How do Markets desks really perform?
Almost every firm with a trading-desk capability periodically reviews the performance of their sales and trading teams. Typically, this is done using internal 2nd line (risk) and 3rd line (audit) resources at regular intervals following a routine process of checking controls and a range of related processes.
In many cases, underlying performance is not regularly compared with internal and external measures.
We have done some recent work to independently examine the performance of trading and sales teams at firms and arrived at some interesting results. The true source of revenue of certain desks was quite different to that assumed or expected by management, risk and audit.
In some cases, the daily analysis of the sources of revenue was not complete due to limited data sets while in other cases the analysis lacked critical components to highlight inefficient use of balance sheet and/or capital.
In other examples, analysis showed the expected changes to the capital calculations (FRTB and SA-CCR) will also impact the performance of Markets businesses. The efficient use of resources such as capital and balance sheet will be imperative for future profitability of desks and businesses.
Careful, independent analysis of performance is very effective in identifying changes to trading and sales strategies over time. When many firm’s staff are still operating out of the office environment, fully understanding these changes can prevent desks straying from the agreed strategies.
Strategy versus performance
It’s rare to find desks without a reasonably well defined strategy which is agreed with management and risk reflecting the preferred levels of customer engagement, trading performance and risk limits.
But are desks actually following the strategy? We have found some significant deviations from the strategy which can be found in the true performance of the desk. While this may not be a problem for an individual desk, it can lead to duplication and inefficient use of resources across a business.
Statistical analysis
An effective starting point for our performance analysis is a statistical analysis of the trading and sales results over a number of years. Key statistical parameters can be readily calculated from the daily P&L data which can then be aligned to the expected outcomes for desks with similar strategies.
Desks can be broadly classified as ‘flow’, ‘proprietary’ and ‘hybrid’. Although each desk is not uniquely one type only, they typically have strategies which place them predominately in one or the other. Each type has quite distinctive statistical markers which can be tested against the actual desk performance.
In a number of cases, we have found significant differences between the stated strategies of desks and the actual performance. In some cases, reviews have unearthed concerning patterns that lead to important questions and further investigation. This can better inform managers of the daily activity of their teams and help them align their business to their preferred goals.
Runs of positive performance
Long or short runs of positive performance are very insightful when combined with the type of desk and the statistical calculations for that desk. In some cases, long runs of positive P&L are expected while in other cases a more random dispersion of results is typical.
Anomalies in the expected performance runs can show either poor performance and/or the existence of internal or firm advantages. For example, significant balance sheet used to fund certain assets at comparatively cheap rates may be within the strategy of the desk; or it may not. Either way, the true performance might show inefficient or inappropriate use of cheap funding rates.
We have found this measure helps identify aspects of performance that can be compared with the desk strategy and expected performance.
Different time frames of reference
Most performance analysis is done on a monthly and annual basis. Budgets are set and monthly results are often used to identify and adjust expected targets across businesses for revenue, cost and capital.
While this is a perfectly reasonable management approach, we have found looking at performance on a more micro scale can yield significant information to better manage the monthly and annual performance as well as identify areas of improvement. In particular, the daily and weekly P&L data can be used to effectively identify these areas.
Potential misconduct?
While this is rare it still exists. In many of the public cases, ‘post-mortem’, detailed reviews of daily performance compared with the expected performance have shown that preemptive actions could have averted the fraud and saved considerable losses, reputations and remediation costs. We have all seen how modest business units have led to substantial losses far exceeding long-run profitability.
No manager (front office or risk) wants to find fraudulent activity and an independent review can uncover both potential and actual opportunities for unapproved trading that might otherwise remain hidden.
Summary
Martialis has skilled market professionals with significant experience in desk performance-analysis covering most asset classes and types of activity.
Although the emphasis is on improving efficiency though careful analysis to ensure resources are used according to the strategy, we have also found serious concerns about potential and real fraud.
Independence is key. A fresh look at familiar data and some different techniques of analysis can add significantly to the understanding of desk performances and, crucially help management unearth and understand anomalies.