Taking on trading desk risk: the lessons of UBS and MF Global

November 22, 2011

Traders work at their desks in front of the DAX indexBy Rachel Wolcott

LONDON/NEW YORK, Nov. 22 (Thomson Reuters Accelus) — When the young UBS trader Kweku Adoboli turned himself in after allegedly having lost $2.3 billion on the Swiss bank’s delta one desk, many asked how such a huge loss could have happened without anyone knowing. The short answer was, in part, that Adoboli’s back-office experience gave him inside knowledge which permitted him to game UBS’ control systems and hide the fraud. The same excuse was trotted out to explain Jérôme Kerviel’s $6.8 billion loss at Société Générale in 2008, but it must surely take more than a stint in the bank office to fool banks’ risk controls systems.

Giorgio Questa, visiting professor in the faculty of finance at Cass Business School in London, said: “Banks have not understood that they will have accidents if they don’t come to terms with risk controls. It’s a question of incompetence. It’s completely clear [in the UBS case] that people weren’t doing their jobs.”

A certain amount of sloppiness in banks’ risk management of trading operations seems to have been a serious contributing factor to catastrophic failures such as the ones at UBS and SocGen. It can often be difficult for risk managers to get a true picture of what traders are doing, because they keep their positions on spreadsheets that are not necessarily connected to the banks’ risk systems. What is more, when traders operate in murky markets like exotic structured products where there is little independent market data available, it can be nearly impossible to value some trades, let alone manage the risk.

Risk management experts have said, however, that the main reason trading desks are notoriously hard to control from a risk point of view is simply that if a trader is doing well, the tendency is for their managers to let them get on with making money, not to rein them in.

Marco Delzio, director of Martingale Risk Italia, a financial engineering consultancy in Rome, told Thomson Reuters: “It’s important to have daily monitoring of traders. My impression is that when the traders are lucky and they think they can make a lot of money, noone complains. Of course at the banks everyone is happy. When the trader makes a loss that’s a criminal action or a fraud, I have the impression that banks might not have a strong incentive to control traders day-by-day.”

What the UBS case and, more recently, the MF Global bankruptcy demonstrate is more attention must be paid to traders’ activities. Risk controls must be reinforced and more stress testing and scenario analysis should be conducted. At the most basic level, however, traders and the trades they make need to be checked. Firms need to educate managers more effectively about how to police their own trading desks. They need to have the proper security setup so a trader cannot confirm his own trades. In addition, firms should improve their operational controls and introduce better-defined independent functions.

Lance Smith, chief executive of Imagine Software in New York, said: “If you are a head of desk you can’t be aware of everything, but you should be evaluating trades to see if they really make sense. You should be asking, ‘where is the real source of profit in this trade?’. You should be asking more questions and getting good answers.”


It may come as a surprise to some that, in an industry apparently so dependent on computer power for split-second algorithmic trading, many traders still rely on the humble spreadsheet for everything from position keeping to running models. Most banks are trying to eliminate the use of spreadsheets on traders’ desktops, but not all have yet succeeded.

Daren Cox, founder and chief executive of Project Brokers, a business intelligence consultancy in London, said: “You can do anything on a spreadsheet. You can change the way calculations are done in cells and it’s just not auditable. From what I see, there are some places left where there still are a lot of spreadsheets in the front office, but almost everywhere I go there is a push to eradicate spreadsheets from the desktops and use the books and records systems. Once you do that it makes tracking risks and positions much easier.”

Spreadsheet use increases the possibility of fraudulent behaviour, and removing them from traders’ desktops should be a high priority for firms looking to shore up operational risk management functions and prevent fraud. Firms performing trading risk management from spreadsheets are leaving themselves open to abuse and human error. The chance of something going badly wrong is high.

Smith said: “If you are trading off a home-grown system it’s hard to enforce rules. Spreadsheets are such a free-for-all. They open up all kinds of possibilities for fraudulent behaviour. Firms need to have systems in place so they can lock down certain functions appropriately. Only certain people should have permissions to do particular functions, while others should be locked out. Firms should have complete control of that.”


