Mismatched trade detection poses risks for banks

Simon Richards is chief executive USA at Fonetic

While some of us have experienced the speed at which sales are completed on the trade floor, most of us have heard tell of it. So it’s no surprise that mismatched trades is a common challenge for most banks, giving rise to the finance technology industry creating a brand new category – solutions for the management of the P&L trading desk, writes Simon Richards

A good illustration of the problem might be when a trade booked for $10m is marked against the actual trade (completed via telephone or chat) which is booked at $100m. In most cases this is not detected until the reconciliation process, when it is too late – the damage is done and the chance for immediate remedial action has passed. These mismatched trades are racking up millions worth of losses at the P&L desk and have consequences for banks, brokers, financial institutions and independent advisors globally.

The crux of the problem is the time it takes for a mismatch to be identified and analysed. Take the example mentioned above. The trade is made on X date with the broker believing the sale was made for $100m but the trader thinking the buying price was $10m. Three days later (trade date X+3) the clerk in the trader’s Back Office finds the anomaly and calls the broker’s Back Office. Both parties agree there is a mismatch and start the process of collating all of the communications relating to the trade to identify where the mistake was made. Bearing in mind this task can take around a week (trade date X+3+7) both parties then have to analyse all the steps of the trade in an attempt to identify if the mistake was made by the broker or by the trader – this can take another five days. So the whole process of sorting out the mismatched trade can take from day trade date X+3+7+5 = 15 days.

But this isn’t the worst bit! The corrected trade now needs to be booked at the correct trading price and markets don’t stop for mismatched trades. The market has more than likely shifted and devalued the sale resulting in a loss of millions of dollars. This creates a ripple effect beyond the original trade as the bank or firm is liable for the loss which will impact on their trading book.

This isn’t just a P&L desk problem. What the industry needs is a solution to link Front, Middle and Back office systems to enable the P&L of the trading desk to better manage risks and to detect and alert the P&L of mismatched trades in real time. But how can the risk of trades be measured? In order to quantify the risk, all communications must be analysed to decipher trading behaviours and understand the context of each trade. Many solutions already offer this service for written communication but the winning solution will have the ability to monitor and analyse voice calls. To be really first class, it should be able to report the early detection – the benchmark is within 15 minutes – before market movement drives greater losses, enabling swift remedial action and mitigating large losses.

What is required is a linguistics-based trade matching solution that can automate the analysis of trades and match voice and text communications with their corresponding trades and traders whilst using trade data and metadata in context. This requires many elements to come together including social media, data analytics, sentiment analysis and unstructured data to create a single multi-channel, multi-language solution that can decode trader behaviour to make recommendations on managing risk.

The most important element of this is context of voice. The technology must be able to analyse the context of the voice content to establish accurate meanings right from the start of the data gathering process. This requires experience and understanding of the trading floor environment alongside an understanding of trading specific language and slang.

Put simply, the word ‘car’ is often used in the context of trades but is also used in everyday conversations- for trade analytics to work it must have the ability to put this word into the context of the complete conversation. What is needed is a solution that goes beyond who’s talking to whom and how often. A relational picture must be built up over time that allows the analysis of all the types of communications put together as well as the relationship between the people involved. Only then will context anomalies show up.

More than ever the financial community needs a technology partner it can rely on; one that helps improve efficiency when it comes to unnecessary errors, misdirected effort and most importantly reduce money lost through mismatched trades. Dare I say the advent of this new category of financial technology – management of P&L risk – will prompt a shift from so call ‘dodgy’ trading to banking that embraces innovative trading?