Rice University CoFES Conference

1
Rice University CoFES Conference

• Derivatives Risk Systems - An IT Perspective

Steve Beasty Managing Director, CTO Credit
Derivatives and Structured Finance JPMorganChase
2
Agenda

• Derivatives, Analytics, and Risk Systems
• Derivatives at JPM
• Analytics and Risk Systems - Who Does What
• Valuation and Risk Calculation
• A Typical Risk System
• Challenges
• Opportunities
• JPMs North American Technology Center
• Vision and Resource Strategy
• What We Plan To Do in Houston
• Growth Plans

3
Derivatives at JPMorgan

• Largest Derivative House on the street today,
  with over 475,000 outstanding trades
• Interest Rate Swaps 260,000
• Interest Rate Options 50,000
• FX Options 45,000
• Exotics/Hybrids 35,000
• Equity Derivatives 45,000
• Credit Derivatives 40,000
• Competitive advantage in derivatives lies in two
  key strengths
• Ability to bring new, innovative products to
  market quickly
• Ability to understand, quantify, and manage the
  market risk of a derivatives book
• Quantitative modelling and information technology
  expertise are critical elements in gaining that
  competitive advantage.

4
How Were Organized Around Providing Analytics
and Risk Systems

• Derivatives Research

• IB Technology (IT)

• Quant/Tactical

• LOB Strategic Tech Teams

• What They Do
• Innovate new pricing and risk models
• Implement models in the form of analytics
  libraries
• Work closely with traders to vet and tune new
  models
• Typical Technology Tools
• - C/C
• - Excel
• Typical Education
• - PhD, Mathematics-Focused (Physics, Engineering,
  etc.)

• What They Do
• Delivery quick time-to-market technology
  solutions
• Integrate DR analytics into pricing and risk
  management tools
• Work closely with traders to solve day-to-day
  risk management business problems
• Projects measured in days or weeks
• Typical Technology Tools
• - Excel/VBA
• - Access
• Typical Education
• - BS Engineering, Math, Com Sci.
• - Heavy Finance background

• What They Do
• Deliver robust, scalable, enterprise risk
  management systems
• Integrate analytics for products that have
  outgrown tactical tools
• Projects measured in weeks or months
• Typical Technology Tools
• - Java/EJB
• - C
• - Smalltalk
• Typical Education
• - BS, MS Engineering, Math, Com Sci.
• - Finance background

Innovation
Industrialization
5
Trade Valuation and Risk Calculation

• Valuation
• A majority of derivative products are valued
  under a term structure model.
• Techniques include tree-models, monte-carlo,
  others.
• Risk Calculation (Delta)
• Trading/Market Risk - Calculations are based on
  an empirical risk framework
• Sensitivity of a trades valuation to each market
  parameter is measured by iteratively perturbing
  each parameter (e.g. 10 year Swap Yield) and
  revaluing the trade.
• Those individual sensitivities are then converted
  into common equivalents, using the sensitivity
  of a common hedging instrument (e.g. 10mm 5 year
  LIBOR swap).
• The total of the common equivalents across a
  traders portfolio of trades tells him what trade
  he would have to put on to hedge his risk.
• The Challenge
• The model works well for simple products under
  small market movements.
• Under large market moves, a trades sensitivity
  will change (gamma). To have an accurate view of
  risk, a recalculation of the sensitivity for
  every trade is required.
• Risk calculations are compute intensive - most
  banks have the capacity to compute them only
  once/day.

6
Traditional/Typical Derivatives Risk Management
System
Valuation Engine
3
4
Trade Repository
Risk Engine
1
2
5
1. New trades are captured/existing trades
maintained 2. Sensitivity/risk calculations are
broken down into a series of valuation
requests. 3. Valuation requests are dispatched
to compute nodes 4. Appropriate pricing routines
are executed to calculate the requested
valuation 5. Valuation results are aggregated
and converted into sensitivity/risk
representations.
7
Implications and Challenges
Examples
Key Challenges

• Across a typical banks derivatives portfolio,
  tens of millions of valuation calculations need
  to be performed daily.
• As product complexity increases, the number of
  valuations increases in a non-linear fashion.
• The cost of IT infrastructure to support this
  business is high and product margins are
  shrinking as complex products are commoditized.

