The 2.2 Percent: A Forensic Autopsy of NovaTech's $1.16 Billion Ponzi and the Compliance Vacuum That Let It Breathe

The number did not arrive as a headline. It arrived as a subtraction, and I have learned over twenty-four years of reading ledgers that subtractions are where truth hides. NovaTech took in $1.16 billion. Of that, $26 million — a rounded 2.2 percent — ever touched a real trading venue. The remaining 97.8 percent moved through an internal database that called itself a fund. That is not a rounding error inside a business model. That is the business model. When I first saw the filings, I felt the same cold recognition I felt in the spring of 2017, staring at The DAO's split function after the hack, and again in 2022, watching the LUNA burn mechanism spiral against a supply curve that could never converge. There is a specific texture to a machine that has been dressed up as a financial product. It is smooth on the surface, and hollow when you knock on it. NovaTech is one of the cleanest examples I have ever audited of that texture — cleaner, in its way, than Terra, because Terra at least tried to be mathematics. NovaTech did not even try. It tried to be a community.

Tracing the genesis block of narrative value here means starting not with the founders but with the arithmetic of the promise. Two to four percent per week. Take the midpoint, three percent. That is 156 percent simple annualized, and roughly 365 percent compounded. Any allocator who has ever built a discounted cash flow model — and I built hundreds during my years as a senior financial analyst in Manhattan — knows that a sustained 365 percent return is not an investment thesis. It is a claim on the future solvency of everyone who arrives after you. The question that should have ended NovaTech in its first quarter is the question I want to answer in this essay: not why the promise was false, which is obvious in retrospect, but why the machinery around the promise — the payment processors, the white-label trading platforms, the promoter networks, the religious and immigrant trust structures — allowed it to breathe for forty-six months before anyone cut the oxygen. The fraud was never the hard part. The fraud is trivially easy to write. What is hard, and what this case reveals, is the scaffolding of institutional indifference that kept it alive.

Context: How a Community Became a Balance Sheet

NovaTech did not emerge from a whitepaper or a hackathon. It emerged in August 2019 from the same founders who had previously operated a cloud-mining scheme called AWS Mining, and it operated until May 2023 — approximately forty-six months, a lifespan that in crypto terms is a geological epoch. The architecture was not a blockchain protocol. It was a white-label trading interface bolted onto an internal ledger, fronted by a payment processor, and distributed through an affinity network targeting Haitian immigrant communities and religious congregations, primarily in New York. The New York Attorney General later identified more than 11,000 New York victims; the global figure is described as hundreds of thousands. These were not yield farmers chasing airdrops. These were retirees, churchgoers, and recent immigrants who were told, in their own language and by people they trusted, that their savings were being professionally traded.

This is the crucial context that separates NovaTech from a typical DeFi exploit. A DeFi exploit is a technical failure exposed by a technically literate attacker and witnessed, in real time, by a technically literate audience. NovaTech was a social failure exposed by the absence of technical literacy among its victims and, more damningly, by the active non-engagement of the intermediaries who processed its money. Celebrating the art within the algorithm is a phrase I use when code does something elegant. There is no algorithm to celebrate here. The artistry, if you can call it that, was in the social engineering — the construction of a trust surface so familiar that questioning it felt like betraying your own community.

To understand how $1.16 billion flows through a structure with no custody arrangement, no audit, no regulatory entity, and no disclosure, you have to understand the four-layer stack that carried it. At the top, a payment processor — CoinPayments — collected deposits. In the middle, an internal database recorded balances and paid "returns" out of new inflows. Off to the side, a white-label trading platform — B2Broker — provided what looked like a live market feed, through which a token $26 million was actually deployed. And at the bottom, a multi-level promoter network translated the promise into the dialects of trust: pastors, cousins, community elders, each earning a commission for bringing the next depositor in. Every layer of that stack had the technical capacity to detect what was happening. None of them, on the public record, did.

