⚠️ All findings are navigational tools derived from automated extraction. They have not been independently verified and should not be treated as established fact. (Unverified) tags apply throughout. Appearance in this analysis does not imply wrongdoing. See COMPLIANCE.md for full professional standards discussion.
After 25+ extraction phases across 1,476,377 DOJ EFTA files (plus 503K cataloged media items across 19 datasets), the publication ledger contains 6,310 unique transactions totaling $2,146,000,000 (Unverified) in documented financial activity across 10 payment types. The auditable subtotal (Tiers 1–3) reaches $1,960,600,000 — 104.4% of the $1.878 billion FinCEN SAR benchmark. The extraction revealed a structured 4-tier trust network with 43 confirmed shell-to-shell wire transfers across 14 shell entities and 8+ banking institutions.
| Metric | SAR Benchmark | Publication Ledger (Unverified) | Delta |
|---|---|---|---|
| Total financial activity | $1,878,000,000 | $2,146,000,000 | +$268,000,000 |
| T1–T3 auditable subtotal | $1,878,000,000 | $1,960,600,000 | +$82,600,000 |
| T1–T3 SAR coverage | 100% | 104.4% | +4.4% |
Why T1–T3 exceeds 100%: The SAR benchmark counts only transactions banks flagged as suspicious. My extraction captures all identifiable financial activity — including non-suspicious legitimate transactions (art auction proceeds, law firm settlements, VC investments, hedge fund returns). The T4 (Unclassified, $185M) tier is excluded from the SAR comparison because those transactions lack sufficient entity resolution for classification.
| Gap Source | Estimable? | Reason |
|---|---|---|
| WEAK/VERY_WEAK tier exclusions | Yes — $5M-$15M | $991M in raw fund_flows_audited entries were excluded as low-confidence. Based on the 93% PROVEN accuracy rate, manual review of the top entries by amount could recover an estimated $5-15M in verifiable activity. This is the only gap I can put credible numbers on. |
| Sealed/withheld documents | No | Court-sealed records are inaccessible to EFTA. Financial activity exists in these documents, but I have zero visibility into the amounts — it could be $5M or $200M. |
| Attempted vs. completed transactions | No | SARs count attempted suspicious transactions; my extraction finds completed wire transfers. The conceptual gap is real, but I cannot estimate its size. |
| Destroyed pre-retention records | No | Bank retention policies may have purged records predating the investigation. The gap is real but unquantifiable from the EFTA corpus alone. |
| Cross-bank SAR duplication | No (but directional) | When the same wire triggers SARs at both the sending and receiving bank, it inflates the benchmark. This means the $1.878B SAR figure itself may overcount — which would reduce the apparent gap. |
Net assessment: I can credibly estimate only one of these gaps ($5-15M from excluded tiers). The others are real information gaps with unknown dollar values. The SAR benchmark itself may contain duplication that overstates the target. I am not going to put specific dollar ranges on things I cannot measure.
See full annotated diagram: NETWORK.md
The extraction revealed a structured hierarchy of trusts and shells operating across four functional tiers:
Tier 1 — Holding Trusts received external investor funds:
Tier 2 — Distribution Trusts redistributed funds internally:
Tier 3 — Operating Shells paid beneficiaries:
Tier 4 — Personal Accounts were terminal destinations:
I traced a single $10 million wire moving through 5 separate entities:
Step 1: Black Family Partners → Southern Trust ($10M — EXTERNAL deposit)
Step 2: Southern Trust → The Haze Trust (DBAGNY) ($10M — Tier 1 → Tier 2)
Step 3: The Haze Trust → Southern Financial ($10M — Tier 2 → Tier 2)
Step 4: Southern Financial → Boothbay ($10M — Tier 2 → Tier 3)
Step 5: Boothbay → Honeycomb Partners ($10M — Tier 3 → Tier 3)
Before Phase 22’s chain-hop filter caught this, my pipeline was counting ONE $10M as $50M. This pattern repeated for $8.5M, $8M, $7M, $5M, and $3M amounts — inflating early totals by $311M.
This layering pattern is consistent with structuring techniques designed to create distance between the original source of funds and their ultimate use.
Jeepers Inc. (DB Brokerage) transferred $51,876,640 (Unverified) to Epstein’s personal NOW/SuperNow account across 21 separate wires. The amounts followed a pattern: 3 wires at $3M each, 8 wires at $2M each, and multiple smaller amounts.
Jeepers Inc. itself was fed by a secondary shell — Jeepers Inc. (non-brokerage entity) transferred $6M to Jeepers Inc. (DB Brokerage) in a single wire. This represents a shell funding a shell funding a personal account.
