Please find our company and products presentation: this new series of slides include the latest news related to our company, and a detailed product description of IRIS, our counterparty credit risk framework.
1. Global Market Solutions SAS
Consulting services for the Financial Industry
Risk solutions provider
Bespoke Solutions for Financial Software
1
Global Market Solutions
2. 1 – Company Presentation
2 – Company Business
3 – Innovative Risk Integrated Solution
4 – Focus on Performances and Costs-of-Ownership
5 – Our Clients and Partners
2
Table of Contents
Global Market Solutions
3. Company Presentation
3
• Founded in October 2010, Global Market Solutions is a software vendor
providing expertise in implementation & integration projects in the Capital
Market area, including (but not restricted to):
• Summit
• ActivePivot
• Murex
• Orchestrade
• The Company business is located in Europe, Asia and America, includes a staff
of 55 members and 23 customers (Commercial and Investment Banks, Hedge
Funds, Software vendors).
• GMS also invest R&D by mainly focusing on following areas:
1. Counterparty Credit Risk and CVA since Mid-2011 (IRIS Software) with
CORE Business expertise on CVA since 1999
2. OTC Clearing since the beginning of 2012
Global Market Solutions
4. 1 – Company Presentation
2 – Company Business: The company provides dedicated software
solutions answering specific business needs, as well as strong
consulting team of experts for local implementations
3 – Innovative Risk Integrated Solution
4 – Focus on Performances and Costs-of-Ownership
5 – Our Clients and Partners
6 – IRIS Release Management & Roadmap (technical / functional)
4
Company Business
Global Market Solutions
5. 5
Company Financial Figures (1/2)
Global Market Solutions
13%
54%
6%
6%
12%
4%
5%
Group Sales Allocation
(by product)
ActivePivot
Summit
Murex
Orchestrade
IRIS
Sungard
Sophis
34%
62%
4%
Group Sales Allocation (by region)
Europe (exc. France)
France
Rest of the World
7. 7
Company Structure
Global Market Solutions
Illustrations of country and administry districts
Marseille
Paris
Bucharest
Frankfurt-am-Main
• Paris and Frankfurt-am-Main
as European Business Centers
•Bucharest for near-shore
center (extra development
capacity)
• Admin / HR office in
Marseille
Company Locations
8. Global Market Solutions
8
Global Market
Solutions
Subject matter expertise
Project management
IT architecture and development
Front-Office and Risk Business analysis
Back-Office and Operation Business analysis
Outsourced Developments
Fixed price or T&M projects
Bucharest and Paris development centers
C++, C#, F# and Java
Financial Software expertise
Summit
ActivePivot
Murex
Orchestrade
Zeliade
IRIS
Risk Management Software provider
IRIS Software: XVA and related portfolio
management Software
Global Market Solutions
9. 1 – Company Presentation
2 – Company Business
3 – Existing Risk Integrated Solution: IRIS was born for answering
highly complex requirements related to Basel III and Credit Valuation
Management: A proven solution where its backbone can be used to
leverage any solution
4 – Focus on Performances and Costs-of-Ownership
5 – Our Clients and Partners
6 – IRIS Release Management & Roadmap (technical / functional)
9
Existing Risk Software Solution
Global Market Solutions
11. IRIS allows a user-defined
upfront EVA assessment
Upfront Pricing
IRIS provides a robust framework to assess the profitability of a trade at inception
Product costs
CVA / DVA
FVA / LVA
Bank Levy
ATEVA
CoC Default
CoC CVA
Bank
specific
Client
specific
Sales costs
Incl.
Incl.
margin
EVA
CVA/DVA
BankLevy
CoCDefault
CoCCVA
PC
SC
11
Global Market Solutions
12. 12
Global Market Solutions
TOTAL
Contract
Value
Non-
Collateralized
Risk Free
Value
Funding Costs
(FVA Repo)
Funding Costs
(FVA
Premium)
LVA:
Liquidity
Value
Adjustment
CVA/DVA
Credit Value
Adjustment
Trading Desk CVA Desk LVA/Collateral Desk Treasury Desk REPO Desk
PnL Attribution to the relevant desks of the
dealing room
13. LVA and FVA: Recap on definitions
13
Global Market Solutions
Business Case:
• Interest-Rate EUR Swap
• Interbank Market
• Floating Leg on EURIB 1M
• Cash Collateral in EUR
LVA Definition:
The Liquidity Value Adjustment (LVA) is the discounted value of the difference between the
collateral rate and the risk free rate on the collateral, and it represents the profit or loss
produced by the liquidation of the Net Present Value of the derivative contract due to the
collateralization agreement.
