INVESTOR DATA ROOM — CONFIDENTIAL

Trust is the missing
layer in modern finance.

Most quant edge is fake. Allocators can't tell real from fabricated without seeing the strategy, and no manager will show it. MIZAN proves the few edges that are real, without revealing a single trade. We hold ourselves to the same bar first: at the honest 95% significance test, our own flagship fails: and we publish that verdict, unchanged. One engine, any asset class. Crypto is what is verified today; the rail is market-agnostic — equities, FX, commodities, macro and rates as managers submit them.

Request access See how it works
STRATEGY · AUDIT CERTIFICATEMZN-TREND-4H-001
Historical data — no future data usedPASS
Survivorship bias — correctedPASS
Out-of-sample held backPASS
OOS Sharpe (walk-forward) · 1.29PASS
Risk-adjusted return · 1.65PASS
Positive years · 7 of 10PASS
VERDICT  VERIFIED
MIZAN
VERIFIED
The Problem

$23 trillion moves on blind trust.

The global hedge fund industry manages $4.5 trillion. Family offices and institutions hold $23 trillion in alternatives. Every allocation decision rests on claims that cannot be independently verified. Independent allocator research now measures the flood: 72% of LLM-finance studies leave key biases unaddressed; nine of ten frontier LLM trading agents show negative stock-selection alpha (SPEC Research, State of AI Alpha, Q3 2026).

GAP 01

Anyone can show good backtest numbers

With AI tools, generating a convincing performance chart takes hours. There is no standard way for an allocator to know if those numbers are real. They rely on introductions and reputation alone.

GAP 02

Sharing the strategy means losing it

The only way to verify your edge today is to reveal your code and data. Most managers refuse, so the strategy stays unverified and the allocator faces a binary: trust or walk.

GAP 03

Due diligence is slow and expensive

Institutional DDQ processes take 6–18 months. This friction keeps 90% of good smaller managers locked out of institutional capital — permanently.

The Trust Gap — Today vs. With MIZAN
What allocators need to know Today With MIZAN
Is this backtest free of hindsight bias?Trust the managerCryptographic certificate
Were out-of-sample periods truly held back?No way to knowMathematically proven
Does live performance match the backtest?Manually cross-checkAutomated audit trail
Has the manager retracted bad strategies?No standard recordPublic retraction log
Can I verify without seeing the strategy?Impossible todayYes — strategy stays private
Why This Didn't Exist

The market wasn't waiting.
It was blocked.

This market wasn't waiting on demand. It was blocked. To prove an edge was real you had to reveal the strategy, and once you revealed it, it wasn't yours. To keep it secret you had to ask for trust. Allocators have lived inside that trade-off for thirty years, and no amount of diligence dissolves it, because the constraint was mathematical rather than commercial.

Zero-knowledge proofs dissolve it. That cryptography only became practical, at this cost and this speed, in the last two years. We didn't invent it. We pointed it at systematic finance: twelve conditions checked independently — sixteen when the credential discloses its exposures and regimes, a 219KB receipt, seconds to verify.

The constraint that made this market impossible is gone. The standard is unclaimed.

How It Works

Four steps from claim
to certificate.

Think of it as the SSL certificate of quant finance. Just as a padlock proves a website's identity without revealing its server code, a MIZAN certificate proves strategy integrity without revealing a line of logic.

GIPS verifies the process. MIZAN proves the math. An audit confirms a workflow ran; a proof confirms the numbers are true.

01

Submit your strategy

The manager submits their trading strategy and historical data to the MIZAN verification engine. Nothing leaves the secure environment. The strategy is never exposed to anyone, not even MIZAN.

02

16 integrity checks run automatically

MIZAN re-derives the result from point-in-time data and cryptographically enforces no look-ahead, a held-back out-of-sample split, walk-forward consistency, the full cost-and-funding model, and a declared gate policy no human can override. Both canonical schools of backtest honesty now run in-circuit and are live — López de Prado's suite plus the Harvey–Liu / Hansen program (Probabilistic Sharpe, SPA, minimum backtest length) on the v11 engine: the Deflated Sharpe Ratio for selection bias, the Probability of Backtest Overfitting (PBO) for overfitting, and Combinatorial Purged Cross-Validation (CPCV) for leakage, each minted as a real STARK credential and re-verifiable at mizan.market/verify. No strategy, including our own flagship, has yet cleared all three bars.

