Mifid IIAug 14 2017

How AI can help advisers with Mifid II

  • To understand what Mifid II is going to bring in.
  • To learn what hidden obstacles might lurk in the legislation.
  • To learn what sort of strategies can help ensure compliance.
  • To understand what Mifid II is going to bring in.
  • To learn what hidden obstacles might lurk in the legislation.
  • To learn what sort of strategies can help ensure compliance.
pfs-logo
cisi-logo
CPD
Approx.30min
pfs-logo
cisi-logo
CPD
Approx.30min
twitter-iconfacebook-iconlinkedin-iconmail-iconprint-icon
Search supported by
pfs-logo
cisi-logo
CPD
Approx.30min
How AI can help advisers with Mifid II

At the core of many AI programmes is Natural Language Processing. NLP is nothing more sinister than using different steps to develop a way for computers to understand human speech, and has been a field of computer science since the 1950s.

Examples of the evolution of NLP include Google Translate, automatic email spam detection, and the condensing of newspaper extracts online.      

When an AI engine reads through a piece of legislation, like Mifid II, it can use its NLP to model and break down each single regulation into its underlying concepts; for example, spotting and classifying the people or organisations that need to act, identifying the goals, and any timescales or deadlines contained in that paragraph of text.  

The AI will then build up a conceptual map of the document, generating an understanding of how these concepts are related to each other and how similar articles within the legislation are to one another.

For example, on this basis, the AI will build up an understanding that a derivative contract is (i) a contract for (ii) an underlying product.

The AI can digest and classify a long piece of legislation like this within seconds. For example, consider this from our trained AI at Waymark Tech, which has broken down a specific requirement from MiFID II Article 16(7):

"...an <actor>investment firm</actor> <deontic-obligation>shall</deontic-obligation> take all reasonable steps to record <information-object>relevant telephone conversations</information-object>...  The records kept in accordance with this paragraph <deontic-obligation>shall</deontic-obligation> be provided to the <actor>client</actor> involved upon request and <deontic-obligation>shall</deontic-obligation> be kept <temporal>for a period of five years</temporal>..."

The AI now has all the relevant information from the legislation to make a decision on whether or not to present this requirement to the compliance officer. It will only do so in this case if the person using the system has already stated that they are an investment firm.

The AI can then present further information, such time conditions, on request, providing compliance officers with a more manageable way to navigate the legislation.

The benefit of adopting technology for this process is clear. Analysing the text will be infinitely quicker, it will minimise the chance of human error, and it will organise and centralise the regulatory text, giving compliance officers only the parts that apply to that organisation – or they have specifically requested.

Streamlining implementation by identifying regulatory overlaps

Another of the great challenges of implementing Mifid II has been identifying overlaps with other existing regulations or packages set to be implemented in the future. Where are the crossovers? If by doing one thing to implement Mifid II, are we now failing to comply with another, entirely different, set of regulations? 

PAGE 3 OF 4