One reason some traders cling to their spreadsheets is that the instruments they trade are too complicated for the typical vendor-supplied risk systems. Most of the time, changing books and records systems to accommodate complex transactions is not feasible.

Cox said: “It’s fairly hard to have the standard books and records systems keep track of complex structured products. You have to shoehorn them into systems. You have to break them up and put the components in different asset class systems, then you have to keep track of them. I think that’s why historically banks have used Excel because of its flexibility, but within that flexibility is also its downfall.”

It is not only a case of a round hole and a square peg when it comes to using risk management technology to monitor structured products. A lot of these trades are so esoteric that it can be difficult for risk managers to understand them, let alone perform stress and scenario testing on them. Furthermore, it can be hard to prove to the regulators that they are under control.

Delzio said: “For firms that are trading structured products and exotic derivatives it’s difficult for the risk management department to understand what the actual trade is. These trades are very complicated and maybe there are three or four people in the bank that can really understand what is going on.”


Another reason risk managers have trouble getting a full picture of what is happening on the trading floor is a lack of independent data, especially for exotic transactions. There is a certain amount of information about these kinds of instruments that is only available on the trading desk. There really is not a lot of publicly available data which can be used by risk managers for modelling, stress testing and scenario analysis.

Delzio said: “In 2008 banks started hiring high-level, well-paid quants in the risk management function, because they wanted to improve [it]. So this small group became a bit bigger, but the problem remains the same. Even if there were one risk manager for each trader it would remain difficult to check what the trader was doing, because of the lack of independent data.”

The lack of data for certain trades means that risk managers rely on traders to provide information. It can be relatively easy for a less scrupulous trader to manipulate data and hide losses if necessary.


Firms need to be more vigilant in controlling traders’ books and, through scenario analysis, to understand some of the potential outcomes of the products they are trading. Most banks already set position limits for traders and have policies that set out daily value-at-risk (VaR) limits that are not to be breached. VaR is a common measure for daily risk at banks, but it is not infallible. VaR calculations are based on pricing functions and mathematical models that do not necessarily represent reality.

Delzio told Thomson Reuters: “It’s called model risk. Sometimes quants cannot model the behaviour of the underlying variable. In this sense it is difficult to have a reliable VaR. That is why we suggest banks perform all the scenario analysis. With scenario analysis the most important part is the extreme risks and it’s important to shift the underlying variables, but not by only a few basis points; by 100-200 basis points.”

Moreover, if a firm is still using spreadsheets to manage risk on its trading desks, conducting meaningful stress testing and scenario analysis is impossible. Contemporary vendor-supplied systems allow risk managers to go in and do stress tests on trades. Encouragingly, more firms are moving to these kinds of systems.


No matter how sophisticated the risk management system used, experts have emphasised that most would be unlikely to be of any use in a situation where a trader was determined to commit fraud. It is the responsibility of the desk heads to keep a close eye on what their traders are doing and to look out for unusual behaviour.

What can really undermine risk management efforts, however, are the traders themselves. Critics of banks’ proprietary trading desks point to the culture on the trading desk, where risk-taking is rewarded and personal downside is limited. Most of the time, the worst thing that happens to a trader who racks up losses is that the individual is fired and goes to another firm.

One asset manager told Thomson Reuters: “The balance of risk and reward is wrong on banks’ proprietary trading desks. For me, proprietary implies the word proprietor, meaning the business owner: the person who put capital into the business. The people who are taking these massive risks are in their 20s and they’ve got no equity in the business. The risk/reward is all skewed wrong. If they get it right, they get a massive bonus. If they get it wrong, nothing happens.”


((This article was produced by the Compliance Complete service of Thomson Reuters Accelus. (http://accelus.thomsonreuters.com/solutions/regulatory-intelligence/compliance-complete/) provides a single source for regulatory news, analysis, rules and developments, with global coverage of more than 230 regulators and exchanges))

No comments so far

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/