How To Actually Calculate Net Worth From Public Data
Most people treat net worth as a headline number from Forbes. It never works that way. You have to reconstruct it yourself from fragments, and the process is uglier than any article suggests. I spent three weeks last year building a net worth model for a prominent female finance figure after a podcast episode quoted a single estimate without citing sources. The published number was off by roughly forty percent. Here is the method I used, the data sources that actually matter, and where everything falls apart. Start with SEC filings. If the person holds equity in a publicly traded company, you will find their exact share count in Schedule 13D or 13G filings. These are free on the SEC EDGAR database. I pulled a filing that showed a holdings discrepancy between the annual proxy and the quarterly amendment, and the difference was twelve thousand shares. At the time, that was about four million dollars. The public net worth figures that week had not caught up at all.
Next, cross-reference with company insider transaction forms. Form 4 filings show buys and sells within two business days. The timing matters. If someone sold stock during a lock-up expiration window, the proceeds might be reinvested elsewhere and still counted as liquid in some models. Do not assume every sale is cash-out behavior. For private holdings, you dig through press releases, patent filings, and state-level business registrations. I once tracked a stake in a defunct fintech by following a founder mention in a CB Insights report, then matching that name to a Delaware LLC search result. The entity showed a capital contribution of two hundred thousand dollars in 2018. Valuation at liquidation came back at zero. The model had to reflect that loss instead of the optimistic figure floating around in media coverage. Real estate is the hardest piece. County assessor records vary by jurisdiction. Some counties publish full parcel values. Others only show the land value and not the improvement value. I use a combination of county GIS portals and a paid service called PropStream for aggregated data. The workaround I ended up using for a difficult property in Cook County was calling the assessor directly. The online system had the valuation stale by eighteen months. The phone representative gave me the assessed value from the most recent reevaluation.
Liabilities are where most models fail. Public records will show liens, judgments, and mortgage assignments. But they will not show personal guarantees on SBA loans, margin debt against stock portfolios, or private credit facilities. I encountered a case where a figure appeared to carry eight million in declared assets but had four million in undisclosed margin exposure with a broker. The actual net worth was half the headline number. There is no clean way to find this. You look for red flags: frequent portfolio rehypothecation patterns, margin call news mentions, or sudden stock pledges reported in regulatory filings. Here is the part nobody talks about. Liquidity-adjusted net worth is different from stated net worth. If sixty percent of the estimated portfolio is in restricted stock or private fund commitments with five-year lockups, the usable net worth on any given day is dramatically lower. I built a liquidity waterfall into the model: unrestricted publicly traded holdings at one hundred percent of market value, restricted stock at seventy percent, private fund interests at fifty percent, and real estate at sixty-five percent after accounting for transaction costs and carrying charges. The tools you actually need are minimal. An SEC EDGAR search, a state LLC lookup, county assessor websites, and a spreadsheet with scenario ranges for each asset class. Do not pay for a net worth estimation service. They run the same public data with worse granularity. The whole process for a moderately complex case takes about six to eight hours if you know where to look. The first attempt took me fourteen because I did not realize the trust structure holding the primary stake required parsing through a separate probate record.
Get the Full Details
One more thing. Net worth estimates for living subjects should always carry a confidence range, not a single number. My final model outputs three figures: conservative, base case, and aggressive. The gap between conservative and aggressive for the subject I worked on was twenty-two million dollars. The media had been citing one point estimate for months. A range would have been more honest and just as informative.
