Global Valuation research has created a new architecture for secure, data privacy respecting analytics. We call it Permissioned Analytics, as the analytics cannot expose any private data that the data owner does not agree to be shared with the author of the analytics.
Permissioned Analytics is a safe way for data owners to allow published analytics to be run on their private data, such as banking or hedge fund portfolios. This approach will have many use cases beyond Financial Services. We illustrate here some use cases that will change how analytics are used in finance.
What is Permissioned Analytics
Analytics which are calculated on a contract portfolio details of which cannot be directly accessed by the initiator of the calculation.
Example: A bank can compute the VaR on a hedge-fund’s portfolio, but has no direct access to the positions of the hedge-fund with other dealers. In this case VaR is an analytic that is permissioned to have access to the full portfolio information at run-time, but can only report the VaR number to the initiator. VaR itself is expressed by aggregation logic to which both the bank and the hedge-fund have to agree as part of the permissioning process.
How Permissioned Analytics works: Illustration of Wrong-way Risk in Clearing
Features of permissioned analytics:
Fine-grained permissions on both data and analytics are under the same security context and permissioning
Permissioning attributes identify the permissioning Party, the calculation Party and data access rights
Full security is achieved by performing calculations on hardware controlled by the permissioning Party, without the need for Confidential Compute infrastructure and by defining the full logic of the analytics tokens
Each Party can validate the analytics source code on the blockchain and their outputs
Roadmap for Permissioned Analytics
Three prerequisites must be present in organisations wishing to use Permissioned Analytics with their partners/clients/service providers:
A permissioned ledger technology with built-in data privacy capabilities and synchronisation of data between members of the network. Create a trust relationship with the publishers of tokenized analytics
Make portfolios available to Esther for analytics execution. Ideally, use data formats that are "compilable" - where software can automatically use the data, such as derivatives contracts, without any data integration activities. An example of this approach is the ISDA Common Domain Model or Esther's own smart contracts
A universal solver that will always execute analytics in the same, consistent way across all members of the network - such as Esther Solver. Esther requires a very small infrastructure footprint and can be made available with every instance of a ledger or permissioned blockchain that requires analytics.
All of these prerequisites exist already. They just have not been fully integrated yet. The integration effort is minuscule in relation to the business benefits of Permissioned Analytics.