TLDR
Among 2,894 FedEx shipments in the Epstein corpus, several exhibit cost-to-weight ratios that match Trade-Based Money Laundering (TBML — laundering money through international trade by manipulating the price, quantity, or quality of goods) indicators: a 2-pound "RADIO FENCE" package costing $1,710 to ship ($855 per pound), and JEGE aircraft parts shipments totaling $5,844 (PAPER TRAIL Project, 2026). These anomalies may indicate high-value contents, over-invoicing (deliberately inflating the price of goods on trade documents), or OCR errors in degraded shipping records where parse confidence averages 40-50%.
The TBML Framework
Trade-Based Money Laundering moves illicit value through the international trade system via over-invoicing, under-invoicing, phantom cargo, or misrepresented contents. The Financial Action Task Force identifies anomalous cost-to-weight ratios as a primary indicator: when shipping costs are wildly disproportionate to weight, either declared contents or declared costs are wrong. The Epstein FedEx corpus — 2,894 shipments recovered from DOJ Data Set 10, spanning 2000 to 2005 — contains entries matching this pattern.
Trade-Based Money Laundering is one of the least understood and most underenforced categories of financial crime. Where traditional laundering moves illicit value through banks, TBML moves it through the international trade system — by over-invoicing or under-invoicing goods (inflating or deflating prices on trade documents), by shipping phantom cargo, or by misrepresenting the contents and value of physical shipments. The Financial Action Task Force (the international body that sets anti-money laundering standards) identifies anomalous cost-to-weight ratios as a primary indicator: when shipping costs are wildly disproportionate to the weight of what is being shipped, either the declared contents are wrong or the declared costs are wrong (PAPER TRAIL Project, 2026).
The Epstein FedEx corpus — 2,894 shipments recovered from DOJ Data Set 10, spanning 2000 to 2005 — contains several entries that match this pattern.
$855 Per Pound
The "RADIO FENCE" shipment is the corpus's most extreme anomaly: a 2-pound package costing $1,710 to ship to Bal Harbor Boulevard, yielding $855 per pound versus the standard $10-15 FedEx 2Day rate for the period. Three possible explanations exist: high-value electronic perimeter security equipment with declared-value surcharges, OCR errors in the cost or weight fields, or an over-invoiced amount serving a different financial purpose. A second RADIO FENCE shipment to Punta Gorda recorded no charge — adding inconsistency without resolution.
The most striking anomaly is the RADIO FENCE shipment. A 2-pound package, referenced in the FedEx record as "RADIO FENCE," cost $1,710 to ship to Bal Harbor Boulevard. At $855 per pound, this is among the highest cost-to-weight ratios in the entire corpus. For context, standard FedEx 2Day service for a 2-pound domestic package in this period cost approximately $10-15 (PAPER TRAIL Project, 2026).
There are three possible explanations. First, the package contained genuinely high-value contents — electronic perimeter security equipment ("radio fence" is a commercial term for wireless boundary systems used on properties) — and the $1,710 reflects declared-value surcharges or special handling for expensive electronics. Second, the shipping cost or weight field contains OCR errors (errors from automated text recognition of scanned documents), and the true values are different from what was parsed. Third, the charge represents something other than standard shipping — an over-invoiced amount that served a different financial purpose.
A second RADIO FENCE shipment appears in the records, sent to Punta Gorda, Florida, with no charge recorded. The inconsistency between a $1,710 charge on one shipment and a zero charge on the other adds uncertainty without resolving the question.
The RADIO FENCE shipment remains unverified. No external corroboration of the product, vendor, or pricing has been obtained. It appeared on the EP06 episode's unverified claims list as the only item without independent confirmation (PAPER TRAIL Project, 2026).
JEGE Aircraft Parts
Five FedEx reference fields contain JEGE identifiers — "2 ENGINE JEGE," "WON3286 JEGE," and tail number N908JE (the Boeing 727). Two shipments from "SHIPPING DEPT" carried JEGE work order references totaling $5,844: $5,633 to Valencia, California and $211 to Auburn, Washington. Unlike the RADIO FENCE anomaly, these have a clear operational explanation. Epstein's fleet required maintenance, components are expensive relative to physical weight, and declared values align with the known cost of aviation maintenance logistics.
