Reviews
Field guide to estimate methodology: the five-step pipeline
Estimate methodology behind every published wealth figure: the pipeline, the assumptions and their direction, and the hole where liabilities belong.
Every published net worth figure is the output of a procedure. Someone picked a starting point, applied a chain of assumptions, rounded the answer, and printed it next to a photograph. This page describes that procedure honestly, step by step, and shows where the error enters.
This page is the site's methodology disclosure: it explains the method, which is why no figure follows for a named person.
We publish no net worth figure for any named person, here or anywhere on this site. That is not squeamishness. The quantity does not exist in a form anyone outside the person's own accountant could check, and printing a number you cannot check is a claim about a stranger's finances that you have no basis to make.
What to take away
- The three steps that decide the answer use no evidence at all: tax, growth and debt.
- Better income data does not fix an estimate, because the invented growth rate dominates it.
- Estimates skew high. Debt is usually left out and spending is understated, and on a leveraged estate the error can exceed the assets themselves.
The five-step pipeline
Nearly every figure in circulation is built the same way.
The five-step estimate pipeline
- Collect visible income
- Guess what was kept
- Guess what it became
- Add visible assets
- Subtract debt (skipped)
The five steps in the pipeline are:
- Gross income.What actually arrived, taken from public filings, collectively bargained pay scales, published prize tables or court records.
- Tax.One effective rate applied across a whole career.
- Investment return.An assumed annual return, compounded over the years.
- Spending.An assumed annual outflow, subtracted from what is left.
- Debt.Liabilities subtracted from assets. This step is skipped in almost every published figure.
Steps two, three and five are the entire distance between money that arrived and wealth that remains. None of them uses evidence. They use defaults.
The five steps are easier to see with one hypothetical, using round rates chosen only to make the arithmetic visible. A person earns 10 million dollars gross over ten years. Tax at 45 percent leaves 5.5 million. Representation, staff and travel at 15 percent of gross take another 1.5 million, leaving 4 million.
Spending at 300,000 dollars a year for ten years takes 3 million, leaving 1 million. A 7 percent return over the decade roughly doubles that 1 million, which is still no headline. Now subtract a mortgage, a loan against a future contract and tax still under dispute. The same person can land at or below zero.
A published figure that skips step five reports 1 million dollars and calls it net worth, before any debt.
The assumptions, and why an outsider cannot check them
| Assumption | Why an outsider cannot know it | Effect on the published figure |
|---|---|---|
| Effective tax rate | Depends on residence, entity structure, timing, deductions and losses carried forward | A single rate applied to a whole career hides swings large enough to change the answer |
| Representation and staff costs | Set by private contracts, and they change over a career | Guessed from a headline percentage rather than a signed contract |
| Spending | Unobservable, and the visible spending is the atypical part | Purchases make news and ordinary outflow does not, so the visible part is not the total |
| Investment return | Depends on allocation, timing, borrowing and losses nobody reports | Dominates the result, for reasons below |
| Debt | Mortgages, margin loans, pledged shares, tax liabilities and personal guarantees are rarely public | Rarely subtracted from the assets in the same figure |
| Ownership structure | Assets held through companies, partnerships and trusts may be controlled without being owned | Turns a judgment call into a number with no marker that a judgment was made |
The step that swamps everything else
Compounding is the quiet problem. An assumed return applied across a long career is raised to a power, so a small change in the assumption produces a large change in the output. The starting income figures might come from real documents, and the answer would still be governed by the one input that was invented.
This has a consequence people miss. Making the evidence better does not fix the estimate. You can find the exact contract terms, the exact fee, the exact prize money, and the published figure will still be mostly a function of the growth rate somebody chose. The precision is entirely in the wrong half of the calculation.
Debt is the hole in the middle
Net worth is assets minus liabilities. Estimates are built almost entirely from the first term.
Liabilities that often stay private include mortgages and second charges, loans against a share portfolio, borrowing inside a company the person owns, and deferred or disputed tax. Add personal guarantees to a lender for a business, plus obligations from a settlement.