03

A mathematical proof is generated

MIZAN generates a 219 KB cryptographic proof, a mathematical fingerprint that proves in minutes and is independently verified by anyone in seconds. It confirms every check passed without revealing how the strategy works.

04

A verification certificate is issued

The manager receives a MIZAN Verification Certificate — shareable with any allocator, independently verifiable in seconds. No trust required. The math speaks for itself.

VERIFICATION CERTIFICATE MZN-TREND-4H-001
Strategy typeMomentum — BTC 4H Trend
Data window2020 → 2026 (12,917 4h bars)
Out-of-sample periodlast 30% · walk-forward
Annual return (net costs)+26.6%
Maximum drawdown · intrabar−20.76%
OOS Sharpe (last 30%)1.29 ✓
Integrity checks16 / 16 PASSED
Proof 219 KB · Proves in minutes · Verifies in seconds VERIFIED
Any allocator can verify this certificate independently in seconds
The Standard

MIZAN is to quant finance
what SSL is to the web.

Before SSL, websites could claim anything. After SSL, a padlock meant something — enforced by math, not by trust. MIZAN does the same for investment strategies.

SSL proves identity without revealing server code. MIZAN proves integrity without revealing a single trade or parameter.

SOC2 certifies operational processes for software. MIZAN certifies research processes for quant managers, the same institutional trust layer, applied to alpha.

Moody's rates bond creditworthiness through independent analysis. MIZAN rates strategy credibility through mathematical verification, not opinion.

The first certificate is ours, and it fails the 95% significance bar, on purpose. MIZAN judged its own flagship by the Deflated Sharpe Ratio (Bailey–López de Prado), got "not significant at 95%," and published that verdict unchanged. It has since refused our own AI as well: a neural network whose inference we prove in-circuit, which made +895% on NVIDIA and still failed. Published in full, re-verifiable by anyone, the model never revealed. A credential its own author cannot force to pass is the only kind worth trusting.

Verification Certificate
ACTIVE
Issued toMIZAN · ATLAS System
Strategy IDMZN-TREND-4H-001
Mint date2026-07-22
Checks passed16 / 16
Proof size219 KB
Verification timeseconds · any laptop
Net annual return+26.6%
Sharpe · after costs1.63
PROOF HASH
0x9f3a2b8c4d1e7f6a5b0c3d9e2f1a4b7c
8e5f2a9b3c6d0e4f7a1b5c8d2e6f9a3b
219 KB
PROOF SIZE
seconds
VERIFY TIME
16 / 16
CHECKS
Live Strategy Performance

The first verified strategy.
Nine years. After all costs.

Before selling verification to others, we verified ourselves. The ATLAS BTC 4H Trend strategy is our flagship credential — reproducible from a real cryptographic proof, walk-forward tested, on a held-back out-of-sample window. Every number below is after transaction costs, slippage, and funding. It is a backtest, not a forecast.

The honest bar

Judged at the 95% significance bar, the Deflated Sharpe Ratio of Bailey and López de Prado, which discounts a Sharpe for how many strategies were tried — this flagship reads "not significant at 95%." We show that verdict on our own credential, unchanged. The numbers below are real, after-cost, and out-of-sample; we simply refuse to overstate their statistical significance. A credential that cannot fail is not a proof, which is exactly why the ones we do issue mean something.

Annual Return
+26.6%
OOS Sharpe
1.63
Max Drawdown
−20.76%
Risk-Adj. Return
1.65
OOS Sharpe
1.29

Binance 4H 2017–2026 · 5bp taker + 2bp slippage + real funding · IS 2017–2022 / OOS 2023–2026

MetricMIZAN StrategyBuy & Hold BTC
Annual return+26.6%+35.7%
Max drawdown−20.76%−83.9%
Risk-adj. return1.650.79
Worst year−6%−74%
Positive years7 / 108 / 10
Time in market31%100%
Verification statusCertified ✓Uncertified
Year-by-Year (After Costs)
2017
+14%
2018
−6%
2019
+53%
2020
+71%
2021
+73%
2022
−2%
2023
+47%
2024
+61%
2025
+26%
2026
−1%
Crash Resilience
2018 crash
BTC Falls 84%
Bitcoin−84%
MIZAN−6%
2020 crisis
COVID Crash
Bitcoin−60%
MIZAN+71%
2022 bear
Terra / FTX
Bitcoin−65%
MIZAN−2%
2025 volatility
Macro Pressure
Bitcoin−35%
MIZAN+26%
Revenue Model

MIZAN earns from verification —
and shares the upside when alpha wins.

