Goldman Sachs & Co
One of our partners was a founding member of the Portfolio Optimization and Indexation system group in the Fixed Income Research department of Goldman Sachs. This group was created in order to develop an integrated software product in order to better service portfolio analysis and optimization for institutional clients.
The software enabled Goldman to enter into new businesses by providing fixed income portfolio optimization services for institutional clients including pension funds and insurance companies. This was previously dominate by other companies such as Salomon Brothers.
The software was designed using Objected Oriented design methodologies using AT&T C++ on an X Windows environment using a Motif based font end and Sybase data base. The software development involved integrating additional software developed by other groups. The heart of the Analytic engine was a polymorphic interface used to wrap other non object oriented analytic engines to provide a uniform interface for analysis of Bonds of various types and markets. This software was also one of the pioneering efforts at Goldman Sachs for using 3 Dimensional Computer graphics for data visualization. This had very impressive results from both a marketing and analysis point of view.
Responsible for most analytic models of the Software system developed. Developed and implemented methodologies for various kinds of analysis for various bond markets and analysis on a portfolio level. Work included leveraging all existing Goldman Sachs analytics and developing a new analytical framework so as to provide maximum benefit to the firm. Further responsible for UI design and implementation and design and implementation of a 3 D graphics library for visualization and presentation purposes. This was at a point that few such libraries were commercially available on Sparc stations.
Helped Goldman's market research and Industry resource analysts analyze several client portfolios including cash flow matching of liabilities with portfolio cash flows, breakdown of portfolios, asset liability management, mortgage market analysis under different scenario etc. Also trained the IRG analyst regarding the underlying analytic methods, spread pricing, term structures, scenario analysis, relative value measures and the underlying option models.
Technology Work Included
- Extensive User Interface Design using Motif/Xwindows.
- Extensive Analytic Software design as described below.
- Data Base access using Sybase and GS Dbo interfaces.
- Full 3-D Visualization Package developed on top of Motif.
- Use of Yacc++ for developing data structures and analytic routines read in from files.
Software Model development work included
- Term Structure Analytics
- Bond Analytics. US Domestic and International markets
- Rates of Return Analytics
- Currency Analytics
- Portfolio Analytics including aggregated cash flows.
Details of work Follow
This is a more detailed description. This could be of interest to people with a detailed fixed income securities and international bond and currencies markets background.
International Term Structure Analytics
- Term structure representation using quadratic B Splines for US Treasury term structure. Used a raw spline for analysis of US Treasury and Mortgage markets and an adjusted spline (one made to go through the on the runs) for the corporate, MTN’s and mot other US Markets.
- International term structures represented as spline for liquid markets particularly UK, Japan, Italy, France, Canada and Germany.
- International term structures represented using discrete bonds for the other markets. Preferably using on the runs if they defined enough of a term structure, else using the entire market for term structure representation and use for rolldown calculations for such markets.
- Full forward curve analysis of term structures of the various countries. Calculation of representation of the curves in any form i.e. as a par, zero, coupon or a discount curve.
- Horizon scenario definition as par, zero or coupon shocks to the appropriate term structure. Could define shocks over a shocked curve for intra day scenario analysis. Could use zero day horizon for instantaneous shock analysis.
- Index generation using the forward curves for among other Libor and CMT indices. Could compute index of arbitrary length at arbitrary spacing in time.
- All interfaces to existing Goldman software except for mortgages.
- Incorporation of most US Domestic markets including treasury’s, corporates, muni’s, money market instruments, cd’s, CMO’s, mortgage passthroughs, Asset backed securities etc.
- Incorporation of most developed international bond markets including gilts, bunds, jgbs, jbds, italian bonds, canadian bonds, french bonds and bills (including OATs and BTANs), austrian, australian, danish, swedish etc.
- P/Y, duration, avglives augmentation for having durations and convexities in various forms including Goldman Sachs'’s own standards.
- Extended spread analysis ranging from par, zero, static and a Goldman market research standard fitted spread. These provide not only rich/cheap measures but are also used for spread pricing both intraday and at a terminal date for horizon Rate of Return calculations.
- Break even analysis. Forward Rates (Price and Yield) for any particular bond based on its repo. Furthermore computation of BE Yield changes and CMT Yield changes. These numbers computed for a hedged bond for the foreign markets.
- Holdings and market value analysis with currencies and accrued.
- OA analysis for callable US Domestic securities. Bob Littermans term structure of credit spreads used for quoting effective spreads and effective durations. This model assumes term structure to credit spreads based on the industries. Option model for corporate’s using the treasury term structure and for municipal bonds using the muni GO curve. Also wrote an interface to mortgage option and ROR model.
Rate of Return Analytics
- Scenario rates of returns computed for all bonds. Scenarios modified by parallel as well as non parallel shocking. Different kinds of phase in for the scenarios. Intermediate scenarios definition also possible for using varying interest rate paths.
- Generic way of managing reinvestments using short end of the appropriate term structure or even reinvesting in other bonds.
- Multiple methods for terminal pricing ranging from constant horizon OAS through constant horizon static spread through specification of and P/Y/Spread value at horizon.
- For callable bonds using optimal exercise for ROR analysis. This required a terminal Black Derman Toy tree for valuing the callable bond at horizon.
- Multiple scenario capability at horizon.
- Developed a currency analysis framework for bonds. One could hedge or compute home or bonds denomination based holdings and market value numbers. Rates of returns also generated in such fashion.
- Forward cross rates used with assumed future spot rates for break evens, rates of returns and matrix of interest rates vs. currency scenario analytics.
- Portfolio cash flow aggregates weighted by par amounts.
- Use of Portfolio cash flows for portfolio optimizations such as immunizations and asset/liability cash flow matching.
- Scenario based cash flows for the portfolio and individual securities.
- Portfolio aggregate numbers. Aggregates computed with the appropriate weighs. Special methods for averaging Rate of Returns for the portfolio.
- Break down of aggregates by completely generic sectors defined by the analyst.
- 3 D view of Portfolio Profiles and for Portfolio cash flows.
3 D Graphics
- Developed a completely general purpose 3 D visualization module for data visualization. This included scenario cash flows, tranche cash flow for CMOs, PSA and GS Prepayment model scenario cash flows for mortgage passthroughs and CMO’s. These graphs are shown in Goldman Sachs 1994 Annual Report.
- Portfolio profiles displayed in 3 D
- Sector break down summaries displayed as 3 D graphs.
- Yield surfaces and Yield curves display.
- Complete manipulation of 3 D graphs and printing as post script report.