Rebalancing when half your portfolio cannot be sold

A portfolio in which half the assets cannot be traded is not rebalanced by selling what has risen and buying what has fallen. It is steered with the money that flows through it. Set wide target ranges instead of fixed weights, measure them including the commitments you have not yet paid, and use three levers in this order: the size of new commitments, the destination of distributions, and only then the liquid holdings. Selling an illiquid position to restore a percentage is almost never the right move.

In January the year-end depot statements are in and the fourth-quarter fund reports are still two months away. That gap is the first problem.

Why the textbook method does not work here

Textbook rebalancing assumes that every position has a price today and can be traded today in any size. A fund stake has neither. Its value is the general partner's estimate as of a quarter-end, reported six to ten weeks later. A direct holding in a start-up is carried at the last round, which may be three years old. A property has an appraisal or your own guess.

So the weights you compute are a mix of today's prices on the liquid side and old estimates on the illiquid side. When share prices fall sharply, the private share of the portfolio jumps although nothing happened in the private holdings. This is the denominator effect, and April 2025 showed how misleading it is: the jump had reversed before the quarterly reports could confirm or deny it, as we described at the time.

The second difference is that you cannot choose the trade size. A fund calls capital when it wants and distributes when it can. An angel holding is sold when the company is sold. The only illiquid position whose size you decide freely is the next one.

Ranges instead of targets

A target of "30 per cent private" is breached on the day it is set. A range is a decision about when to act and when to leave things alone.

Say a portfolio of EUR 3,000,000 looks like this: EUR 1,350,000 liquid (ETFs, bonds, cash), EUR 750,000 fund stakes at NAV, EUR 300,000 direct holdings in start-ups and EUR 600,000 equity in two rental flats. The owner has set ranges:

After the fall the portfolio is worth EUR 2,730,000. The liquid share is a touch below its range, the private share still inside. A fixed-weight rule would now demand selling EUR 95,000 of private assets. That is impossible, and a rule that cannot be followed teaches its owner to ignore it. The range says: nothing needs to be sold, and attention is needed on the liquid side.

How wide should the ranges be? Wide enough that an ordinary bad year in listed shares does not breach them. Fifteen percentage points for the large blocks is not sloppy; it reflects how imprecisely the illiquid values are known.

Count what you have promised, not only what you have paid

The table above leaves something out. This owner has EUR 400,000 of unfunded commitments: capital promised to funds and not yet called. Those calls will be paid from the liquid part and will turn into private assets.

Run that forward on the fallen portfolio. Liquid drops from EUR 1,080,000 to EUR 680,000, private rises from EUR 1,050,000 to EUR 1,450,000. The shares become 24.9 per cent liquid and 53.1 per cent private. Both are far outside their ranges, and no new decision has been made; these are obligations already signed.

Distributions will arrive in the meantime and soften the picture, but they are uncertain in timing and amount and the calls are not. An allocation measured only on paid-in values understates the private share for as long as the investor is still building the programme. We would measure both: the current view, at NAV, and the committed view, with unfunded commitments moved from liquid to private. The range should hold in the first and should not be wildly breached in the second.

The levers, in the order we would use them

New commitments. This is the main lever and the slow one. A commitment signed today is called over three to five years, so it changes the allocation of 2029, not of this year. In the example the decision is simple: no new fund commitment until the committed view is back near 40 per cent. In the opposite case, after a strong run in listed shares has pushed the private share below its range, the answer is not a double commitment this year. A steady annual amount, adjusted by a fifth up or down, keeps the vintages spread and avoids committing most at the top.

Distributions. Every distribution is a rebalancing decision that costs nothing to execute. If the private share is high, distributions go to the liquid side and stay there. If it is low, they fund the next commitment. Writing this rule down in advance prevents the common default, in which distributions sit in a current account until something comes up.

The liquid part. This is where actual trades happen, and it has two jobs that can conflict. It is the adjustment for the whole portfolio, and it is the reserve for calls. The reserve comes first. We set out how to size it in planning liquidity for capital calls. Inside the liquid part, rebalance with new money and with distributions before selling anything, because sales cost tax. For a German private investor, selling EUR 100,000 of an equity ETF bought for EUR 60,000 realises a gain of EUR 40,000; after the 30 per cent partial exemption, EUR 28,000 is taxed at 26.375 per cent including solidarity surcharge, which is EUR 7,385 gone from the portfolio for the sake of a percentage.

Property debt. For real estate the adjustable quantity is the loan, not the building. An extra repayment when a fixed-rate period ends raises property equity and lowers liquidity; a higher loan does the reverse. It is available about once every ten years per property.

Selling an illiquid position comes last. A fund stake can be sold on the secondary market, usually at a discount to NAV and with the manager's consent, as we explained in selling a fund stake. That is a tool for a real liquidity problem or for a change of strategy. As a way of moving an allocation by three points it is expensive.

The mistakes we see most

Acting on a stale ratio. If the private share rose because listed shares fell last week, wait for the marks. If it rose over four quarters of reports, it is real.

Paying calls out of distributions in one's head. After a fall nobody wants to sell securities, so the hope is that distributions will cover the next calls. A call has a due date and a default clause. Plan it from the liquid side.

Treating all private holdings as one block. A fund in its eighth year that is distributing and a fund in its second year that is calling have opposite effects on liquidity.

Forgetting look-through. If the liquid part is a world ETF heavy in US technology and the private part is venture funds, moving money between the blocks changes the label more than the exposure. That was our subject in how much AI do you own.

A one-page policy

The useful output of all this is a single page: the ranges for three or four blocks, the two views in which they are measured, the rule for distributions, the annual commitment amount and the condition under which it is cut or raised, and the size of the reserve for calls. Valued shows the unfunded commitment per fund from the booked calls and distributions, with a forecast of the calls and distributions still to come, which gives the committed view without a separate sheet; the page itself is the investor's own. Write it in a calm month, and read it before acting in a bad one.