Zero management fee. Zero AUM tax. A flat credential to get verified, and a success fee only on capital the marketplace sources, paid only when the strategy wins. On that capital: a 20% performance fee, split manager 15% · MIZAN 5% · allocator keeps 80%.

STREAM 01

Marketplace Success Fee

On capital MIZAN's marketplace sources, a 20% performance fee on profits — split so the manager keeps the majority: manager keeps 15%, MIZAN takes 5%, allocator keeps 80%. Crystallized annually, above a high-water mark — no fee on recovered ground. MIZAN earns for sourcing the capital, the rails and the proof — only when the strategy wins. Scales with capital flow, not per-strategy negotiation.

5%
MIZAN TAKE · MANAGER KEEPS 15% · ALLOCATOR 80%
STREAM 02 · ON-RAMP

Verification Credential

Managers pay a flat $6,000/year per active credential — initial mint plus quarterly chain-extensions. Institutional tier at $25,000/year adds the full 16-condition disclosing report, priority Free forever, by design: running the verifier costs nothing, and registering trials in the honest-N ledger costs nothing — charging for either would corrupt the statistics the standard exists to protect. Founding cohort: the first 25 managers mint their first credential free. proving and named support. If a manager raises capital themselves, this is all they pay: MIZAN never charges a fee on capital it didn't help source. You pay for the examination, not the outcome: pass or fail, the price is the same, and re-anchoring after a MIZAN verifier upgrade is free.

$6K/yr
FLAT · SELF-RAISERS PAY ONLY THIS · INSTITUTIONAL $25K
STREAM 03 · INFRASTRUCTURE

Allocator Seat + Bespoke Verification

Allocators pay $25,000/year per seat: credential dashboard, watchlists, extension & break alerts, API, and $25,000 per strategy for bespoke verification: the allocator's own gate conditions, a full re-run, and a written institutional report. Browsing the marketplace and running the public verifier stay free. Cash-positive infrastructure layer that funds operations even while the marketplace take ramps.

$25K/yr
SEAT · BESPOKE REVIEW $25K/STRATEGY
Founding allocator program · 3 seats
The first three institutions: bespoke verification waived, gate conditions tuned to their desk, founder-direct line. When the seats are taken, list price — permanently. Allocators who help tip a standard shouldn't pay like the ones who wait for it.

The marketplace and its success fee are contemplated future features; they may require licensing in relevant jurisdictions and will not be offered until authorized. Mizan Verification, Inc. (a Delaware corporation).

Revenue Projections
AUM
Capital routed
×
15%
Avg net return
×
5%
MIZAN take
+
$6K/yr + $25K/yr
Credentials + seats
=
$18M
Year 3 ARR
Year Credentials Capital Routed Profits Generated Marketplace Fee (5%) Total ARR
Y1 10 pilotslaunch $5Mfirst integrations $750K15% net $37.5K+ credentials $400K
Y2 350 issuedtraction $250Mnetwork forming $37.5M15% net $1.9M+ credentials $5M
Y3 1,000 livenetwork standard $1BFoF + institutions $150M15% net $7.5M+ credentials $18M
The Marketplace

The product the revenue model will run on.

The engine is asset-class agnostic and verifies any market. Verified today: the directional trend family — more markets as managers submit them. The marketplace engine — allocations, capital-routing, and fee accounting — is implemented and unit-tested in code; no capital has flowed through it yet, and that is the next milestone. Allocators self-direct capital into their own brokerage accounts; custody never leaves them, and MIZAN never custodies, executes, or advises. This is where the marketplace success fee is designed to be earned.

studio.mizan.market/allocate — product preview
MIZAN allocator interface — asset-class-agnostic verification, funded through the allocator's own brokerage account

Illustrative product preview, the marketplace engine is built in code and unit-tested; no real capital has flowed through it yet. Live today: the verification engine, the self-serve product, chained tracks (credentials that grow as data accrues), and the first issued credential.

Market Opportunity

A missing standard
in a $23 trillion market.

Every major institutional trust layer was eventually standardized — credit ratings, audit standards, cybersecurity certifications. None existed until someone built the first one. MIZAN is a cryptographic verification credential for quant finance — built to become the standard.