Plan D LLC — an entity with near-zero inflows in the master ledger ($1,125 from a remittance advice) — disbursed $18,000,000 (Unverified) to Leon Black across 4 wires ($8M, $5M, $3M, $2M).
The near-zero inflow against $18M outflow indicates Plan D LLC’s funding came from sources outside my current extraction scope — possibly earlier wires, inter-trust transfers not captured in the audited tables, or direct bank deposits not visible in the EFTA corpus.
Deutsche Bank appeared in 78 wires across the master ledger, functioning as the primary custodian/intermediary:
| Destination | Amount (Unverified) | Wires |
|---|---|---|
| HAZE_TRUST | $31,287,087 | 8 |
| SOUTHERN_TRUST | $8,896,579 | 2 |
| BUTTERFLY_TRUST | $6,350,000 | 4 |
| INDYKE (estate attorney) | $5,798,525 | 7 |
| JPMORGAN (interbank) | $275,405 | 2 |
The “(DBAGNY)” suffix on entities like “The Haze Trust (DBAGNY)” confirms these were accounts at Deutsche Bank — Deutsche Bank acting as custodian and executing wire transfers on behalf of Epstein entities.
The Haze Trust received substantial funds from art auction houses:
| Source | Amount (Unverified) |
|---|---|
| Sotheby’s → The Haze Trust (Checking) | $11,249,417 |
| Christie’s Inc. → The Haze Trust (Checking) | $7,725,000 |
| Total art proceeds | $18,974,417 |
These represent art sale proceeds being deposited into Epstein trust accounts — the art market functioning as a liquidity conversion mechanism.
Gratitude America MMDA disbursed $6,253,493 (Unverified):
| Destination | Amount (Unverified) | Share |
|---|---|---|
| Morgan Stanley/Citibank (investment) | $5,500,000 | 88% |
| First Bank PR | $300,000 | 5% |
| Medical charities (Melanoma, Cancer Research, Moskowitz) | $425,000 | 7% |
| Other small disbursements | $28,493 | <1% |
88% of Gratitude America’s outflow went to investment/banking accounts. Only 7% went to actual charitable purposes. The entity name projected a charitable mission while primarily functioning as an investment vehicle.
NES LLC disbursed $553,536 (Unverified), of which $538,617 (97%) went directly to Maxwell. Remaining disbursements: $13,940 to Pottery Barn, $979 to Visa Card.
Consistent with NES LLC functioning as a personal expense shell for Maxwell.
| Metric | Value |
|---|---|
| Publication ledger transactions | 6,310 unique |
| Publication ledger total | $2,146,000,000 |
| Payment types classified | 10 |
| T1 (Epstein-Controlled) | $1,610,000,000 |
| T2 (Known Associates) | $343,000,000 |
| T3 (Extended Network) | $7,600,000 |
| T4 (Unclassified) | $185,000,000 |
| Wire transfers in master ledger | 481 |
| Bank statement transactions verified | 1,202 (13 banks) |
| Entries with dates | 350 of 481 wires (72.8%) |
| Entries with court exhibit references | 122 (verified_wires tier) |
| Shell-to-shell transfers identified | 43 |
| Unique entities in master ledger | 228 (entity-resolved) |
| 5-axis PROVEN accuracy (v6.2 spot-check) | 93% (28/30) |
| Contamination bugs caught & fixed | 9 |
| Total inflation removed by bug fixes | $691.8M |
| Total legitimate recovery from bug fixes | $484.1M |
Sealed document access: Financial activity exists in currently sealed court records, but the dollar value is unknown. Unsealing efforts would close the primary information gap.
Plan D LLC funding source: The $18M outflow with near-zero inflow suggests funding mechanisms outside the EFTA corpus. Bank subpoena records may reveal the source.
Jeepers Inc. full chain: The $51.9M pipeline from Jeepers to Epstein personal accounts warrants tracing the upstream funding source of Jeepers Inc. itself.
WEAK/VERY_WEAK tier manual review: $991M in raw fund_flows_audited entries were excluded as low-confidence. Manual review of the top 100 entries by amount could recover $5-15M in verified additional activity.
Cross-bank SAR reconciliation: Comparing entity-level extraction against bank-specific SAR filings could identify which banks reported which transactions and reveal any SAR-level duplication in the benchmark itself.
Analysis current as of February 25, 2026 (Phase 5L). All findings derived from publicly available DOJ EFTA document releases. All amounts are (Unverified) automated extractions. Appearance in this analysis does not imply wrongdoing.