Global Custodian Spring 2018 | Page 51

[ M A R K E T develop data tools that help its asset man- ager clients make investment decisions. “We are applying data analysis tools and using artificial intelligence to help drive data decisions for clients. Canada is a centre of excellence for AI and RBC has several links to universities that are pro- ducing data scientists that we can lever- age for our business and provide insights into how to use the data,” he adds. Data dangers With such a heavy emphasis on data, both from the custodian and the asset managers, there is a danger of data over- load. Custodians need to be both wary of all of the sources of data they are gathering, and also aware that they may be providing too much data to the client, something which State Street’s Pagliaro believes could result in clients turning the other way and ignoring it. “If you are doing pre/post-trade anal- ysis, you have your standard sets of data you can get by, but any more could be overload, which clients tend to avoid,” says Pagliaro. With newer collection methods and data sources being discovered every day, Pagliaro adds a lack of standardisation in these sources and process could result in skewed or incorrect insights by the client. “In the investment management space, when you track data flowing through their accounting systems, that operational data is very valuable but it suffers from a lack of standardisation. It tends to be a very manual process to handle all of these data sets,” he says. Furthermore, asset managers that are so heavily relying on finding patterns in their data to make decisions are in danger of putting together wrong correlations be- tween data sets. Without a comprehensive due diligence process for custodians and third-party providers, they are in danger of formulating incorrect conclusions. “Because people are striving to become better traders or becoming more efficient, they are looking more into data sets and run the risks of finding correlations that do not exist,” says Anders Kirkeby, tech- nical fellow, vice president for enterprise architecture at SimCorp. “What we are doing is taking the data from different vendors and applying rules R E V I E W | B I G D ATA ] to mitigate any differences between those data sets that will provide something that is a golden copy. “Clients are looking at the cost of working with individual brokers and custodians because they all have different data qual- ities. We have clients that have hundreds of counterparties that they deal with, but they are a group that continue to serve up poor quality of data. For that, they are either going to have to re- negotiate their contracts with them or end business altogether.” Asset managers also have to ensure when they are outsourc- ing data management is that they do not rely entirely on their “A proper handle on these vast amounts of data requires these institutions to focus on their business objectives and on building out their data capabilities step-by-step.” MIKE CLARKE, DIRECTOR OF PRODUCT MANAGEMENT FOR EUROPEAN CUSTODY, DEUTSCHE BANK third-party partner. Regulations such as MiFID II require all firms to be responsible for the quality and accuracy of their data, even if they outsource it. The result of which, argues Kirkeby, could mean data man- agement being brought back in-house in order to oversee the governance of that data. Next stage Following the evolution of big data into data analytics, arguably the next stage of this cycle is robotics and AI. These data management and data governance tools are increas- ingly being replaced with machine learning, which can be used to consumer, consolidate and utilise in one place. A number of custodians have already begun to implement machine learning and AI into their data lakes. However, new technologies should not only be used as a means to replace lega- cy processes to do the same thing. Custodians have long been criticised of failing to innovate in the data space; they would follow the guidelines in manging it but there would be no process to improve it. Data analytics must include data assessment, as well as produc- tion. With asset managers moving into new and more complex products such as tracker funds, new technologies must be used more carefully. With data management becoming more and more a part of the custodian’s business model, they must realise they the technolo- gies they will harness has to do more. Spring 2018 globalcustodian.com 51