Duplicate Trades

There are things that can go wrong with Front Office systems that may go undetected. One common issue is duplicate trades.

Whilst most duplicate trades are detected by “filtering controls” that look for a repeated Transaction ID, there are instances where Front Office systems or process creates a unique Transaction ID for the same trade.

This may occur as a result of a number of reasons, some of which being: user error, a fault in the Front Office system, a coding issue in a bespoke process that extracts raw trade data, edits on existing trades that erroneously fires off a new, cloned transaction instead of issuing a CANC-NEWT on the existing transaction.

Duplicate transactions may be difficult to detect. This is because some transactions may look like duplicates but they are not. For example, a block-fill order that is executed may be comprised of individual transactions that share the same characteristics such as buyer, seller, trading datetime, quantity, and price. These transactions are actually separate transactions executed in the market and each one is subject to transaction reporting.

One approach to detect duplicated transactions is to execute custom processes that ignore block-fill transactions and target transactions having the same data in key fields.

When duplicate trades are created as a result of an edit to an existing transaction, these can be detected by more sophisticated algorithms that rely on historical trade data. Whilst these algorithms can be resource-intensive, they are effective in analysing each new transaction and comparing it against all previous transactions sharing that same trade datetime field.

Many firms may be unaware that their Front Office systems are producing duplicated transaction until an issue is spotted in a routine Quality Assurance check. These routine checks take a sample of data and apply a series of Quality Assurance and logic checks across the fields. They tend to be less resource-intensive as they only analyse a small sample of data rather than the entire data set.

Once duplicate trades have been identified, then it is a matter of detecting the origin of the duplication and putting controls in place to prevent the issue from reoccurring.