UBS Lost Billions on a Desk Trading Exchange Traded Funds
In 2011 a trader on the bank's ETF desk was found to have concealed unauthorised positions. The bank had already been through the financial crisis and rebuilt its risk framework.
The Desk
Adoboli worked on a Delta One desk, which trades products designed to track an underlying asset closely, including exchange traded funds and swaps. The name refers to a delta of one, meaning the instrument moves essentially in line with what it tracks.
Such desks are meant to run limited directional risk. They facilitate client business and hedge resulting exposure, earning spreads rather than taking market views.
In September 2011 UBS announced losses of roughly 2.3 billion dollars from unauthorised positions on that desk. Adoboli was subsequently convicted of fraud.
The Method
The concealment used fictitious hedges. Real directional positions were entered, and offsetting trades were booked that did not exist, so the reported net exposure appeared within limits.
The specific vulnerability involved trades booked with forward settlement dates that did not require immediate confirmation, creating a window during which the fake side of the book was not verified externally.
This is the same mechanism as Société Générale three years earlier. A control that verifies the net position can be defeated by inventing the other side of it.
Why This Case Is More Troubling
The uncomfortable element is the timing. UBS had been among the institutions worst affected by the financial crisis, had received government support, and had conducted an extensive review of its risk management.
An institution that had recently examined its controls in detail, with every incentive to get it right, still experienced a concealed position of this size three years after an almost identical case at a peer.
That suggests the difficulty is structural rather than a matter of insufficient attention. Any control that relies on reconciling reported positions can be defeated by someone who understands the reconciliation and can influence what enters it.
The Detection
The position was ultimately identified through questions from the control function about specific trades, and Adoboli disclosed the situation in an email to a colleague.
As in other cases, the loss had grown through attempts to trade back to flat. Concealment creates a requirement to recover the hidden loss, which requires more risk, which increases the hidden loss when it fails.
What Controls Actually Work
The measures that address this specific failure mode are narrow and practical. Monitor gross exposure rather than net. Require independent confirmation of every trade including forward settling ones. Treat cancellations and amendments as a monitored risk signal rather than as administrative noise. And enforce genuine consecutive leave.
None of these are sophisticated. They are cheap, and they address the mechanism directly, which is why they appear in every post incident review of this class of event.
The Bottom Line
UBS lost billions to the same mechanism that had hit a competitor three years earlier, despite a recent controls overhaul. Reconciliation based controls fail against someone who can invent the other side of the reconciliation.