The accounts agreed.
The roll-up did not.

Run one household through two systems and you can get two answers. Neither is wrong. Finding exactly where the difference comes from, to the decimal, is not a feat. It is the minimum.

The question that starts it

Take one household, four accounts, seven years, run through two systems. In the illustrative example below, ours says the household had returned 59.7% since inception. The other system says 60.70%. Same accounts, same custody, same period, and a gap of 1.01 percentage points.

Nobody is wrong. Here is the whole of it, because the useful thing is not which number you prefer — it is being able to say exactly where a difference comes from.

First: the accounts agree

Ending market value, net contributions, management fees and income reconciled to the dollar. Every account-level since-inception return matched as well. We display one decimal place; the underlying figures agree beyond it.

So no account is being measured differently. The entire difference lives in one place: how four account returns become a single household number.

Two accepted ways to roll up a household

Asset-weighted composite. Each period, weight every account's return by its beginning-of-period market value, then chain the periods geometrically. This is the standard asset-weighted composite method. It is what we use.

Daily aggregate. Treat the whole household as one portfolio, revalue it daily, and weight every cash flow from the exact day it lands. A true daily time-weighted return at the aggregate level.

Both are standard. Both are defensible. When accounts move together and flows are small they land in the same place, and most of the time you will never notice which one your provider uses.

Where they part company

They diverge when a large flow with an off-market return arrives mid-period. Year by year:

Six of the seven years agree inside a quarter of a point — within display rounding. One year differs by half a point, and because returns compound, that single year carries the whole cumulative gap. Replace that year alone and the two cumulative figures reconcile.

What happens in that year of the illustration

One account was funded with a large deposit near the end of June. The money sat in cash and earned close to nothing for the remainder of the year. That year was a down year, driven by the equity accounts.

So the question is narrow and specific: how much should a flat cash sleeve, arriving in the middle of a losing year, soften the household's loss?

The daily aggregate method lets that cash participate only from the day it arrived, at roughly zero — pulling the household's year toward zero, so slightly less negative.

Asset-weighted composite weights each account's return by its value at the start of each sub-period, which weights that same cash differently and leaves the year slightly more negative.

Neither treatment is wrong. They are two accepted answers to one genuinely ambiguous question, and on a year of that size a half-point spread is a structural property of the methods, surfaced by the timing of a deposit. It is not a defect in either report.

Why we would rather tell you this than not

A provider who cannot explain a difference will usually tell you the other report is wrong. That is the answer you get when nobody has reconciled anything.

The work that matters is the reconciliation: agree the accounts to the dollar first, isolate the roll-up, then find the single period that carries the gap and say what happened in it. In this illustration our method comes out lower. That is not a problem to be managed — it is the point of doing the exercise.

None of this should read as a wow moment. Two systems agreeing on the accounts, and someone able to explain the roll-up, is the expectation for any provider who calculates rather than prints. Treat it as the minimum.

What to ask your own provider

Which roll-up method do you use? Asset-weighted composite, or daily-valued time-weighted at the household. Either answer is fine. No answer is not.

Do the account-level returns reconcile to the custodian? If the accounts do not agree, the household number cannot be argued about yet.

Can you show me the year that carries a difference? Any competent provider can isolate it. The ones who cannot are not calculating — they are reporting whatever the system printed.

Figures are illustrative: fictional, not derived from any account, and shown only to illustrate how two accepted roll-up methods diverge. They do not represent any strategy or blend available on the platform and are not an indication of any actual or expected result.

How composites are built →