Each day, asset managers and their service providers must track complex OTC derivative trades. They do this by manually reading each confirmation and finding the corresponding trade record to see if they match. A major asset manager challenged us to apply artificial intelligence to instantly perform this same complex matching process. We applied the Levenshtein Distance algorithm to accomplish this. The algorithm, invented by a Russian mathematician, matches two string records and calculates the number of operations needed to match them. When two strings match perfectly, the algorithm assigns it 100 percent. We developed a string identifier comprised of different elements of a confirmation to train our algorithm. When we completed our system, over 80 percent of the OTC confirmations were perfect matches, proving we could substantially relieve workloads in the middle office.