FVA Definition:
The Funding Value Adjustment (FVA) is the discounted value of the difference between the
real rate to apply and the risk free rate on the uncollateralized part of the contract. It
represents the profit or loss produced by the liquidation of the Net Present Value of the
derivative contract due to partial non-collateralization.
The Funding Value Adjustment applies only to the non-collateralized part of the operation.
14. IRIS Collateral & Funding at Portfolio Level
14
Global Market Solutions
1 3
2
4
Collateral Management
LVA/FVA Desk
Proper Indicator Calculation
Proper Funding Curve & Collateral Rate
Wrong-Way Risk
Eligibility for Liquidity Coverage Ratio (LCR)
Impact on Net Stable Funding Ratio (NSFR)
Advanced Collateral
Optimization
LVA Desk Specificities
Advanced What-if Requirements on:
Security Collateral Haircut Stress
Rating-based Thresholds Stress
Ability to test when a collateral received becomes ineligible
Imperfect Collateral Usage
Security based Collateral (Coll. Quality Management)
Asymmetrical CSA’s
Rating-based Threshold
Fully Cash-Collateralized Trades
High proportion of large dealer OTC products will
be cash-collateralized:
Ability to deliver Cheapest-to-deliver collateral by
using the right OIS Discounting
Simple What-if CTD Needs at Collateral Set Level
Additional Margin Requirements
15. Risk-adjusted
Valuation Engine
• Server side modules
• Client side modules• Natively Multi-threaded
• Grid compliant
• CPU / GPU / MIC Hybrid
Architecture
• In memory cubes
• Strong ETL
• Various input
formats
• FPML compliant
• Markit and
Bloomberg compliant
• Dedicated API for each
module
• Core features built with
APIs
• External plug-ins available
(e.g. for Summit Trades)
• Interface with Zeliade
IRIS Technical Requirements
15
Global Market Solutions
16. IRIS Software design (architecture)
Data
Integration&pre-processing
Static
Data
Trade
Data
Market
Data
IRIS.Server IRIS.Client
DataEnrichmentAPI
Client Model API Plugin
*Intra-Risk
Correlation
Model API :
• Calibration
• PCA
• Diffusion
• Pricing
Reporting
• Profile Display
• Settings
Real Time Analysis :
- Incremental Update
- WhatIf Scenarios
Screen Customisation
• View Extension
• New Trade View Design
Real-Time Calculation Agent
APIExtensions
Messaging(RabbitMQ)
• Subscribe/Publish
• Trades
Risk-Factor
Calibration
Aggregation
Pricing
Diffusion
16
DataClean-up&curvestripping
Pre-processingAPI
*Risk-Risk
Correlation
Matrix Reduction
*Intra-Risk Correlation is typically a PCA used for reducing the dimension of a risk factor (e.g. from 50 Brownian to 3 Brownian in a Forward Curve),
whereas the Risk-Risk Correlation Matrix Reduction is a PCA reducing the nb. of Brownian involved (which is huge for Equity & Commodities)
17. IRIS Server: Pricing integration Possibilities
17
Global Market Solutions
Multi-threaded CPU
architecture on both
Brute Force MC and
American Monte Carlo
Ability to use default
diffusion or pricing
routines supported by IRIS
on CPU architectures
Dedicated GPU/MIC
designed Pricers for
massive parallelism
Cash-Flow Model and Black-
n-Scholes Models written in
CUDA/OpenCL allowing large
porfolio pricing
IRIS Model APIs calling
external pricing libraries
from ZQF (Zeliade Quant
Framework)
ZQF privides a wide range
of diffusion and pricing
models allowing fast
integration across all asset
classes
IRIS Model APIs call any
existing client library and
able to inter-act with
IRIS Framework
Library can be either
integrated as a native C#
component or C++ lib
through a dedicated bridge
CORE IRIS
Classic CPU-grid
Architecture
CORE IRIS
Advanced GPU/MIC
Architecture
IRIS powered
by ZELIADE
IRIS powered by
bespoke library
IRISServer(withModelAPI)fordiffusion/prcing
18. 1 – Company Presentation
2 – Company Business
3 – Existing Risk Integrated Solution
4 – Focus on Performances and Costs-of-Ownership: How to quickly
reduce your FCO?