NEAR-TERM

$500B+ quant hedge funds

10,000+ systematic funds globally. Every one needs credibility with allocators. Target: 5% penetration = $250M TAM. Reachable within 3 years at current growth trajectory.

MID-TERM

$4.5T hedge fund industry

All hedge funds face due diligence friction. MIZAN is designed to become the standard DDQ supplement: increasingly expected for allocator relationships at institutional level.

LONG-TERM

$23T alternatives market

Family offices, pensions, endowments. All move on trust today. MIZAN is designed to become the trust infrastructure: the Moody's of machine-generated alpha.

Why Now

AI made fake edge free to produce

Anyone can generate a convincing 10-year backtest in hours using AI tools. The supply of unverifiable performance claims has exploded. Allocators cannot distinguish real edge from sophisticated simulation.

  • AI strategy generation available to anyone with an API key
  • Backtest fabrication is undetectable without third-party verification
  • Number of "quant funds" growing faster than institutional DD capacity

Verification technology just matured

The mathematical tools powering MIZAN — cryptographic proofs that verify computation without revealing inputs — have only become fast enough for finance in the last 2 years.

  • Proof size: 219 KB for a complete 9-year backtest audit
  • Verification speed: seconds: instant for any allocator, anywhere
  • Strategy privacy: complete: zero information revealed about the logic

Today, credible verification means a manual audit at roughly $50,000 per strategy per year, a cost that prices out emerging managers. At the same time, SEC scrutiny of performance claims is rising, and examiners increasingly reward numbers that have been independently verified. The need is real; the timing is now.

The $6,000 credential is the wedge. The business is the rail: a take-rate on capital moving through a $23 trillion market that today runs on self-graded homework. Rails that become standards are not valued on revenue multiples — Moody's, MSCI and ICE are the comp set.

Competitive Moat

Three layers of
defensibility.

The hardest thing to copy is not the technology — it's the reputation for honesty. MIZAN's moat compounds with every certificate issued, every retraction logged, and every allocator who relies on the standard.

Cryptographic Moat

Verification uses mathematical proofs that are computationally infeasible to fake. No one can issue a MIZAN certificate without passing all 16 checks. The math is the gatekeeper, not a human reviewer who can be pressured or paid.

Reputation Moat

We retracted our own strategies publicly when they failed. That public retraction log is our strongest asset. Every honest retraction increases the value of every remaining certificate. No incumbent can copy this without reversing decades of marketing.

Network Standard Moat

Once 50 allocators require a MIZAN certificate, every manager needs one. Once 500 managers are certified, every allocator needs access. The standard becomes self-reinforcing — exactly like SSL or credit ratings before it.

Timestamping isn't verification

Tools like validityBase hash a track record onto a blockchain — proof you committed to numbers, not that they're honest. MIZAN runs the backtest's integrity checks (no lookahead, walk-forward, point-in-time data) inside a zero-knowledge circuit, and proves the numbers were generated honestly, the methodology itself, not just the timestamp. Commitment versus computation. They stop where we start.

No two-sided cold-start — we're our own first supply

Most marketplaces die bootstrapping supply and demand at once. MIZAN doesn't. We don't need external quants to launch — MIZAN Labs' own gate-cleared strategies are the initial supply, seeded through the same locked gates and subject to the same public retractions. Allocators see real, credentialed supply from day one. External managers are the growth lever, not the launch dependency — no two-sided bootstrap to solve.

Self-supply doesn't compromise the credential: the gate is objective, every proof is independently reproducible, and MIZAN publishes its own failures (40+ retracted). Our own strategies clearing a bar we publicly hold ourselves to is evidence the standard is real, not a conflict.

Competitive Landscape
Hedge fund DDQ
Status quo

6–18 months, hundreds of hours, no standardization. Completely manual, completely subjective. Cannot verify computation — only reviews documents and meets management. MIZAN: seconds, cryptographic, permanent.

Third-party auditors
Traditional

Audit firms review books and processes but cannot mathematically verify that a backtest is free of look-ahead bias. Human opinion, not cryptographic proof. Expensive, slow, relationship-dependent.

Performance databases
eVestment, Preqin

Collect self-reported numbers with no verification. Any manager can submit any numbers. They aggregate claims — MIZAN verifies them. That distinction is the entire business.

