How the Belief Gap is computed

This page previously said our published fair value comes from a five-year discounted cash flow. It does not, and our main methodology page has always said so. The published estimate is an earnings-power calculation: normalized earnings times a capped growth assumption, floored at book value per share and held within a band of the current price. The error was here, on this page, and it is corrected rather than quietly deleted because a methodology page that edits itself without saying so is worth less than one that admits a correction.

What the Valuation Lab exposes is a second, separate model. It takes the published fair value, solves for the free cash flow per share that a five-year discounted cash flow would need to arrive at that same number, and then lets you move the growth rate, discount rate and terminal growth from there. It is a way of asking what assumptions the published figure is equivalent to, not the calculation that produced it.

The Belief Gap runs that lab model backwards: it solves numerically (bisection, bounded between 0% and 30% annual growth) for the growth rate at which the lab's value equals the current market price. That rate is the market-implied growth — the growth you must believe for today's price to be fair on those assumptions. If no rate between 0% and 30% reproduces the price, we say so instead of clamping to a fake number.

A 13F reports each position's dollar value and share count per quarter, so value divided by shares is the fund's implied average price for that quarter. For each of the largest reported holders of a stock we walk up to eight quarters of that fund's consecutive filings and find the quarter the position first appeared, or grew by more than 25% quarter over quarter. That quarter's value/shares is the filing-implied entry price. No price history feed is involved: the filing itself carries the cohort's approximate cost basis.

Sanity guards: filings with zero or negative values or shares are dropped, and an implied price outside 0.2x to 5x of the current price is kept but flagged low-confidence (marked with an asterisk) rather than silently discarded. A position held through our whole filing window with no material add is also flagged low-confidence, because the earliest quarter we can see is a floor on the entry, not the entry itself.

Each entry price, and the mean Wall Street analyst target where real coverage exists, goes through the exact same inversion as step 1. The output is one comparable axis, growth-you-must-believe, on which the market price, each super investor's entry, the analyst mean and our model's own funded growth assumption all sit. The belief gap is the distance between what the market demands and what informed buyers historically underwrote, and the verdict sentence under each axis is assembled from those numbers by fixed rules.

Congressional trades are absent on purpose. Those disclosures report only wide dollar ranges, with no share counts and no price band, so no honest entry price can be derived from them. When a cohort's entry price cannot be computed from a filing, it does not appear; nothing on the axis is estimated from a midpoint or a guess.

A filing-implied average price is not a cost basis: the reported value is measured at quarter end, the fund may have bought throughout the quarter, and 13Fs arrive up to 45 days after the period ends. The inversion inherits every weakness of the published fair value it is anchored to, which is documented on the main methodology page, and adds one of its own: the discounted-cash-flow framing it solves in is a restatement of that figure rather than an independent check on it. And a low market-implied growth is not a buy signal: the market is frequently right about why a price is where it is. The Belief Gap converts four incommensurable signals into one comparable axis; it does not tell you which point on the axis is correct.