5 – Our Clients and Partners
6 – IRIS Release Management & Roadmap (technical / functional)
18
Performances and Cost Benefits
Global Market Solutions
19. Performances: Solving the problem in 3 ways
19
Global Market Solutions
• Major Banks / Investors / Clearing Houses have a orthogonal problem to solve:
• Large Vanilla Portfolio (~150K up to 500K transactions) with “basic”
pricing algorithms (usually requiring a Brute Force Monte Carlo only)
• Smaller Exotic / hybrid Portfolio (<100K transactions) with high-complex
and proprietary pricing models usually requiring more advanced
simulation process (e.g. American Monte Carlo)
• Further requirements to adjust the time bucketing on the cash-flow dates of the
portfolio (huge increase of the time grid size)
• Improving ROI by finding the cheapest-to-deliver technology upon Portfolio
valuation complexity is a silver bullet for the industry:
• CPU: Using complex and evolutive pricing models on existing architecture
• GPU: Leveraging on cutting-edge technology for stable vanilla pricers
• MIC: on-going R&D for evaluating performance benefits of intel MIC Cards
without involving costly source code rewriting
20. Performances: Problem solving suggestion
20
Global Market Solutions
Number of transactions in the Portfolio (by transaction type)
PricingModelComplexity
Bespoke
ProductsHigh-complex
products requiring
American Monte
Carlo (path-
dependant
products)
Credit Derivative
Products
Inflation
Products
Cross-asset
Products
Standarized
Products
(often
cleared)
Vanilla Product Pricers (GPU area)
Fra / bulletPayment
Swap
capFloor
Complex Pricers (CPU area)
Equity Derivative
Products
FX Cash Products
Large IRD Vanilla Porfolio requiring
massive parallel-computation
21. IRIS on GPU: Benefits of massive parallelism
21
Global Market Solutions
• Super-computers are now all based on
hybrid CPU / GPU architecture
allowing more flexibility during
computation:
• GPU are powerful linear algebra
engines and optimized for high
dimensional problem solving (optimum
benefit for very large portfolio)
• Careful use of Common Shared Memory and Constant Memory will lead to the
best performance and maximize the GPU benefit.
• Dedicated algorithm to optimize financial computation will lead to an additional
benefit across the sub-portfolio loaded by each CUDA kernel
• CPU and MIC Architecture will remain key components for advanced
computation processes (exotic and hybrid products) requiring a far too expensive
refactoring for GPU
22. GPU Speed-up benefit for large Vanilla Portfolio
22
Global Market Solutions
4816326412825651210242048409681921638432768
0tan0ar0
0tan1ar1
0tan2ar2
0tan3ar3
0tan4ar4
0tan5ar5
0tan6ar6
0tan4ar4
0tan16ar16
0tan4ar4
0tan12ar12
0tan20ar20
0tan19ar19
0tan8ar8
0tan5ar5
Nb. Of Monte Carlo Simulations
Speed-up Ratio as GPU / CPU
(computation time)
Portfolio Size
(as nb. Of transactions)
0tan5
ar5-
0tan6
ar6
0tan4
ar4-
0tan5
ar5
0tan3
ar3-
0tan4
ar4
When dealing with significant amount of transactions (>50K), and using enough simulation
paths (>5K), the massive parallelism benefit of CUDA on a GPU platform highlights speed-up
ratio bigger than 5x, encouraging the use of this technology for huge vanilla portfolio.
23. GPU Speed-up benefit for large Vanilla Portfolio
23
Global Market Solutions
• GPU architecture demonstrates huge capabilities to parallelize pricing routines
across each cash-flow of a transaction (Swap in our study)
• Proprietary algorithm to optimize financial data and technical use of the GPU
architecture allowing Forward Rates and Discount Factors being calculating once
for the whole portfolio
• The relative efficiency of CUDA vs. HOST on float numbers is obvious in almost all
cases, as the size of the portfolio increases, CUDA achieves factor 5x against its
CPU-based HOST counterpart.