MIZAN
New standard

Mathematical proof, not human opinion. Instant verification, not months of review. Strategy stays private. Public retraction log. Early in a category that cannot exist without the technology we've built and are already running.

Verifies Alpha ↑ · Generates Alpha ↓
CONFLICTED — RUNS OWN CAPITAL
NEUTRAL — NO STAKE
Embedded verification
Darwinex · Numerai · Fund Auditors
Verification tied to their own fund or platform. Not a neutral standard the market can trust.
THE OPEN QUADRANT
Neutral verification layer
MIZAN · ATLAS
Verifies any strategy on locked, human-override-proof gates. No fund is structurally capable of occupying this position. MIZAN is building the standard from it.
Generates alpha · Has stake
Funds · Pods · Strategy Shops
Sell or run their own alpha. Structurally conflicted as verifiers — the moment a fund also verifies, its verdicts are worthless.
The status quo
Self-reported backtests
The default today — allocators handed PDFs and PnL screenshots. No verification. This behaviour, not any company, is what MIZAN displaces.
Y-axis: Generates vs Verifies alpha X-axis: Has stake vs Neutral That emptiness in the top-right is the opportunity
Why neutrality is structural

The moment a verifier also runs capital, its verdicts are conflicted. A fund cannot verify neutrally without ceasing to be a fund. Neutrality is a position only an infrastructure layer can hold permanently.

  • The distinction is structural, not ethical: alpha sellers and trust sellers serve different masters.
  • Every strategy shop is a potential customer: they need independent verification precisely because they cannot provide it themselves.
  • Not outclaimable: MIZAN does not compete with quants. It certifies them.
Why the standard wins

Strategies decay and get copied. A trusted verification standard compounds toward winner-take-most. Once allocators accept the MIZAN certificate, every quant who wants capital seeks it.

  • The standard, not a strategy: we verify everyone's; no one can copy the referee.
  • Honesty as brand: 40+ strategies tested and retracted (9 published in full with root causes), including a live-fill bug we caught and disclosed the same day.
  • Standards compound: the credential embedded with the first fifty allocators becomes the default.
Verification Roadmap

The lead is a set of
closed boundaries.

The engine already proves a backtest ran clean: point-in-time data, no lookahead, costs and funding applied, out-of-sample, walk-forward, a declared gate policy. That closes dishonesty within a trial. Everything below closes a boundary around it. Each is disclosed today as an open boundary at mizan.market/standard: a system that named zero open boundaries would be lying. Closing them one at a time is the moat: a competitor with more capital still has to close each one itself.

TRIAL LEDGER
LIVE · forward-only
CLOSES · MULTIPLE TESTING

A forward-only, Merkle-committed pre-registration log: every trial's specification is committed before it runs. The number of attempts behind a reported t-stat becomes cryptographic instead of living in the researcher's head, which makes the deflated Sharpe ratio enforceable rather than self-reported. This is the exact gap the field's leading researchers still describe as unsolved.

● LIVE
N read in-circuit since guest v6
SIGNED PROVENANCE
~2–3 weeks
CLOSES · DATA FABRICATION

Today the proof binds the computation to a committed dataset; the dataset's authenticity anchors to a canonical commitment. Binding that commitment to a signed source — exchange-signed data or a data-authority attestation — moves fabrication from an open boundary that is disclosed to one that is blocked. This is the deepest trust root in the system.

○ PLANNED
LIVE CHAINING
~3–4 weeks
CLOSES · BACKTEST-TO-LIVE GAP

The credential proves the backtest, not that live trading matched it. Chaining an attested live-execution signature to the credentialed strategy turns a research credential into a track-record credential, the object allocators actually allocate against. Labeled clearly as not-yet-shipped until it is.

○ PLANNED
INDEPENDENT AUDIT
external
CLOSES · TRUST IN THE ENGINE ITSELF

The internal security review is published in full. A name-brand third-party audit of the circuits turns "trust the math" into "and unaffiliated experts verified the math", the credibility capstone an allocator's counsel expects. No new capability; pure assurance.

○ FORTHCOMING

Sequencing is customer-led: the Trial Ledger ships first because it carries the most leverage and is the smallest build; the order of the rest is set by which boundary a design partner needs closed first. At the end of this list, the credential has no closeable honesty gap left — only the philosophical residue no cryptography can touch, which we will keep disclosing, because that discipline is the brand.

Public Retraction Log

We retract bad strategies
publicly.