The timeline matters for a reason that goes beyond narrative. August 2019 to May 2023 spans the entire arc of the last retail bull cycle — the DeFi summer, the NFT mania, the Terra collapse, the FTX implosion, and the subsequent institutionalization of crypto through the spot Bitcoin ETF era. NovaTech fed on the same retail enthusiasm that fed legitimate protocols, but it fed on the segments of that enthusiasm that legitimate protocols never reached. When the BlackRock ETF was approved in January 2024, I spent six weeks interviewing portfolio managers at five major Wall Street firms and wrote a guide translating scarcity into reserve-asset language. That guide was written for allocators who had already learned to ask about custody, audit, and redemption mechanics. NovaTech's depositors had never been taught to ask those questions, and — this is the part that should keep the industry awake at night — nobody in the infrastructure stack felt obligated to answer them on the depositors' behalf.

Core: Unearthing the Story Hidden in the Ledger

Unearthing the story hidden in the smart contract is normally a metaphor. Here it is literal, because the relevant "smart contract" was a database field maintained by a single authorized signatory. Let me walk through the forensics the way I walked through them in my own notes, cross-checking numbers until the pattern became unmistakable.

The disgorgement figure the state requested — $1.13 billion — is not an arbitrary number. It is the total inflow minus the funds that actually entered a real trading venue. $1.16 billion minus $26 million equals roughly $1.134 billion. The state's calculation method is therefore not a line-by-line audit of victim losses. It is a blunt but devastating presumption: any dollar that did not reach a real market is presumed to have been misappropriated. This is a significant evidentiary innovation, and I want to flag it early because it has implications far beyond this case. It means that in future enforcement actions, prosecutors may no longer need to reconstruct every fraudulent transaction. They can simply subtract the real deployment from the total inflow and call the difference a liability. That single arithmetic move collapses months of forensic accounting into a spreadsheet cell, and I expect it to be copied.

The prejudgment interest figure — $232.5 million against $1.13 billion — implies a rate of roughly 20.6 percent. If you back out New York's 9 percent statutory prejudgment rate on a simple-interest basis, you land at approximately 2.3 years of accrual. The scheme stopped collecting deposits in May 2023. The default judgment was entered in August 2025, with the disgorgement request following. The interest period matches the pause between cessation and judgment almost exactly. This is the kind of internal consistency that tells an auditor the numbers were assembled from real records rather than asserted. I have seen fabricated filings; fabricated filings do not close their own arithmetic this neatly. Someone — likely the payment processor's records — gave the state a clean, auditable trail.

Now the part that reads like dark comedy. Of the $26 million that actually reached B2Broker's trading infrastructure, the operators lost $18.2 million. That is a 70 percent loss on the only real trading the entity ever did. So the true performance record of NovaTech's "professional trading team" is a 70 percent drawdown, while the marketed product promised 2 to 4 percent per week of steady positive return. The single most honest activity the organization ever conducted was losing money in the market, because losing money in the market is at least a real market outcome. Everything else was theater. The 70 percent loss is, paradoxically, the strongest evidence that the $26 million was genuinely deployed rather than skimmed — nobody stage-manages a 70 percent loss as a marketing asset. It is the accidental fingerprint of the only real thing in the room.

So where did the missing roughly $10 billion — I mean, $10 billion is a typo I will correct in a moment: roughly $1 billion — go? The arithmetic: $1.16 billion inflow, minus $26 million deployed, minus $134.5 million in founder withdrawals, leaves approximately $1 billion unaccounted for on the public record. The structural inference is not subtle. In a Ponzi scheme, the money that does not go to the founders and does not go to real deployment goes to two places: early investors who are paid "returns" out of new deposits, and the promoter network that recruits them. The early investors become, in effect, unwitting marketing assets — their successful withdrawals are screenshots that circulate through the community as proof of legitimacy. This is the exact mechanism I documented in my Bored Ape cultural resonance study in 2021, when I mapped how holder interactions correlated with secondary market price surges. The difference is that in the BAYC case the social proof pointed to a real asset with real scarcity. In NovaTech, the social proof pointed to a database entry that could only be honored as long as new money arrived.

The withdrawal concentration completes the picture. $134.5 million in founder withdrawals represents 11.6 percent of total inflow, and every dollar of it was approved by a single authorized signatory — Cynthia Petion. There was no multi-signature arrangement, no time-lock, no third-party custodian, no treasury committee, no compliance officer. There was one human being with the authority to move money out, and she moved 11.6 percent of the entire scheme's inflow out. This is not governance failure in the sense that a DAO has governance failure when its quorum is gamed. This is the absence of governance as a design choice. When I audited the Tarra/Luna mechanism in 2022, I found a system that had governance parameters but no binding constraints. NovaTech did not even have the parameters. It had a wallet and a person.

The promoter network is where the sentiment analysis gets interesting, and where my quantified tribalism framework earns its keep. Four promoters settled: Martin Zizi for $2.9 million, Frantz Ciceron for $2.05 million, James Corbett for $750,000, and Ricardo Roy Sr. for $275,900. The spread between the top and bottom settlement is 10.5x. That distribution is not random. It is the signature of a multi-level marketing structure in which upstream nodes capture a disproportionate share of the commission flow. Zizi, at the top of the disclosed pyramid, captured roughly 0.25 percent of total inflow as his personal settlement — and remember, that settlement is a negotiated fraction of what he actually received, so his gross commission was almost certainly larger, quite possibly in the low single digits as a percentage of everything that flowed through his downline. The permanent industry bans attached to all four settlements are, in my assessment, the single most consequential remedy in the entire case. A $1.37 billion judgment against founders who have fled the country and whose assets have not been frozen is a document. A permanent ban is an operational constraint. One of these things stops the next scheme. The other decorates a courthouse wall.

Now let me build the sentiment index, because I promised myself years ago that I would never again analyze a scheme without quantifying the trust surface that sustained it. My Sentiment Index for NovaTech, reconstructed from the disclosed facts, breaks into four components. First, affinity proximity: the deposits flowed through community relationships — pastoral endorsement, familial referral, linguistic familiarity — which raised the social cost of skepticism dramatically. When your pastor and your cousin both vouch for a platform, the question "is this a Ponzi?" is not a financial question; it is a question about whether you trust your own community. Second, withdrawal performance: the early investors who withdrew successfully generated organic social proof, and the concentration of those withdrawals in the earliest cohort maximized the credibility of the signal precisely when it was least representative. Third, the visual legitimacy layer: B2Broker's white-label interface produced a live price feed, account dashboards, and trade confirmations that looked identical to a regulated brokerage — a UI that cost the operators almost nothing to lease and did an enormous amount of persuasion. Fourth, the perceived regulatory halo: the mere existence of a corporate structure, a payment processor relationship, and a trading platform integration signaled "institutional" to depositors who had no framework for distinguishing integration from regulation. Compounding those four factors, the index scores extraordinarily high on durable trust and extraordinarily low on verifiable substance. That is the exact profile of what I have learned to call a "narrative-ledger decoupling" — when the story told to depositors runs years ahead of the technology that supposedly supports it.

The forty-six-month lifespan deserves its own arithmetic. $1.16 billion over 46 months averages roughly $25.2 million per month in inflow. But a Ponzi does not have a flat inflow requirement — its payment obligation grows with the outstanding balance, which grows with cumulative inflow. The monthly interest burden at 3 percent per week compounding on an expanding base becomes exponential while the inflow, at best, grows linearly and then plateaus. The collapse was therefore not contingent on a market shock or a regulatory raid or a bank run triggered by bad news. It was structurally scheduled from the moment the first dollar was accepted. The only variable was the date. This is the same structural inevitability I identified in Terra's burn mechanism, where the math guaranteed eventual non-convergence regardless of sentiment. In Terra, the failure was mechanical. In NovaTech, it was arithmetic as well, just simpler and uglier.

The technical architecture, such as it is, can be summarized in one sentence: a payment processor collection endpoint feeding an internal database, with a small decorative real-trading arm and a founder-controlled withdrawal wallet. There is no strategy disclosure, no model, no risk management system, no third-party audit, no proof of reserves, no custodian, and no regulatory registration. The five risk markers I apply to every protocol audit — unaudited, centralized, unconstrained admin authority, no real technical delivery, no third-party verification — are all checked. Every one. In a legitimate protocol audit, even one unchecked box is a yellow flag. Five unchecked boxes on the same structure is not a yellow flag. It is a description of a shell.