• CUDA will always be the best choice to achieve high speed computations for
vanilla pricers covering:
• Cash Flow Pricer for linear products
• Black-Scholes Pricer for European Options
24. IRIS on intel MIC:
a well known co-processing architecture
24
Global Market Solutions
• Complex Pricing Models cannot be
easily redesigned for running on
GPU:
• Solving the Cost Problem of
existing code rewriting
• Shortening the Time-to-Market
• Sourcing CUDA Skills is difficult
• Alternative solution is to adapt
existing source code to cope with a
well known INTEL co-processing
architecture, allowing fast and
secure delivery process
• Available since the end of 2012, the first generation of Intel Xeon Phi product
family (coprocessors codenamed "Knights Corner")-computers are currently being
benchmarked against the existing IRIS CPU-based CORE architecture
• Future generations of Intel Xeon Phi products will also address enterprise
datacenters and workstations, making Ad-hoc simulations easily done for end-users
25. CPU-Grid Baseline Costs
25
Global Market Solutions
• The total cost of acquisition (TCA) is roughly only 20% of TCO
• Procurement tends to forget that low-cost architectures often produce higher
costs of ownership
• Labor, maintenance, power and cooling currently drive a higher cost (some 3 to
4 times higher) than acquisition alone
[PERCENTAGE]
[PERCENTAGE]
[PERCENTAGE][PERCENTAGE][PERCENTAGE]
[PERCENTAGE]
[PERCENTAGE]
[PERCENTAGE]
[PERCENTAGE] [PERCENTAGE]
Total Cost-of-OwnershipPurchase
Internal Labor
Outage Risk
Software Maintenance
Hardware Maintenance
Electric Power & Cooling Systems
External (contract) Labor
Data Center Floor Space
Schedule Downtime
Service Level Agreement Penalties
27. Global Market Solutions 27
Make versus Buy: A bank view point*
* Coming from HSH Presentation at the CVA Marcus Evans in London (03.02.2014)
28. Global Market Solutions 28
Make versus Buy: IRIS Framework
as best possible compromise
Buy
As-is Benefits
Make As-is
benefits
• No black-box effect, since IRIS is an open framework
• Start with a light Implementation and add-up things as you go
• No compromise: all functionalities can be extended through dedicated
API’s = leverage on existing CORE features + bespoke devs
• Flexible in amendments: when new pricing, aggregation or outputs
are needed, enrich your solution with bespoke developpments
• Generic and established best practises solution
• Benefit from the latest functionalities available in new releases
• Global Market Solutions helps you achieving proper integration
• Global Offer Consulting + Software solutions = All experts are there!
• Many years of experience in integration projects
• Professional Project Team with various skills
SofwareVendorBenefit
InternalSoftwareBenefit
29. 29
Global Market Solutions
IRIS Benefits: How to shorten your Time-to-Market
on key business features?
IRIS Input
Data Viewers
IRIS Pivot
Table-
based
Reporting
IRIS
What-if
Scenarios
1. IRIS Input Data Viewers: Ability to
quickly benefit from Portfolio /
Market data and Static Data
Viewers.
2. IRIS What-if Scenarios: Ability to
quickly deploy the tool for Sales /
Trading People – powerful UI for
your CVA Desk.
3. IRIS Pivot Table-based
Reporting: Most of your reporting
requirements are met through the
use of dedicated Pivot Tables for all
your relevant end-users
Choose IRIS Solution
Implement IRIS
Framework
Shorten
Solution delivery by
1 year
30. 1 – Company Presentation
2 – Company Business
3 – Existing Risk Integrated Solution
4 – Focus on Performances and Costs-of-Ownership
5 – Our Clients and Partners: A successful track record in numerous
countries proving the ability of the company to conduct projects
anywhere
30
Our Clients and Partners
Global Market Solutions
33. 33
Global Market Solutions : contact details
Sales & R&D
Global Market Solutions
7 Cité de l’Ameublement
F-75011 Paris
Finance & HR
Global Market Solutions
29-31 Bld Charles Moretti
F-13014 Marseille
Global Market Solutions SAS
Global Market Solutions
Paris – Frankfurt – Bucharest
sales@globms.com
Please visit us on the Web: http://www.globms.com
When Risk becomes Opportunity
Editor's Notes
When the dealer sells the EUR Swap, he receives an amount in cash equal to the present value. Then he posts this cash as collateral. The buyer pays the collateral rate on the cash collateral.