Every time we found a flaw in our own research, we documented it publicly and stopped trading immediately. No incumbent will ever do this. It is our most defensible moat, and it cannot be purchased.

1H BB SQUEEZE STRATEGY
Live-fill testing confirmed a hidden fill bug in the backtest. 8 test trades, win rate 12.5% vs 75% claimed — the gap exposed entries beyond the stop price being booked as instant wins. The strategy was halted and publicly retracted the same day the bug was found.
Retracted 2026-06-13 · Caught in live-fill testing
FILL ASSUMPTION BUG
Published Sharpe 16.09 → Honest: negative
BTC+ALT OVERLAY FAMILY (v1–v19)
Strategy used data from the end of each month to decide that month's trade — equivalent to knowing the future. Hidden leverage of 6.9× caused 7 liquidation events invisible at monthly sampling. Published "+7,177% CAGR" was mathematically impossible. Honest replay: −86%/year.
Retracted 2026-06-12
LOOKAHEAD + HIDDEN LIQUIDATION
US EQUITY MOMENTUM FAMILY (v41–v46)
Point-in-time membership test revealed survivorship bias: the engine held today's 322 survivors back to 1997, buying future winners before their inclusion dates. Removing the bias: CAGR drops from 27% to 7.7%, Calmar from 1.26 to 0.40. The entire selection layer was invalid.
Retracted 2026-06-12
SURVIVORSHIP BIAS
INDIA GOLD ENGINE v5
Monthly-close metrics showed −4.6% max drawdown. Daily mark-to-market at 3× leverage revealed a −51% intra-month drawdown over 5 days. The monthly sampling masked catastrophic intraday risk. Strategy halted the same day the analysis was completed.
Retracted 2026-06-13 · True MDD: −51% vs published −4.6%
HIDDEN LEVERAGE MDD
Every retraction makes the remaining certificates more valuable — this is our compounding trust moat
See the full Retraction Ledger →
Technology

Three systems. All running now.

Not a roadmap item — live production infrastructure operating 24/7. Built before we raised.

Sample verification output  ·  run the live proof at studio.mizan.market  ·  16 / 16 cryptographic gates pass
The strategy is hidden.
The proof is not.
Independently re-verify a real 219 KB STARK credential against the canonical price data — without trusting MIZAN, the issuer, or the data provider. The panel below is a captured sample of the engine's output — run the live proof at studio.mizan.market.
STRATEGY
BTC-BOLLINGER
CREDENTIAL SIZE
219 KB
CHECKS
12 cryptographic
VERIFY TIME
seconds
STRATEGY REVEALED
NEVER
OPEN LIVE STUDIO  ↗
LIVE Want to mint and verify other strategies yourself? The full MIZAN engine is live at studio.mizan.market — pick a strategy, mint a credential, re-verify it independently. Real cryptography, ~300 ms per call.
or run locally: ./verifier credential.bundle prices.csv
LIVE 24/7 · ATLAS — Autonomous Research Engine

The machine that never stops looking.

A fault-tolerant autonomous engine that runs continuous research cycles. Every hypothesis must first become a falsifiable research question: a stated mechanism, a named loser, a pre-declared kill-criterion — before it is ever tested. The locked harness then stamps each result confirmed or falsified, remembers what it kills so it is never re-tested, and adversarially re-audits every survivor (parameter-perturbation, cost-stress, independent skeptics) before a human ever sees it. No human approves or overrides. The engine polices itself.

Data refresh
→
Universe scan
multi-asset
→
Research question
mechanism + kill-test
→
Locked harness
confirm / falsify
→
6 gates
all must pass
→
Vetted candidate
audited → human
WHAT RUNS AUTONOMOUSLY
▸
Continuous data refresh: real-time market data across the full strategy universe
▸
Nightly research cycle: dozens of strategy variants tested and ranked automatically
▸
Automated promotion: only strategies clearing all hard limits advance
▸
Live health monitoring: watchdog alerts on any failure, silent when green
LOCKED RESEARCH GATES — ALL SIX MUST PASS
OOS SHARPE
≥ 2.0
IS SHARPE
≥ 1.5
MAX DD
≤ 25%
TRADES (n)
≥ 300
OOS HALF A
Positive
OOS HALF B
Positive
Gates are hard-coded. The AI proposes strategy specs as constrained parameters — it never writes executable code and cannot alter its own gates.
HELD TO THE INSTITUTIONAL BAR

