Pink Iguana

Home » Uncategorized » Nimo Development to do list

Nimo Development to do list


This is a list of to do items for the static balance sheet simulation code in github/jsandber/finding_nimo as of Feb 2016:

  • Simulate taking market data from the user from main() to classes
  • Thread the user input market data to the static security simulation
  • Generate benchmark elapsed runtime for the static security simulation with user market data
  • Simulate taking a user’s accrual portfolio from main() to classes
  • Map a user’s accrual portfolio to the current Reference Server security models
  • Generate benchmark elapsed runtime for the static security simulation with user market data and user accrual portfolio
  • Devise, test, and implement some exception checking, handling, and reporting for incomplete market data coverage of required security models
  • Simulate taking a user’s initial balance sheet + amendments ( balances not to be computed from the Reference Servers models).
  • Map statically simulated products to empty balance sheet accounts using user input denoting the mapping.
  • Map statically simulated products to initialized/amended  balance sheet containing externally generated balances and rates.
  • Expand the set of native products from cards, deposits, and fed funds to a first approximation of the FRB product set.
  • Implement and test  test modules for the initial classes.
  • Fit the product balance and rate models to the historical FRB data – there is going to be some experimentation here as there are many forms of linear and non-linear regressions we can fit. We don’t want to spend a lot of time on it but it would be  good to have a reasonable view and some idea of the expected forcasting error (which I would expect is relatively large). The reason we do not feel the need to spend time on it it the GSIBs and the FRB are going to have their own take on what the right level of regression is. We are not likely to guess correctly.
  • Devise three user tests from the FED market data and aggregate portfolios
  • Implement the driver repositories for the FED data and portfolio
  • write the main() function that pulls from the local FED repo and runs the static balance sheet simulation – actually you can run the FED code as a direct product simulation.
  • Run internal demos of the FED  Simulations
  • Generate some elapsed runtime scaling with  increases in the number of modeled products – we should be good to a billion securities particularly if we are going to push the angle of statically simulating all USD assets and liabilities for the FED under a TBD standard FED model.
  • Expand the balance models to include subtypes: retained, acquired, and runoff – this is the standard hack for supporting aggregated deposits, cards, loans, or mortgages. Again we have no part in selecting the aggregation levels, that is a GSIB users call. The Reference Server does not know anything about the level of aggregation in the GSIBs securities.
  • Pull these classes into a AWS Server implementation  for static balance sheet simulation
  • Design the Server initialization and reset processes
  • Design, document, and implement the Reference Server API
  • AWS Reference Server testing
  • Generate papers and prose for the demos
  • Line up demo meeting with folks at the NY Fed. I see Gupton and Finger are in DC. I see Testa in NY. We can work out the basics of the testing idea for the CCAR submissions  – How would CCAR Static Simulations be automatically tested if you could do  a billion different CCAR simulation in a day? It might take a couple weeks to get good results on the topic but we can drop a reasonably compelling position paper outlining how that might go.




Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: