According to the Justice Department, the conviction stems from the sale of a non-functional automated cryptocurrency trading system marketed to investors between June 2017 and August 2018. The case matters to anyone operating in the algo-trading space: a federal court has now treated a broken execution engine as a prosecutable artifact of fraud.
The product that never traded
Prosecutors presented evidence that Block Bits Capital's flagship tool, internally named Autotrader, was pitched to investors as a complete and operational proprietary algorithm. Internal state: non-functional. Execution path: absent. Dillman allegedly knew both. Investor capital, pooled under the expectation of automated execution, was redirected to personal payments and to speculative positions in unrelated crypto ventures. Those positions lost heavily; reporting back to investors claimed significant profits. The trial lasted 10 days before U.S. District Judge Richard Seeborg.
Key parameters from the docket:
- Defendant: Japheth Dillman, age 48
- Entity: Block Bits Capital
- Charges: wire fraud; conspiracy to commit wire fraud
- Window of misconduct: June 2017 – August 2018
- Tool marketed: Autotrader (proprietary automated trading software)
- Tool reality: non-functional as represented
- Investigating bodies: FBI; IRS Criminal Investigation; SEC San Francisco Regional Office
Why this matters to algo operators
The verdict isolates a specific failure mode: marketing a trading system without an execution layer beneath it. In a functioning stack, the separation between signal generation, order routing, and risk controls is enforced by code review, latency benchmarks, and audit trails. None of those artifacts existed here. The conviction treats the absence of a working system — not merely poor performance — as fraud. For fund operators, the standard going forward is binary: either the strategy runs against live or simulated order books with verifiable fills, or it does not run. Declarations of profitability absent that underlying execution record now carry direct legal exposure.
Sentencing is scheduled for December 8 at 9:30 a.m. before Judge Seeborg. Statutory maximum: up to 20 years in federal prison and a $250,000 fine per count. Dillman remains free on bond pending sentencing.
What to verify in any fund offering
- Execution evidence: on-chain fills, exchange API logs, or broker statements matching claimed positions.
- Backtest auditability: code repositories, fixed datasets, out-of-sample periods. If neither exists, the Sharpe ratio is decorative.
- Capital segregation: wallet addresses and custody arrangements traceable from deposit to deployment.
- Personnel disclosures: identification of co-conspirators and prior regulatory actions against principals.
The case will not move the market. It does, however, reset the minimum due-diligence threshold for any automated trading product that solicits external capital.