Every candidate's Sharpe is deflated for multiple testing with the Deflated Sharpe Ratio (Bailey–López de Prado) — searching thousands of strategies inflates the best one's Sharpe by chance, so we discount for how many were tried. We judge against the honest 95% significance bar, the same bar our own flagship fails, which we disclose rather than hide. The overfitting and leakage tests (PBO and CPCV) now run in-circuit and are live alongside the DSR, and on the v11 engine the second school (Harvey–Liu / Hansen) too, the complete framework, both schools, each minted as a real STARK credential and re-verifiable at mizan.market/verify. Every hypothesis is pre-registered with a stated mechanism and a falsification criterion before it is ever tested. Cryptographic enforcement of the trial count, the Merkle Trial Ledger, with N read from the ledger in-circuit — is live in the DSR credential; its honest boundary is forward-only (N counts every trial from the moment the ledger opened, and experiments run before it existed cannot be recovered). This is why most of our own candidates fail, and why a pass means something.

PRODUCTION · mizan-proof

Cryptographic Verification Engine

Advanced cryptographic proofs applied to financial performance. The strategy is verified without ever being revealed, the same principle used by the most secure systems in the world, now applied to quant finance.

01
Returns on claimed data
Without revealing the strategy
02
No lookahead bias
Verified by harness, not inspection
03
Stats computed correctly
Sharpe, CAGR, MDD, walk-forward
04
Strategy never revealed
Logic, signals, params — sealed
CREDENTIAL SPEC
✓
219 KB cryptographic proof: independently verifiable receipt
✓
16 cryptographic checks (12 base · 16 when disclosing exposures & regimes) — data integrity, gate policy, verdict, strategy concealment, digest
✓
Verifies in seconds: anyone, anywhere, no MIZAN involvement required
RECORDING NOW · Proprietary Microstructure Data

A moat that compounds continuously.

Tick-level order book depth, funding rates, open interest, and basis — recording continuously across 6 major coins. Every day this dataset runs, competitors fall further behind.

BTC
ETH
SOL
BNB
XRP
DOGE
▸
Full order book at tick frequency: 20-level depth + trade prints
▸
Funding, OI, basis across perps: captured continuously since June 2026
✓
Signal exists, not yet deployable: deep-book order imbalance on BTC has a real, sign-stable gross rank-IC, but maker Sharpe is ~0.76 after correcting a cost model bug — below the 1.30 bar. Fill/adverse-selection harness in development.
Competitors who start recording today are already 6+ months behind. The gap compounds every day.
Team

Built by a quant who
polices himself.

The founder found and published every bug in his own research before anyone else could. That discipline is the product, and the reason the verification standard means something.

FOUNDER

Mohammad Muavia

Shipped 18,000+ lines of Rust before raising a dollar, the cryptographic engine works in production today
Built ATLAS: an autonomous research system running 24/7 across 18 markets, finding and discarding strategies that fail honest validation
Discovered and published 4 major bugs in his own strategies, halting the affected strategy each time immediately
Verified MIZAN's own strategy first — the first customer is the founder
Live systematic trading since 2026, with real financial consequences for every research decision
Building the standard from first principles: no certificate issued unless the math passes
ANGEL INVESTOR & ADVISOR

Dr. Anish Mohammed

Co-Founder & CTO, Panther Protocol: ZK cryptographer
CEO & Co-Founder, Prufold Labs: verifiable-AI cryptographic trust layer
Reviewed Ethereum's Orange Paper
20+ yrs cryptography · research incl. Microsoft Research
Ripple advisory board
Advising on the MIZAN verification engine: a cryptographer's lens on the core proof-system architecture
Angel investor in MIZAN
Built Before We Raised
Verification engineLive ✓
First certificate issued2026-06-14 ✓
Proof size219 KB
Verification speedseconds
Public retraction log9 published ✓
Checks per audit16 / 16
Strategy data window9 years
Own strategy verified firstYes ✓
Traction

The pull is inbound.

Engine live. Self-serve product built and tested. First credential issued end-to-end. Credentials chain into growing, re-verifiable tracks. Engaged users from 15+ countries; a dozen ran live prove-and-verify operations, unprompted. Built solo, before a dollar was raised.
An external quant from an algorithmic-trading firm ran the live engine unprompted and proved two strategies.
Senior institutional interest is inbound. The specific names are shared on calls, not on websites.
Execution Plan

12 months to
becoming the standard.

