Most quant track records are overfit, backfilled, or impossible to check. MIZAN proves the few that aren't — without ever seeing the strategy. The engine is live and the registry is minting, external quants included. The destination: every systematic strategy on earth carries a MIZAN proof — the way every bond carries a rating and every company carries an audit. This page is the path from built to standard; the staged, dated version lives on the roadmap.
This was the biggest solvable problem in quantitative finance. The laws of markets cannot be repealed; this deadlock could be — and it is closed. What follows is what gets built on top of a closed problem.
Verification is asset-class-agnostic — the engine doesn't know what it's auditing. The sizing is built up from who pays what, not down from global wealth: the alternatives market, its quant / systematic layer, and the independent quants inside it who need a portable credential.
The unit economics are the sizing: a recurring per-strategy credential fee, multiplied by credentials outstanding, up the $23T → $8T → ~$800B serviceable chain. We don't need 1% of the market — we need to become the standard the best of it runs on. Commercial detail travels with the investor materials.
The technically irreducible part — the part most startups are still promising — is already shipped and operating. Everything in this section is real and independently verifiable today. Everything marked Vision elsewhere on this page is a forward-looking goal, and is labelled as one.
VTR-1 — The Verifiable Track Record Standard, published on SSRN and frozen: 41 numbered requirements, three conformance levels, nine known-answer test vectors, CC BY 4.0 — free for anyone to implement, including competitors. The "standard" destination below already has its founding document.
18,000+ lines of Rust on RISC Zero. A 225 KB STARK, ~81 ms to re-verify, no trusted setup — 17 conditions checked, and the strategy is never revealed.
A public registry of real credentials — external quants included — each verified end-to-end and minted as a portable, tamper-evident credential: sealed by a real STARK, timestamped, publicly re-verifiable. The first was anchored to Bitcoin and an RFC 3161 timestamp; the wall has been growing since.
Every credential is scored against the full Bailey–López de Prado framework — deflated Sharpe, PBO, and combinatorial purged CV — and, since era v11, the Harvey-Liu/Hansen program too. Our own flagship reads "not significant at 95%" on deflated Sharpe — shown on purpose. A credential that can fail on us is the only kind worth trusting.
Credentials that grow: the same sealed strategy re-proven over a longer window as data accrues, every link bound to its parent. The first live chain shipped August 2026: a strategy committed and anchored in Bitcoin (block 962013) before the bars it would be judged on existed — hindsight structurally excluded. Public at /track/.
studio.mizan.market — signup, strategy submission → queued proof → issued credential, and a public per-credential page. Built and tested.
Re-verify a credential entirely client-side, no server — including the STARK seal itself, with no trusted setup. Try it.
Allocations, capital-routing, and fee accounting built and tested in code. It carries no real capital and will not until licensed.
Point-in-time (survivorship-free) equities and ML / black-box paths — engines built; the PIT-equity feed licensed and integrated July 2026. The sealed ML path makes MIZAN verifiable AI for finance: performance proven, model never revealed.
A full internal adversarial security audit run and remediated — self-audit in public. Independent third-party audit still forthcoming.
Nine killed strategies published in full with documented root cause; 40+ retracted overall — the honesty record that makes a PASS mean something. The ledger.
Continuous, fault-tolerant strategy research across asset classes — AI proposes falsifiable research questions; a locked harness confirms or falsifies each.
Multi-venue microstructure recorded continuously since June 2026 — compounds every day.
What moves is a two-sided network. Quants on one side, allocators and platforms on the other, the credential flowing between — and every credential makes the wall worth more to the other side. The cross-side effects are structural and the machinery is built; and it compounds with every credential added. From there it compounds without our sales effort, because a record only grows.
Every page on this estate argues about honesty. This section is about what happens after — the world the standard exists to build. It is a vision, stated as one, and we are building toward it in public, one verifiable step at a time.
You are somewhere no allocator will ever visit — Lagos, Jakarta, a bedroom in Karachi, a desk in Belgrade. You have spent three years finding something real. Today your options are ugly: reveal it to someone with more lawyers than you, or stay invisible forever.
Now run the other tape. You prove it instead. The proof carries everything an allocator needs — honest computation, real costs, out of sample, the full count of what you tried — and none of what they could steal. Your name goes on a public wall next to the verdict. Your track grows forward, anchored, un-backfillable, while you sleep.
And then the direction of the conversation reverses: capital comes looking for you. Not because you networked. Because you are provably real in a world drowning in fakes — and provably real became the scarcest asset in finance the day AI made fakes free.
No pedigree required. No warm intro. No firm name. The wall does not know where you went to school. The first genuinely meritocratic pipe between talent and capital that finance has ever had. Eight things you could never prove →
Every manager who reaches your desk arrives pre-deflated: the search behind their number already counted, cryptographically, before you spend a single diligence week. The flood of AI-generated brilliance that buries your inbox sorts itself — the fakes cannot pass the gate, so they stop applying.
And the upside is bigger than the protection: you see real edge years before your competitors, because the bedroom quant with the genuine signal shows up on the wall long before they show up in a prime broker’s introduction. The allocators who read the wall first get first pick of a talent pool nobody else can see yet.
Diligence that took quarters takes seconds. Trust that took reputation takes mathematics. The edge goes to whoever verifies first.
When verified strategies become allocatable, be first — the allocation list →
The registry is Bitcoin-anchored and append-only — which means the order of arrival is permanent. Founding credential #3 will read “founding credential #3” forever, the way an early block number does.
25 founding seats. The last free ones this company will ever issue.
The engine is live, the standard is published and frozen, 77 credentials sit on the wall including our own refusals, and the first chained tracks are growing in public. The marketplace — capital routed against a credential — is built in code and unit-tested; it carries no real capital yet and will not until licensed. That is the distance between here and the dream, and we publish every step of it. Verify anything on this page yourself — that is the whole point.
Revenue is success-based and market-independent: recurring credential fees, allocator seats, and — once licensed and only then — a platform fee on profits from marketplace-sourced capital. No management fee, no AUM tax, no allocator carry. The marketplace and its fee are contemplated future features; they may require licensing in relevant jurisdictions and will not be offered until authorized.
The next generation of strategies will not be written by people — they will be generated by AI agents, by the million. An agent has no career, no references, no handshake. It has exactly one possible form of credibility: cryptographic proof of what it has done. Every trust mechanism finance runs on today is biological; the economy arriving is not. Verification is not a product in that world — it is the only trust layer that can exist. The credential system for human quants is the passport system for machine capital, and it is already running.
Why the seat is defensible once taken: the registry is append-only and every era verifies forever — so the standard's deepest asset is its age. A larger competitor can rebuild the engine in two years; nobody can rebuild 2026 seniority in any year. The moat is made of calendar, and it compounds daily.
Trust infrastructure is the most durable franchise in finance — neutral, mandated, and impossible to dislodge once it becomes the default. Every era installs a verification standard once the market outgrows handshakes: 1909, the first credit rating. 1923, the first audit standard. 1999, GIPS. Systematic finance is the last large market without one — its standard is being installed now, and MIZAN is built to be it.