In this short paper, Point Group’s Head of System, Vahid Mesri, explains the thinking behind the Point Investment Data Intelligence (IDI) platform.
Jorge Luis Borges' short story "The Library of Babel" describes a universe that is a vast library containing an infinite number of hexagonal rooms. Each room contains an equal number of books, and each book is made up of all possible combinations of the letters of the alphabet.
This means that the library contains every book that has ever been written, will be written, and could be written.
There are parallels between this story and the philosophy behind our investment data management system. The Point IDI platform creates and maintains clean investment books of record, with start and end values for every asset.
Although not infinite, using this data, the platform produces a large dataset of analytics across many time periods and asset groupings. This dataset is pre-calculated, stored in a data warehouse and updated automatically as market data changes and trades are updated.
This clean and large investment analytics dataset is of great value to any financial organization. It allows portfolio managers to carry out research without having to make on-demand requests for risk and performance metrics. They can extract any dataset using the tool of their choice and compare data points to make smart investment decisions instantaneously.
Clients can also view their data, slice and dice their portfolios however they want, critically third-party systems can utilize this dataset to present different views of the investment portfolio (accounting, CRM, etc.).
The keystone behind eXeer's architecture is the "data-first" philosophy. Instead of relegating data to the background, we position it at the forefront, emphasizing its primacy. This data-centric approach means that our users always have a holistic view across a domain’s full data spectrum.
To illustrate the significance of this strategy, especially in the realm of asset management, let's delve deeper.
Portfolio data – focusing on the investment view rather than accounting - is always subject to revision, as new information emerges.
NAV prices are updated, and custodian banks and trading platforms adjust details of trades and publish those changes to their clients, sometimes after a considerable delay. Prices of illiquid assets adjust historically in a portfolio to reflect fair values. Back-dated equalisation payments are made by fund managers which affect balances and costs.
It's evident why comprehensive guides exist for back-end operations to navigate these modifications.
Yet, as these shifts occur, a myriad of stakeholders continuously mine this data. Portfolio managers need latest holdings to re-balance portfolios and design trades. Client teams and relationship managers need to access information on portfolio positions and performance to update their clients. While routine client updates are standard, impromptu reports are frequent, especially during market upheavals. Management too is interested to have the latest revenue forecast using up-to-date AUM figures.
A data-first investment data management system helps ensure all these stakeholders have access to the most accurate and up-to-date data. This improves decision-making, client engagement and ultimately lead to better investment performance.
Broadly, the data-first model aids in two primary ways:
First, in addressing fluctuations in 'input data', the system can harness automation. This is not a new idea: using data feeds provided by banks, trading platforms and market data providers, the system can apply changes on portfolio data and monitor variables like performance and valuation. The aim here is to minimise human intervention and change the back-office role from the conventional 'executor' to 'supervisor'
Another advantage of a data-first system is its approach to maintaining 'output' data that remains fresh and accurate even as input data evolves.
This system not only ensures consistency between upstream (IBOR) and downstream (e.g., client holding total in CRM system), but also swiftly recalculates crucial analytics, such as performance and valuation, when there's a change in input data.
For instance, when trade or price adjustments occur in a portfolio, the system automatically reassesses performance analytics across different periods. Data consumers, like client relationship teams, who are subscribed to these analytics, receive immediate notifications about any changes, empowering them to take relevant actions, such as re-issuing client reports.
Creating a data-first system isn’t devoid of challenges.
Just like the library of Babel contains many books with nonsensical text (owning to every letter permutation), a large analytical dataset built with every combination of analytics alphabet - type of analytics, time-period, and calculation subjects - will contain analytics that are not relevant to every particular investment question.
The feasibility of updating all relevant analytics while data changes take place, will also be in doubt. To avoid this glut of information and maintain ‘always up to date’ analytical datasets, the Point platform makes smart decisions about the analytics that it automatically pre-calculates.
For example, it is not necessary to calculate analytics for every possible time-period. Instead, it is more efficient to focus on the time periods that are most relevant to the analyst's needs, then observe user behaviour in requesting on-demand calculations and update the list of pre-calculated time periods accordingly
In Borges's short story, some librarians believe that since all books exist in the library, there should exist a book somewhere that is a perfect index of the library's contents.
Drawing inspiration, we’ve dedicated considerable efforts to craft the optimal "index" for our data, fine-tuning our approach to make Point's data accessible and user-friendly for our clients.
Vahid Mesri - Head of System Development
Vahid has over 20 years of experience as a head of technology, software engineer and data analyst across the technology and financial sectors including in asset management, insurance and pensions.
Prior to joining Point Group, Vahid was the head of technology for a publicly listed asset management firm where he designed and built bespoke trading and compliance software to manage and monitor large client portfolios and in-house funds. Vahid has deep experience in exploiting large, complex financial data sets to deliver value for businesses and end customers alike; he has built and run data analytics solutions ranging from performance engines to hedge fund and equity fund modelling strategies.
Vahid leads on the development of our proprietary investment technology systems and architecture.
He has an M.Sc. in Machine Learning and Data Mining from Bristol University.