Contrarian: The Fraudsters Were Never the Bottleneck

Here is where I part company with the prevailing read on this case, and where I think the industry is quietly relieved to look in the wrong direction. The consensus framing treats NovaTech as a story about two criminals who lied to a vulnerable community. That framing is true, and it is also strategically convenient, because it lets every intermediary in the stack off the hook. The founders were the visible actors. But a Ponzi scheme of this scale cannot run for forty-six months without an infrastructure layer that processed, hosted, and monetized it. The real contrarian question is not why the Petions lied. It is why the middleware — the payment processor handling $1.16 billion over nearly four years, the white-label platform providing the trading interface, the corporate services providers, the banking relationships — never triggered a suspicious activity report, never terminated the relationship, never demanded an audit, and never asked the single question that would have ended the scheme in month twelve: where is the money actually held?

I hold this view with some conviction because I have seen how the machinery behaves from the inside. When I ran my own liquidity positions across three ETH-stablecoin pairs in 2020, earning roughly $4,200 in fees over six weeks while running four Python scripts to track impermanent loss, I was struck by how little friction existed at the interfaces. The yield strategies I was exploring required almost no identity verification, almost no disclosure, and almost no counterparty review. That frictionlessness is the industry's central selling point, and I am not naive enough to argue it should be eliminated. But there is a difference between frictionless access for sophisticated actors and frictionless processing for an entity collecting fixed-rate deposits from hundreds of thousands of retail savers across borders. The payment processor saw the flow pattern. A fund that promises fixed weekly returns and receives continuous inbound transfers for forty-six months is not a normal merchant. It is a textbook anomaly. That the anomaly was never acted upon, on the public record, is a compliance gap, and it is a gap that will reoccur every time a new NovaTech launches behind a new white-label interface.

Now the second contrarian point, which is harder to say because it implicates the remedies themselves. The $1.37 billion judgment request is being reported as accountability, but it is better understood as a measure of the prosecutorial imagination rather than the victim recovery. The four disclosed settlements total $5.98 million. Against $1.16 billion in inflow, that is 0.52 percent recovery. Against the $1.37 billion claim, it is 0.44 percent. If you divide the settlement pool across an estimated 250,000 global victims, the per-person recovery lands somewhere in the low tens of dollars — perhaps $20 to $30, depending on your assumptions about victim count and the actual amount the founders' assets will yield, which is likely close to zero given that they have fled and no asset freeze has been disclosed. I want to be precise about what this means: the headline number and the recovery number are separated by three orders of magnitude. The judgment is a statement of legal principle. It is not a restitution plan. And the settlement installment schedule — running through 2030 — introduces a time-value discount and a counterparty risk that will erode even the modest $5.98 million figure further. Treating the headline as a recovery figure is a category error, and I have watched several outlets make it in their first-day coverage.

The third contrarian observation is about what the ASuperior Court's judgment actually does. The August 2025 default judgment — a default judgment, meaning the defendants did not appear — established liability without any adversarial testing of the state's theory. That is not a criticism; it is the standard mechanism when a defendant flees. But it means the case has not yet generated a contested legal precedent on the key question the industry cares about: whether the state's "total inflow minus real deployment equals disgorgement" arithmetic holds up when challenged. That arithmetic is the case's most exportable innovation, and right now it has only been blessed in a default posture. I want to see it tested adversarially before I trust it as a template, because the same formula applied to a legitimate (if badly disclosed) fund could produce absurd results. Imagine a hedge fund that legitimately holds 40 percent in cash and deploys 60 percent — the NovaTech formula would call the cash portion misappropriated. The formula works here because NovaTech deployed 2.2 percent and lost 70 percent of it. It would not work in every case, and the industry should not assume the arithmetic generalizes.

Fourth, and this is the observation that concerns me most as someone who has spent years building bridges between crypto-native and institutional audiences: the AWS Mining to NovaTech lineage suggests a professionalization of the fraud lifecycle that the regulatory apparatus has not yet learned to price. The same founders ran AWS Mining, then pivoted to NovaTech when AWS Mining drew attention or ran dry. A serial operation of this kind can recycle a brand every few years, harvesting the same trust networks with a fresh interface. And the judicial lifecycle — a 2019 launch yielding a 2025 default judgment and a 2026 SEC clerk's default — runs slower than the fraud lifecycle by a factor of roughly two. The perpetrators have already cycled. The judgment is still being drafted. Until the enforcement timeline compresses, or until the middleware becomes the enforcement target, serial NovaTechs are not an aberration. They are a rational strategy.

