parent portfolio. The parent rebalances between the sleeves and each sleeve rebalances within itself, so the whole behaves like one portfolio whose look-through weights move between resets.

Also known as: nested portfolio, portfolio of portfolios, sleeves, multi-portfolio allocation, combining lazy portfolios, fund of funds, core and satellite portfolio.

## How it works

Start with two complete portfolios you like, say the Golden Butterfly (equal fifths of US stocks, small-cap value, long Treasuries, short Treasuries and gold) and the All Weather portfolio (mostly Treasuries, with stocks, gold and commodities). Instead of merging their holdings into one long list, hold each as a sleeve, a portion of the parent with its own rules, and decide how much of the whole each sleeve gets. A plain fund can sit beside them as a sleeve of its own.

The two layers rebalance separately. Inside each sleeve, the template keeps its own schedule and resets to its own mix. At the parent level, the sleeves drift with their own returns and the parent resets them to their target shares on its schedule. The result differs slightly from a single flat portfolio with the same look-through weights, because the flat one would reset every fund to target at once, while the nested one lets the funds inside a sleeve drift against each other until that sleeve's own reset.

Risk is read one layer at a time. At the parent level, each sleeve is one return stream, and its share of the parent's risk depends on its volatility and how it moves with the other sleeves, exactly as for funds in a plain portfolio. Two sleeves that hold similar things move together and count as one bet, however different their names. A nested portfolio is like a team of teams: the coach decides how much playing time each squad gets, and each squad picks its own line-up; whether the squads add up to a balanced side depends on how differently they play.

Nesting also keeps decisions separate. A sleeve can be a published template followed as written, a strategy with its own rules, or a single fund, and changing the parent's split does not disturb what any sleeve holds inside. That makes it easy to ask questions such as how much of a portfolio to give each philosophy, or what happens when one sleeve is swapped for another, while every sleeve stays recognisable.

## Formula

**Look-through weight of a fund**

```
w_f = \sum_s W_s w_{f∣s}
```

where:

- w_f: share of the whole portfolio in fund f, at the parent's and the sleeves' target weights
- W_s: the parent's target weight for sleeve s (the parent's weights sum to 1)
- w_f|s: fund f's target weight inside sleeve s (0 if the sleeve does not hold it)

This is the starting mix. Between resets the actual look-through weights drift, at two levels.

**One sleeve's monthly return, then the parent's**

```
R_{s,t} = \sum_f x_{f∣s,t - 1} r_{f,t}, R_t = \sum_s X_{s,t - 1} R_{s,t}
```

where:

- r_f,t: total return of fund f in month t
- x_f|s,t−1: fund f's actual weight inside sleeve s at the end of month t − 1 (its target weight right after the sleeve's own reset)
- R_s,t: return of sleeve s in month t
- X_s,t−1: sleeve s's actual share of the parent at the end of month t − 1 (its target right after the parent's reset)
- R_t: return of the whole portfolio in month t

Each layer's weights drift with its holdings' returns and are set back to target on that layer's rebalance months, at the end of the month after its returns.

**Risk share of a sleeve**

```
RS_s = \frac{W_s (\Sigma W)_s}{W^⊤\Sigma W}
```

where:

- RSₛ: sleeve s's share of the parent's risk; the shares sum to 1
- W: the vector of the parent's target weights for the sleeves
- Σ: covariance matrix of the sleeves' monthly returns over the common window
- (ΣW)ₛ: the s-th entry of Σ times W: sleeve s's covariance with the whole

The same formula as for funds in the risk contribution guide, applied to sleeves. It does not look inside the sleeves; a fund held by two sleeves shows up in both of their streams.

## Worked example

A parent portfolio holding the Golden Butterfly and the All Weather portfolio as two sleeves of 40% each, plus SPY (the S&P 500) as a third sleeve of 20%, with the sleeves reset to those shares every December, from Dec 2004 to Sep 2026, about 22 years. For comparison: the classic 60/40 stocks-and-bonds portfolio over the same months.