Five FedEx reference fields contain JEGE identifiers: "2 ENGINE JEGE," "WON3286 JEGE," and tail number N908JE (the Boeing 727). Two shipments from "SHIPPING DEPT" carried JEGE work order references totaling $5,844 ($5,633 to Valencia, California and $211 to Auburn, Washington). These are consistent with aircraft maintenance parts logistics — engine components and avionics parts regularly ship via overnight courier to maintenance, repair, and overhaul facilities (PAPER TRAIL Project, 2026).
The cost-to-weight ratios on aircraft parts shipments are inherently high because aviation components are expensive relative to their physical weight. A turbine blade or avionics module can cost tens of thousands of dollars while weighing only a few pounds. The $5,633 charge for the Valencia shipment, while large in absolute terms, falls within the range of declared-value surcharges for insured aviation components.
Unlike the RADIO FENCE anomaly, the JEGE aircraft parts shipments have a clear operational explanation: Epstein's fleet required maintenance, maintenance requires parts, and parts require shipping. The declared values align with the known cost of aviation maintenance logistics.
The OCR Confidence Problem
FedEx anomaly analysis must account for source-data condition: records were extracted from scanned invoice pages in DOJ Data Set 10 via OCR, with average parse confidence estimated at 40-50% for third-party shipments. At that confidence, "2.0 lbs" might read "20 lbs," and "$1,710" might be "$171.0" or "$17.10" with a misplaced decimal. Script 11 and Script 12 flag entries for human review because automated analysis cannot distinguish a genuine anomaly from an OCR artifact.
Any analysis of FedEx cost-to-weight anomalies must account for the condition of the source data. These records were extracted from scanned invoice pages in DOJ Data Set 10 through OCR processing (automated text recognition from scanned documents). The average parse confidence for third-party FedEx shipments is estimated at 40-50%. At that confidence level, a weight field of "2.0 lbs" might actually read "20 lbs" or "2.0" might be "20.0" — and a cost field of "$1,710" might be "$171.0" or "$17.10" with a misplaced decimal (PAPER TRAIL Project, 2026).
This is a fundamental limitation. The TBML indicators that flag these shipments as anomalous are only as reliable as the underlying data. A cost-to-weight ratio of $855 per pound is extraordinary. A ratio of $85.50 per pound (if the cost is actually $171) is unusual but not remarkable. A ratio of $8.55 per pound (if the weight is actually 200 lbs) is entirely normal.
Script 11 (FedEx parser) and Script 12 (FedEx analysis) flag these entries for human review precisely because automated analysis cannot distinguish between a genuine anomaly and an OCR artifact. The anomaly detection algorithm identifies outliers in the cost-to-weight distribution, but the verification step requires returning to the source PDF and visually inspecting the original invoice.
What TBML Analysis Can and Cannot Tell Us
The FedEx anomalies do not prove Trade-Based Money Laundering — they identify shipments whose characteristics match TBML indicator profiles. TBML detection produces high false-positive rates at institutional scale even with clean data; OCR-degraded data at 40-50% confidence pushes false positives substantially higher. The analysis accomplishes prioritization: among 2,894 shipments, it surfaces the handful warranting source-document verification. The RADIO FENCE shipment remains the corpus's single most anomalous cost-to-weight entry.
The FedEx anomalies do not prove Trade-Based Money Laundering. They identify shipments whose recorded characteristics match TBML indicator profiles. The distinction matters because TBML detection, even at institutional scale with clean data, produces high false-positive rates. With OCR-degraded data at 40-50% confidence, the false-positive rate is substantially higher.
What the analysis does accomplish is prioritization. Among 2,894 shipments, it surfaces the handful that warrant source-document verification and, potentially, external investigation. The RADIO FENCE shipment remains the corpus's single most anomalous cost-to-weight entry. Whether that anomaly reflects an expensive piece of security equipment, a data quality artifact, or something else entirely cannot be determined from the shipping record alone.
References
PAPER TRAIL Project. (2026). FedEx shipment database [Script 11].
PAPER TRAIL Project. (2026). FedEx analysis exports [Script 12, 27 CSVs].
PAPER TRAIL Project. (2026). FedEx third-party analysis [Data set].
PAPER TRAIL Project. (2026). EP06 references [Data set].