Methodology

1. The math

V0 is the most recent real, dated valuation mark — a priced round or a secondary/tender sale, never interpolated — at or before your July 1 join date. S0, your ownership share at grant, is your estimated grant divided by V0. Retention is the running product of (1 − dilution) across every primary funding round after you joined, where dilution is capital raised divided by post-money valuation — except for OpenAI's profit participation units (PPUs), which carry a retention of 1, because PPUs track tender and secondary marks rather than round-by-round dilution. Gross paper value is S0 × retention × the current valuation mark — before tax, before vesting, and before you could ever actually sell it.

2. Sources, per company

Every valuation mark below is a real, dated, transacted event — a priced funding round or a secondary/tender sale — never an interpolated or estimated number. This is the receipts: every mark this site uses, where it came from, and how it's labeled.

3. Honesty caveats

• Gross, pre-tax, illiquid paper

  Every number on this site is gross (before any withholding), pre-tax, and illiquid paper — it is not cash you could spend today. The illiquidity caveat applies to the startup side of any comparison, not to an incumbent baseline you might have supplied: startup equity generally cannot be sold on the open market the way public-company compensation can.

• Assumes 100% vested

  The model assumes the grant is fully vested (see vested_assumed: 1.0 in every result). In reality, almost nobody in these cohorts is fully vested: on a typical 4-year vesting schedule, a 2024 cohort is realistically only about 25–50% vested as of today. The number you see is the fully-vested ceiling, not what anyone in that cohort could actually claim right now.

• Three known winners, not the average startup

  This site models OpenAI, Anthropic, and SpaceX specifically — three companies that are, in hindsight, among the most successful private companies in history. It is not the expected value of joining "a startup," and it says nothing about the much larger population of startups, including AI startups, that are worth far less or nothing at all. This is survivorship bias, presented deliberately and by design: the product picks winners on purpose, and you should read every number with that in mind.

• OpenAI figures model PPUs as an approximation

  OpenAI's 2021–2024 grants were profit participation units (PPUs) — capped-profit instruments repriced at tender offers, and later converted to equity as part of OpenAI's corporate restructuring — not round-diluted equity. This methodology treats PPU value as tracking OpenAI's tender/secondary marks directly (retention = 1; no dilution product is ever applied to a PPU). That is a labeled approximation of a genuinely unusual instrument, not an exact accounting of the PPU-to-equity conversion mechanics.

4. Level crosswalk

L4–L7 is Google's ladder. Every other company on this site uses a different title band for a comparable level of seniority; this is the mapping used everywhere on the site, including the picker's "what's my level?" link.

5. Join-date convention

Every cohort is modeled as joining on July 1 of its cohort year. It's a fixed convention, not a guess about you specifically — it keeps every cohort comparable and every golden test locked to a single, documented date. If you use the personalized "your real offer" input, you can supply your own actual join month; the model falls back to July 1 whenever you don't.

6. Changelog

Every edit to the underlying valuation and grant data bumps a version and is logged here, dated.

• 2026-08-312026-08-31Initial MVP matrix.
• 2026-08-312026-08-31Added Anthropic Series A (May 2021, $461M post) and Series B (Apr 2022, $3B post) — the 2021/2022 cohorts are now computable.

Computed from marks as of 2026-08-31.