Complex Model Development & Validation

We helped our client, a top 5 index provider, develop, test, validate, reconcile and operationalize 500+ index models that involved complex instruments like options, futures, & other derivatives.

Problem Statement

• Client had a suite of complex models that needed to be validated afresh or migrated from archaic frameworks like Excel/Informatica.
• New models with increasing complexities and with derivative instruments as underlying were to be validated, a task which was impossible using client’s current framework.

Solution – Development of new model validation framework

• Technology Identification
  • Scope of the project was assessed and model complexities were ascertained
  • Since data requirements were enormous, a framework built on new-age technology was imperative
  • MATLAB was selected as the choice of framework owing to existing libraries, and customer support

• Development
  • Using MATLAB and SQL procedures, an initial set of code libraries were created for generic re-usable functions such as data retrieval, input/output, publication, etc.
  • Model-specific rules were then coded for each investment strategy, and the models were validated against the client values
  • Use of MATLAB afforded scale and efficiency, which enabled validation of models of even higher complexities than before

• Maintenance
  • The service offering included regular operational and maintenance services
  • Since all these models are running live, a dedicated operations team looks after on-the-go issues pertaining to underlying data, model rule changes, resulting values, etc.

Achievements

• A team comprising developers and domain experts was setup in under 15 days, and became operational in under a month
• Over 500 such complex models have been successfully developed and tested, and are being currently maintained regularly
• Entire development and maintenance run by Indxx, allowing client to focus on business development and sales efforts

Parallel & Independent Index Reviews

We helped our client setup independent & automated framework to review and rebalance portfolios, ranging from a few hundred to over 10,000 securities. This helped them ensure correctness, manage risk through quick reconciliation and better adhere to investors’ expectations.

Problem Statement

• Client had around 12000 securities in initial pool to review each review/rebalance period.
• There was no documentation of the rules/exceptions followed by 3rd party calculation engine.

Solution – Development of Review Tools

• Model Validation
  • Setup standard libraries for selection and weighing rules.
  • Setup of data APIs to get processed data.
  • Setup of a master Python code covering calendars and investability checks.

• Develop and Test
  • Master Python code run once every day through automated scheduler.
  • Automated scheduler fetched calendars and identify the calendars for which review process is active.
  • Performed review and rebalancing for all the indices linked to active calendars.
  • Internal quantity checks and controls on reviewed and rebalanced portfolio were implemented.

• Parallel Comparison & Reconciliation
  • Automated comparison of client’s and Indxx files was executed.
  • Rebalanced and reviewed portfolios were sent to SFTP location where they could be ingested from by client.
  • Generation of a detailed discrepancy report.
  • Detailed analysis outlining source of differences and corrective action proposed.

Achievements

• Generating reviewed and rebalanced portfolios along with a detailed report with entire universe having all the data points used in portfolio construction, with exclusions and calculation at each stage.
• Discrepancies were analyzed and root cause analysis and remedial action plan was proposed.
• The developed tools that helped to automate the review/rebalance process saved the client a huge 42% on operating costs.

Index Data Golden Copy

We helped our client identify & rectify discrepancies in its historical index data stored in multiple locations, that led to better client management as well as accurate product development and maintenance.

Problem Statement

• Client received a request for historical data of flagship indices from fund managers, HNIs and family offices.
• There were severe discrepancies in historical data with multiple versions for same dates at different locations.
• This was impacting new product development where historical compositions were used as starting universe, and prices & FX rates from historical files were used for simulations.

Solution – Creation of a new Data Governance framework

• Governance Process
  • Scope of the project (DBs, time period and indices) was defined.
  • Data reconciliation process was created.
  • Requirement document was created for development team.

• Development
  • Using Python and SQL procedures, governance process was automated presented for analyses by data governance team.
  • Tools were created for automated comparison between different DBs and conflict resolution mechanisms were suggested.
  • Except for special cases (not defined in requirements and/or needing further root cause analysis), golden copy process could be completed for nearly 10,000 indices in 1 day.

• Golden Copy
  • Index compositions, divisors, levels, calendars & index market cap were validated for parent indices.
  • Securities’ reference data, open prices for PR, GR and NR versions, close prices, market cap, free float and exchange rates were validated for securities in parent indices.
  • Iterative process was run to fix new/existing discrepancies at each iteration.
  • All special cases were handled per guidelines from stakeholders.

Achievements

• Client made nearly US$ 1 million within 3 months of successful implementation of data governance framework.
• Backtest of new products were more accurate as correct data for historical compositions and prices were used by backtest application.