Between 2011-2012 it is estimated 9.1 billion euros in trading losses were incurred through operational problems. These losses were not due to bad positions, nor because of fraud, but because during execution and the settlement process something went wrong.
If JPMorgan trades with Goldman Sachs, that trade will go through some kind of external confirmation venue, and both sides will view the others transaction. But when the trades are internal, banks often choose to lower their costs by adopting an internal solution.
These systems are typically excel-dominated in the middle or back office and lacking the infrastructure to detect errors and fraudulent behavior. Worse, internal unmatched trades can go unnoticed for a long time, polluting the downstream systems. The longer a problem is out there, the more expensive it is to correct. The more systems it goes into, the more changes needed to be made.
"This is big," says Neil Vernon, development director at Gresham Computing, a reconciliation and transaction control software provider. "It has nothing to do with fraud, it's simply that banks have complicated infrastructures and things can go wrong in that infrastructure." One of the analogies he likes to use is FedEx. For $10, they will ship a package from NY to London and they will tell you where the package is along the way. "For a trade, I can't get those updates. It's a spaghetti of complex infrastructure. The vast majority of time they pop out the other side with no problems, but a small percentage fail from system 1 to system 2, and we can't get that fixed in real time." From his viewpoint nothing really changed since the study in terms of improving control environments.
In the case of the London Whale, the controls were in spreadsheets, open to manipulation, and the trader took advantage of the T+3 oversight of the unmatched trades.
Need to Automate
The primary response to economic and regulatory pressure in this area is to remove manual middle-office spreadsheet based controls, refereed to as user development applications (UDAs).
"You have to automate as much of your control environment as possible," says Vernon. Solutions like Gresham' Clareti Transaction Control (CTC), a reconciliation software that allow for the same certainty with internal trades as external trades, have seen strong and growing interest from firms looking to replace excel infrastructure.
The software can identify within seconds if a trader booked the other side of a trade, preventing the escalation of transactional loss events.
Since its launch in the beginning of the year CTC has detected problems every day, but not one of them has been due to fraud. All have been attributed to booking errors, minor fat-finger problems -- all innocent trading mistakes that ultimately would have given rise to problems noticed days later.
Risk In Non Standard Products
Some products are highly standardized, and the way JP Morgan processes the messages would work the same as Goldman or Societe Generale. The pure number of transactions lend itself to errors in this area but the real risk, Vernon explains, is in the markets for non-standard products like OTC derivatives.
"In those worlds there is a lack of standards and best practices, and the messages and formats you receive tend to be more complicated and less standard. So we have a high volume of data, and you have to have to deal with large variety of data formats." Legacy UDAs will continue to struggle with these lack of standardizations, limiting a firm's ability to safely enter new markets.