PHASE 01 · Months 1–3

Verified supply

Self-serve portal already built & live · onboard external supply
  • Self-serve credential portal built & live (studio.mizan.market/app) — onboard external managers onto it
  • Onboard first quant managers across markets
  • Issue first credentials · publish public registry
  • First allocator conversations — family offices & systematic shops
TARGET: 10 verified strategies live
PHASE 02 · Months 4–6

The marketplace opens

First capital flowing through MIZAN
  • Allocator dashboard live — fund strategies through your own brokerage account
  • First allocator capital routed through the marketplace
  • Marketplace success fee earning on routed capital
TARGET: first routed capital
PHASE 03 · Months 7–9

Network forming

Marketplace take scaling across markets
  • More verified strategies + allocators onboarded
  • Public credential registry at scale
  • MIZAN credential becomes a due-diligence standard
TARGET: path to $400K ARR
PHASE 04 · Months 10–12

The marketplace is the business

Marketplace success fee = core revenue
  • Capital routing scaling across markets
  • Marketplace take is the core revenue line
  • Positioned for the next raise
TARGET: $400K+ ARR run-rate
The Ask

$1.5M seed.
12 months to standard.

We are raising a $1.5M seed round to build the team, scale the verification engine, and sign the first allocator API contracts. The product works. The certificate exists. We need capital to make it the standard.

Engineering: verification engine scale, API infrastructure, dashboard
$600K40%
Sales & Partnerships: allocator relationships, first 10 API contracts
$375K25%
Research & Auditing: certifying the first 500 external strategies
$300K20%
Operations & Legal: compliance, contracts, infrastructure
$225K15%
ROUND: $1.5M seed  ·  INSTRUMENT: SAFE  ·  RUNWAY: 18 months  ·  TERMS: shared with committed investors

The round is small because the product is already built. This is funding for distribution, not discovery.

WHY INVEST NOW

The product exists. The first certificate was issued before we raised. The retraction log is public. We are not asking you to fund a hypothesis — we are asking you to fund the distribution of a working standard — no neutral verification standard exists yet.

12-MONTH MILESTONES
Month 3
50+
Paying managers
Month 6
5+
Allocator API seats
Month 9
$1M+
ARR run-rate
Month 12
500+
Active certificates
Month 12
10+
Allocator seats
Month 12
$2M+
ARR target
INVESTOR RETURN SCENARIOS · on $1.5M invested
Conservative (Y2–3)
$25M+
Co. value · 10× ARR
Investor return
1.2×
$1.75M on $1.5M
Base (Y4–5)
$270M
Co. value · $18M ARR × 15×
Investor return
12.6×
$18.9M on $1.5M
Bull (Y7–10)
$5B
Co. value · 20× ARR
Investor return
233×
$349M on $1.5M

The standard gets built.
The question is who builds it.

Allocators need a way to trust performance. Managers need a way to prove it. The technology exists. The first certificate has been issued. MIZAN is designed to become the standard, and it is open for partners now.

[email protected]
Schedule a call Connect on LinkedIn Review the product

MIZAN is the verification rail for systematic finance: cryptographic proof that a track record cleared a declared institutional risk gate — after costs, out of sample, free of lookahead — without revealing the strategy. Proofs re-verify in about seconds against a source-available verifier, trusting no one. MIZAN publishes its verification boundaries, its refusals, and its own retractions.

CONFIDENTIAL — FOR QUALIFIED INVESTORS ONLY. This document contains forward-looking statements and projections based on current assumptions that may not prove accurate. Past performance of the MIZAN ATLAS strategy does not guarantee future results. All performance figures are after transaction costs including fees, slippage, and funding rates. The MIZAN Verification Certificate verifies the integrity of a backtest methodology; it does not constitute investment advice, a guarantee of future returns, or regulatory approval of any investment product. MIZAN is a verification and technology provider, not an investment adviser, broker-dealer, or fund. It does not custody, execute, or advise on capital, and does not manage third-party capital. The marketplace and its fee are contemplated future features; they may require licensing in relevant jurisdictions and will not be offered until authorized. Mizan Verification, Inc. (a Delaware corporation). This material is not an offer or solicitation in any jurisdiction. Prospective investors should conduct their own due diligence and consult qualified advisors before making investment decisions. © 2026 MIZAN. All rights reserved.