Product/Business Totals
30 Yr USD/JPY X-CCY Swap
Challenge Cost

• Scalability and support cost of Unix servers is
  high
• Ex 32-cpu Sun server - 750K
• JPMs infrastructure employs over 1100 cpus

Challenge Data
Challenge Complex Products

• Complex Trades can require 1000s of valuations
  each
• Risk/Valuation Engine split is not optimal for
  many complex credit products
• Risk engines need to allow flexible definition of
  new risks

• Aggregation of millions of results stresses
  hardware memory limitations
• Interactive analysis of results stresses even the
  best OLAP tools

8
The Opportunities Moving Forward...

• Time-to-market will continue to be the key
  differentiator in the market
• more flexible analytics to support product
  variations without code changes
• plug-n-play of new analytics into risk systems
• Risk calculations move inside the black box of
  the Analytics libraries
• risk calculation efficiency greatly improves
• relieves the data volume challenge
• Intel/NT/Linux-based replaces Sun/Unix as the
  high-performance computing choice
• Far superior price/performance profile
• Greatly improved fault tolerance profile
• High-performance compute infrastructure becomes a
  cross-industry utility
• common need for distributed calculation
  capability across many industries
• capacity-on-demand, pay-by-the-cycle model
  reduces operating cost, increases flexibility
• People Implications - more overlapping skills
  between quants and IT people
• Quants will need more of an appreciation for the
  practical limits of the technology
• IT developers will need to have a better
  understanding/appreciation for the models

9
JPM IB Technology - Our Vision and Resource
Strategy
Resource Strategy Goals
Organizational Goals - The IB Technology Vision

• Define flexible resource supply model which can
  grow and contract with market demand
• Determine the location of resources (location
  strategy) to identify the potential to move
  resources to lower cost locations
• Reduce cost of business-as-usual support through
  application of differentiated service levels,
  sourcing through strategic partners, and
  application of alternative delivery models
• Secure access to a strategic source of
  development skills which can supply a
  cost-effective set of expertise and talent where
  and when required

• Derive results through strong partnerships with
  the business
• Enable client driven business strategy
• Be excellent solution providers
• Be a learning organization
• Develop and retain core competencies

Currently 95 of resources are located in primary
locations target is to source 38 from secondary
locations.
Current
Future Location Profiles

• Houses strategic positions/organizational
  leadership
• Manages technology where client interaction is
  critical

Primary (e.g. NYC/ London)
Premium Cost Profile

• Same time zone as primary site
• All build functions with the exception of client
  facing technology
• Includes non-client facing functions
• Staffed with JPM employees and Sourced TM

Secondary Site By Region (e.g.Glasgow/Houston)
Moderate Cost Profile

• Asian resource strategy includes
• Moving from high cost locations (e.g. Tokyo) to
  lower cost locations (Singapore / Sydney)
• Cost savings throughout the region are increased
  by staffing with resources from India

Asia
Moderate Low Cost Profile

• Low cost location, able to perform
  standard/maintenance tasks
• Includes potential to move resources from India
  to secondary or primary locations

Offsite, India
Low Cost Profile
10
Houston Will Provide Services Across A Broad Set
Of Applications and Technologies
Houston Offerings

• The ability for primary sites to extend their
  teams and realize reduced costs
• A portal, enabling tertiary sites (India) to be
  come operational at reduced costs
• A support site, where operational staff are able
  to support applications that are run around the
  globe
• Secure access to a strategic source of
  development skills which can supply a
  cost-effective set of expertise and talent where
  and when required
• A project site where lower cost resources can be
  deployed and housed for a specific initiative
• Why Houston? Sizeable market for skilled
  technologists, related industry (energy trading),
  University pipeline

Technologies
Business Applications

• Web Technology Java, Websphere, etc.
• Traditional client/server
• AS/400
• Others

• Equities Trading
• Corporate Credit Risk Management
• Client Information and Profitability

• Debt Capital Markets Origination
• Fixed Income Trading and Risk Management
• Fixed Income Back Office

Software Development Roles

• Technology Developer Leading Edge Technologies,
  New Development
• Technology Developer Mature Technologies New
  Development
• Technology Developer Leading Edge Technologies
  Maintenance Enhancement
• Technology Developer Mature Technologies
  Maintenance Enhancement
• Technology Developer Package Integration/Customi
  zations
• Regression and Other Testing
• Development SA DBA

11
The Houston Technology Center Will Grow
Significantly From 2003 Onward

• Houston Center opened in July 2002, Fannin Street
• India Portal model implemented with Accenture
  with 7 consultants in September 2002
• Current staff totals 60 targeted to grow to 75
  by year-end 2002
• Staff expected to grow to 200 by year-end 2003

Critical To Our Success

• Ability to recruit the best and brightest
  Business Technologists in Houston
• Ability to build and maintain a base of product
  and industry knowledge in the Houston team
• Access to a junior/entry level pipeline of talent