October 11, 2022. A trader borrows $110 million from a system with no rule against it.
Here’s the moment I keep circling back to: a man sits down, opens two accounts he controls, and starts trading against a decentralized exchange called Mango Markets, using the platform exactly as it was coded. No back door. No stolen password. Just the rules, followed to the letter. Avi Eisenberg didn’t hack anything.
A federal jury looked at what he did and called it fraud in April 2024. Thirteen months later, a federal judge looked at the same facts and threw the whole verdict out.
Same trades, same code, same on-chain record, but two opposite legal answers.
That contradiction is the reason I’m writing this. This isn’t a story about crypto being lawless. It’s a story about what actually decides a “code is law” case: not the slogan, not the size of the loss, but the price-feed record and the trade sequence a court can test (twice, if it has to).
The artifact: a price feed, not a philosophy
Here’s what actually happened, stripped of rhetoric. Mango Markets valued its MNGO perpetual futures using a price oracle that pulled from a small set of outside exchanges, including FTX, AscendEX, and Serum. On October 11, 2022, Eisenberg bought large amounts of thinly-traded MNGO on those venues within minutes, from two accounts he controlled. That pushed the oracle price the protocol relied on sharply higher. Against that inflated paper value, he borrowed roughly $110 million in other crypto assets from the platform’s own liquidity and walked away with it.
He called it a profitable trading strategy. Prosecutors called it manipulation. (Somewhere, a PR consultant is already filing “creative liquidity strategy” away for the next earnings call.) I’d say neither label is the evidence. The evidence is the sequence: which trades moved the oracle, on which venues, in what order, and what that price then let the smart contract do automatically, without anyone at Mango Markets approving the loan by hand.
The record cuts twice
The Department of Justice charged Eisenberg with commodities fraud, commodities manipulation, and wire fraud. A Southern District of New York jury convicted him on all three counts in April 2024.
Case closed, you’d think. It wasn’t.
On May 23, 2025, U.S. District Judge Arun Subramanian vacated every one of those convictions on a Rule 29 motion for judgment of acquittal. Two grounds, both technical. First, venue: Eisenberg traded from Puerto Rico, and prosecutors couldn’t show the trades or the platform had a sufficient New York connection to be tried in that district. Second (and this is the part I find genuinely interesting), on the wire-fraud count, the judge found the government hadn’t shown Eisenberg made an implicit false statement, because Mango Markets, in the court’s words, was “a platform with no rules, instructions, or prohibitions about borrowing.” You cannot lie to a system that promised nothing. (Depending on who you ask, that’s either a legal loophole or the entire marketing pitch of DeFi.)
The Department of Justice appealed to the Second Circuit. As of mid-2026 that appeal is still pending, so this story isn’t over. Prosecutors argue the judge read the law too narrowly, and that “the plain meaning of the word ‘borrow’ itself conveys an intent to repay.” They also argue the platform’s code was “not meaningfully different” from the software banks and brokerages already run, and that treating it differently in court “would unsettle traditional understandings of fraud.”
Notice what both sides are actually fighting over. Not whether Eisenberg drained $110 million (nobody disputes that). They’re fighting over whether the platform’s own rules, or the absence of them, created an obligation he broke. That fight lives entirely inside the technical record.
Why this needs an expert witness, not a lawyer’s argument
I’ve read both sides of the argument, and here’s the trap: a lawyer can read Mango’s user guide and argue “borrow” implies repayment. A lawyer can also read the smart-contract code and argue it imposed no such promise. Both readings are plausible on the page. Neither one, by itself, tells a court what actually happened, in what order, on which venues, in the minutes that mattered.
That’s the narrow technical question a court-appointed software expert exists to answer: which trades moved the oracle price, in what sequence, on what external venues, and what did the smart contract do automatically once that price crossed the threshold that let Eisenberg borrow. The materials are named and finite: exchange trade logs, the oracle’s price-update history, the smart contract’s borrow and liquidation logic, the wallet-level transaction record. The method is repeatable: reconstruct the timeline, show what caused what, and state plainly what the data cannot resolve, like intent. A generalist engineer can describe how a price oracle works in the abstract. A court-appointed expert reconstructs what this oracle actually did, on this day, and survives being asked to defend that reconstruction under cross-examination.
That is the same record that convicted Eisenberg in 2024 and helped acquit him in 2025. The facts didn’t change. What changed was how precisely the technical question got asked and tested.
In a case like this what decides it is a software expert witness written to hold up under cross-examination.
The takeaway
If you’re litigating a DeFi exploit, an algorithmic-trading dispute, or any case where “the code allowed it” is going to be argued in a courtroom, here’s the one thing I’d tell you: don’t let the fight stay at the level of slogans, “code is law” versus “that’s obviously fraud.” Push it down to the record: what data fed the decision, what sequence of events produced the loss, and whether the system’s own published rules created the obligation the other side says was broken. That record is where these cases are actually won.
The above is general information only and does not constitute legal advice. Case facts are drawn from the sources cited.