The portfolio of portfolios grew 8.1% a year, with volatility of 8.6% and a deepest decline of -20.4%. The 60/40 portfolio, over the same months, grew 8.1% with volatility of 9.4% and a deepest decline of -30.3%: about the same growth, with a smoother ride and a much shallower worst fall. The Sharpe ratio was 0.96 against 0.88.

Risk sits unevenly between the sleeves. SPY was 20% of the money but 29.1% of the risk, because a single stock fund swings more than either of the two mixed portfolios; the Golden Butterfly and All Weather sleeves together carried the rest, less than their combined share of the money.

The two famous portfolios turned out to be close relatives. The highest correlation between any two sleeves, 0.86, was between them: both lean on long Treasuries, gold and US stocks, in different proportions. Counted as co-movement groups, the three sleeves made 1.00 independent bets, so by that measure the parent is one diversified bet held three ways, not three different ones.

That does not make the parent pointless: it changed the mix of growth and decline compared with 60/40. It does show what nesting is for: checking, at the level of whole portfolios, whether a combination adds the diversification its parts promise, and seeing which sleeve carries the risk.

Data through Sep 2026. Reproduce id: 5dcd6722db9fa3de (this guide, its lab setup and the data month).

## How MarketHeist computes it

- **Data**: Monthly total returns (dividends reinvested). Funds younger than the period shown are extended backward with older funds, spliced on returns, not prices.
- **Window**: The worked example and the lab use the common window of the sleeves and the comparison, Dec 2004 to Sep 2026 (262 months), ending at the data month of this page.
- **Nested portfolio**: Each sleeve is the monthly return stream of a whole portfolio (as the MarketHeist MCP reads it, with that portfolio's own rebalancing rule and overlay) or of a fund. The parent holds the sleeves at their target weights and rebalances them at the stated cadence, as the portfolio engine composes a portfolio of portfolios. Risk shares come from the sleeves' covariance; bets count the sleeves' co-movement groups (average correlation at or above the grouping threshold) the way the risk guides do.
- **Sleeves**: A template sleeve is the template's own monthly return stream, simulated with its own target weights and its own rebalancing schedule; a fund sleeve is the fund's monthly total return stream. Nothing inside a sleeve is changed by the parent. (`composeNodeSeries` in packages/portfolio/engine/nodes.ts)
- **Parent**: The parent is simulated with the same function as any portfolio, treating each sleeve's stream as one holding: every sleeve starts at its target share in the first month of the common window, drifts with its own returns, and is reset to target at the end of each rebalance month (December for yearly). No costs or taxes at either layer. (`simulatePortfolio` in packages/portfolio/engine/portfolio.ts)
- **Window**: The common window of all sleeves (and the comparison), ending at the latest data month; a sleeve's window is that of its youngest fund, extended backward with older funds or indexes where available.
- **Risk shares and bets**: Risk shares wᵢ(Σw)ᵢ / w'Σw on the parent's target weights, with the population covariance of the sleeves' monthly returns. Independent bets count co-movement groups of sleeves (average-linkage clusters whose average correlation is at least 0.5), as for funds. (`exposureOverlap` in packages/portfolio/engine/overlap.ts)
- **Nesting depth**: In the app and the MCP, a sleeve can itself hold sleeves, down to a fixed depth, and cycles are refused. (`MAX_SLEEVE_DEPTH` in packages/portfolio/engine/nodes.ts)
- **Lab**: Two to five sleeves, each a template or a fund; the parent rebalances yearly, quarterly or monthly. (`NESTED_MAX_SLEEVES` in content-engine/lab/compute4.ts)
- **Engine functions**: `simulatePortfolio` (packages/portfolio/engine/portfolio.ts), `riskContributions` (packages/portfolio/engine/risk.ts), `exposureOverlap` (packages/portfolio/engine/overlap.ts), `effectiveBetsFromGroups` (packages/portfolio/engine/overlap.ts)