Any one can be large next to the visible assets. None generates a press release.
A figure that omits liabilities is not net worth. It is gross assets under the wrong name.
To measure household wealth, statisticians survey both sides of the balance sheet and document the method at length. The Federal Reserve's Survey of Consumer Finances is a working example of what that costs and why it cannot be done from outside for one named person.
Why the same person carries different numbers everywhere
Look up one person on several sites and you will get several answers. The reason is not that each site did its own research and reached a different conclusion. It is usually the opposite.
One site posts a number. A second site needs it, takes it, and sometimes adjusts it for a recent event. A third takes the second. After a few rounds the figure looks corroborated, and a search shows four sources agreeing. They are one source and three copies, and the original may never have held a document at all.
The tell is that the differences between sites are usually round adjustments rather than independent results. Independent estimates of a hard quantity scatter. Copies with edits cluster.
Why estimates are round, and why they stick
Two properties give the game away.
Published figures are almost always round. Real balance sheets are not round. Roundness is the fingerprint of a number that was reasoned to rather than counted.
Published figures are also sticky. They tend to sit unchanged for long stretches and then jump, and the jump usually follows a news event rather than any change in the underlying evidence. A quantity that only moves when a headline moves is tracking coverage, not wealth.
What a defensible estimate would have to show
It is possible to write honestly about somebody's finances. It is just much less satisfying than a number. A defensible attempt would publish all of the following, and almost none of what you find does.
- The documents actually held, named individually, with dates, rather than "sources" or "reports".
- Which quantity is being estimatedgross earnings over a period, assets, or assets minus liabilities. These are different and the difference is enormous.
- Every assumption stated as a number the reader can change, including tax, costs, return and time horizon.
- A range rather than a point, wide enough to reflect the assumptions above.
- An explicit statement of what was not obtainable, especially liabilities.
- The date the underlying documents were filed, not the date the article was published.
Apply that list to any net worth article you meet. The result is consistent enough to be a useful habit.
What is actually knowable
Public companies disclose executive compensation in filings, discussed in detail on executive net worth.
Some sports pay is set by collectively bargained scales and published prize tables, covered on athlete net worth.
Court records, including probate in many jurisdictions, can expose financial detail never meant to be public. The judiciary explains what is obtainable in its guide to court records.
What none of those documents give you is net worth. They give you one slice, and the slice is not the whole.
Reading them well means knowing which slice you hold. The contract side is taken apart on contracts and salaries, the sponsorship side on endorsement income, the way lists get assembled on richest rankings, and the difficulty of a lifetime total on career earnings.
Common questions
Isn't an estimate better than nothing?
Only if its error is small enough to support the use you put it to. An estimate whose range spans an order of magnitude cannot support a sentence like "is worth", because that sentence claims a precision the method never had. Nothing is better than a number that is wrong in an unknown direction by an unknown amount.
What about figures attributed to a person's own representatives?
Treat them as positioning. A statement about someone's finances from someone paid by that person is evidence about what they want published. It may be accurate. You have no way to test it, and the incentive runs in a known direction.
Why do these articles exist at all, if the numbers are unreliable?
Because the search demand is enormous and the cost of producing a figure is nearly zero. That is worth knowing before you read the next one.
What should I do with a net worth figure I see quoted?
Ask which of the six items above it publishes. If it names no document, states no assumptions, gives no range and never mentions debt, you are looking at the output of the pipeline at the top of this page, and you now know what that is worth.
In this guide
- A closer look at estimate methodology estimate in 2027Estimate methodology inputs ranked by strength: a working hierarchy of evidence for financial claims, from adversarial court records down to repeated assertion.
- Scoring estimate methodology methodology: what the absence causesEstimate methodology and the missing correction loop: a number nobody can ever prove wrong is not cautious research, it is an unfalsifiable claim.
- Getting highest estimate methodology right the first timeHighest estimate methodology rankings for 2027: why estimators cannot be ranked for accuracy without a truth to check against, and what can be compared instead.