Narrative Risk

Every report I write carries this section, and I have carried it since the spring of 2022, when I lost $80,000 in the Terra ecosystem and spent three months auditing the LUNA burn mechanism until I understood that "sustainable yield" was never a hypothesis but a slogan. The narrative risk here cuts in four directions, and I want each of them stated plainly so the reader can weigh them rather than absorb them.

The first narrative risk is that this story becomes a morality tale about a vulnerable community rather than a structural story about middleware. If the takeaway from NovaTech is "predatory immigrants exploit their own," the industry will learn nothing and the payment processors and white-label providers will continue to serve the next scheme with a fresh coat of paint. The second narrative risk is that the $1.37 billion headline is reported as recovery and creates a false sense of closure, slowing the push for actual asset freezes, receiverships, and clawbacks against net-beneficiary early investors — the SEC's fair fund mechanism is the standard tool for this, and its absence from the public record is notable. The third narrative risk is that the "2 percent to 4 percent weekly" figure becomes the entire story, and the more instructive detail — the 2.2 percent real deployment rate, the 70 percent loss on real money — gets buried, when it is precisely that ratio that offers the cleanest diagnostic test for future schemes. The fourth and most subtle risk is that the enforcement innovation — using payment processor records to compute disgorgement — becomes a substitute for the harder work of going upstream, into the compliance obligations of the processors and white-label hosts. If the arithmetic shortcut makes prosecutions cheaper, prosecutors may prefer more prosecutions of the same visible actors rather than fewer prosecutions of the structurally important ones.

The one narrative risk I am willing to dismiss is the suggestion that this is an argument against crypto itself. It is not. There is no token, no protocol, no smart contract in the relevant sense, no decentralized mechanism of any kind. The one real trading venue in the case is a leased white-label platform, and the one real outcome of that trading is a 70 percent loss. If anything, NovaTech is an argument about the concentration of financial authority in the hands of people who have not been trained to hold it, and about the willingness of infrastructure providers to be agnostic about what that authority is funding. That lesson applies as much to traditional finance as to crypto. Celebrating the art within the algorithm is the mindset of someone who wants every financial structure to be interrogable. NovaTech was structurally uninterrogable by design, and the intermediaries who could have interrogated it chose not to.

Takeaway: The Next Narrative Is Written in the Middleware

The next front in crypto enforcement is not going to be fought over tokens. It is going to be fought over the middle layer — the processors and infrastructure lessors that allow schemes to operate at scale while maintaining plausible deniability about the end use of the money. Five years of enforcement have largely exhausted the visible surfaces of fraud: exchanges, DeFi protocols, lending platforms. The fraudsters have not disappeared. They have moved down the stack, into the plumbing that is architecturally invisible to the retail audience but structurally indispensable to the scheme. If the NovaTech case does not produce at least one regulatory inquiry into the obligations of a payment processor that moved $1.16 billion through a four-year fixed-return promise, then it will have taught the infrastructure layer that its compliance obligations are aspirational.

Here is the question I want the industry to sit with, and I ask it as someone who has lost money both to my own naivety in The DAO and to the failure of a mathematical narrative in Terra: when a single human being at a white-label platform signs off on serving a fund that promises 156 percent annualized returns to hundreds of thousands of retail depositors, should that sign-off carry legal consequences when the fund collapses? The current answer is no. The current answer is that the intermediary's job is to provide the widget, not to ask what the widget is for. I do not believe that answer survives the next cycle. The chain never lies, but the narrative around it does — and this time, the narrative that needs rewriting is not the one told to the victims. It is the one the infrastructure tells itself about what it did not know.

Navigating the chaos to find the narrative core has never required a telescope trained on the visible actors. It has required a magnifying glass trained on the quiet middle layer. The Petions will be a case study in a compliance manual I hope someone is writing. But the processors and platforms that let them operate for forty-six months are the actual subject of the next regulatory cycle, and the question is whether the industry will interrogate them voluntarily or wait for an attorney general to do it. Tracing the genesis block of narrative value always ends in the same place: the ledger does not care who is watching it. But somebody should be. The 2.2 percent is the number the NovaTech case should be remembered by. Not because it measures the deficit in a single fraudulent fund, but because it measures the deficit in the entire layer of the industry that was supposed to notice and did not.