## Portfolio of portfolios (nested portfolios) compared with its neighbours

A portfolio of portfolios compared with other ways to combine holdings

| Structure | What is rebalanced | Risk read at | Typical use |
| --- | --- | --- | --- |
| Flat portfolio | Every fund to its target, at once | Each fund | A single list of funds and weights |
| Portfolio of portfolios (nested) | Funds inside each sleeve on the sleeve's schedule; sleeves on the parent's | Each sleeve, and each fund inside a sleeve | Combining whole portfolios or philosophies |
| Core and satellite | A large core and small satellites, often on different schedules | Core vs satellites | A low-cost core with a few active positions |
| Fund of funds | Each underlying fund by its own manager | The underlying funds, as a whole | Packaged products; extra layer of fees |
| Multi-strategy portfolio | Strategy sleeves by their own rules, beside fixed sleeves | Each sleeve | Combining rules-based strategies with buy-and-hold |

## Limitations

- Risk shares are computed on the sleeves' return streams, so they say how much each sleeve contributes, not which fund inside it is responsible; a fund held by two sleeves is counted inside both.
- Rebalancing at two layers is frictionless in the simulation. In a real account, each sleeve's reset and the parent's reset are separate trades, with their own costs and taxes.
- The common window is set by the youngest fund in any sleeve, so adding a sleeve that holds a young fund shortens the history for the whole portfolio.
- A portfolio of portfolios that each looked good in the past is still a selection made with hindsight; combining them does not remove that.
- Long histories are extended backward with older funds or indexes where a fund is younger than the window, so early years approximate rather than reproduce them.

## Frequently asked questions

**What is a portfolio of portfolios?**

A portfolio whose holdings are whole portfolios, each kept as a sleeve with its own mix and rebalancing, combined at weights the parent sets. It lets two or more complete strategies, such as two lazy portfolios, live side by side in one account and be measured as one portfolio.

**Is a nested portfolio the same as merging the holdings?**

Not quite. Merging gives a flat list of funds that is reset all at once. Nesting keeps each sleeve's own schedule, so the funds inside a sleeve drift against each other until that sleeve resets, and the sleeves drift against each other until the parent resets. The starting look-through weights are the same; the paths differ, usually by a little.

**Does combining the Golden Butterfly and the All Weather portfolio diversify?**

Less than the names suggest. Both hold long Treasuries, gold and US stocks, and their monthly returns have been highly correlated, so in MarketHeist's co-movement count they form one group. Combining them changes the proportions of the same building blocks more than it adds new ones.

**How is risk split between sleeves?**

Each sleeve's share of the parent's risk is its weight times its covariance with the whole portfolio, divided by the portfolio's total, computed on the sleeves' monthly returns. A volatile sleeve, such as a single stock fund beside two balanced portfolios, carries more risk than money.

**Can a sleeve be a strategy instead of a portfolio?**

Yes. In MarketHeist a sleeve can be a rules-based strategy, such as a trend-timing rule or a momentum rotation, held beside buy-and-hold sleeves. The multi-strategy portfolio guide covers that case.

## References

1. Markowitz, H. (1952) Portfolio Selection. Journal of Finance 7(1).
2. Qian, E. (2005) Risk Parity Portfolios: Efficient Portfolios Through True Diversification. PanAgora Asset Management. Risk contributions of portfolio components.
3. Meucci, A. (2009) Managing Diversification. Risk 22(5).
4. Brinson, G. P., Hood, L. R. and Beebower, G. L. (1986) Determinants of Portfolio Performance. Financial Analysts Journal 42(4). The weight of the top-level allocation decision.

## Try it

App: https://marketheist.io/portfolio/new

MCP prompt (MarketHeist MCP server, https://marketheist.io/mcp):

```
Using MarketHeist.io, analyze a portfolio that is 40% the Golden Butterfly, 40% the All Weather portfolio and 20% SPY, rebalanced yearly: CAGR, max drawdown, Sharpe, each holding's share of the risk and the number of independent bets, compared with a 60/40 portfolio over the same months
```

Historical simulation, not advice.
