Razor Risk is a risk management software framework provided by TMX Technology Solutions that allows financial institutions to measure market, credit, and other risks across all global activities in real-time. The Razor Risk Market Risk module provides comprehensive market risk functionality including value-at-risk calculation, reporting, drill-down analysis, scenario analysis, and stress testing to help users meet regulatory requirements. It utilizes both historical simulation and Monte Carlo simulation for value-at-risk and allows drill-down of results to the individual transaction level.
2. About Razor Risk
TMX Technology Solutions recognizes that to proactively measure and manage risk it is necessary to manage the
total exposure of a financial institution across of all of its global activities. TMX Technology Solutions’ products,
including Razor Risk, have been created to help transform the way banks, brokers, central counterparties and
stock exchanges in many countries measure their risk and manage their capital.
3. Overview
TMX Technology Solutions’ award-winning Razor Risk framework provides near real-time risk and real-time, pre-deal calculations that enables
management to view in one consistent firm-wide view the exposure to individual entities. Our clients use Razor Risk’s advanced analytics
and scenario calculations to achieve best practice in managing their complex portfolio hierarchies as well as risk exposures for credit, market,
clearing and liquidity risk within a single application.
Since Razor Risk is a framework and not a risk measure, practitioners can easily incorporate new sources of risk and accommodate innovations
in best practice risk management. Razor Risk also assists financial institutions in satisfying their requirements under the Basel Regulatory
Framework and IOSCO’s Recommendations for Central Counterparties.
Razor Risk has helped improve the way banks, central clearing counterparties, stock exchanges, hedge funds and other financial institutions
across the globe measure their risk and manage their capital.
PRODUCT STRUCTURE
Razor Risk comprises an integrated set of modules to provide a multiple-risk and cross-asset solution that can easily be configured to suit
financial institutions’ individual risk needs and appetites. The combination of Razor Risk’s modern and open design with a modular, geographical
and volume-based licensing model enables our clients to install the functionality required and to scale and adapt as business requirements and
regulatory requirements change.
THE RAZOR RISK MODULES ARE:
Core Services (base architecture)
Market Risk
Credit Risk
Limit Management
Economic Capital
Basic Analytics
Advanced Analytics
Basel III Trading Book
Market Risk Module Overview
Razor Risk’s Market Risk module provides comprehensive market risk functionality to enable a large financial institution to fully meet internal
and regulatory market risk management requirements. Razor Risk’s Market Risk module is fully-integrated with Razor Risk’s Credit Risk and Limit
Management modules. These modules are fully-integrated with a single trade repository, enabling integrated market and credit risk management.
THE RAZOR RISK MARKET RISK MODULE INCLUDES THE FOLLOWING FUNCTIONALITY:
Value-at-Risk (VaR) Calculation
Historical simulation
Monte Carlo simulation
Credit VaR
Specific Risk
Partial Risk
Reporting, Drill-down and What-if Analysis
Extensive market risk reporting within Razor Risk
OLAP reporting for sensitivity analysis
Capability to develop external reports on the Razor Risk results
Full drill-down to the transaction or scenario detail to analyse all market risk results
What-if analysis for new, amended or deleted trades
4. Market Risk Workflow
Base-lining of new VaR results to a previous base-line
Incremental processing of any amendments for real-time VaR analysis
Full audit trail and logging of all aspects of the market risk calculation
Scenario Analysis
Pre-defined scenarios
User-defined scenarios
Stress testing
Extendibility
New pricing models or products can be added to the calculation
Results can be aggregated externally to Razor Risk and aggregated into the overall calculation
Liquidity and Funding Risk
Reporting of all cash flows by currency for all deals
Inclusion/exclusion of interim coupons
Back-testing
Back-testing of VaR results to validation model
Back-testing against theoretical P/L (calculated by Razor Risk)
Back-testing against actual P/L (fed to Razor Risk)
Value-at-Risk Calculation
Razor Risk supports historical and Monte Carlo simulation for VaR calculation (HSVaR and MCVaR, respectively). Both methodologies leverage
Razor Risk’s distributed architecture, enabling calculations to be performed quickly and accurately. Razor Risk uses full revaluation of trades for
the calculation of both HSVaR and MCVaR.
The time horizon, confidence interval, base currency and number of scenarios are configurable for both methodologies. Confidence interval,
base currency and time horizon are defined at the portfolio level so different settings can be used for different portfolios.
Razor Risk’s portfolio aggregation capability enables users to drill down to any desired aggregation level, for example,currency, product type,
risk type, business unit or combinations of any of these. These aggregation levels are user-definable, and there are no pre-conceived notions
of hierarchies inherent to the data design. Razor Risk supports full drill down to the transaction level to determine contribution to the overall
risk results. Razor Risk also enables VaR results at any aggregation level to be compared against a baseline set of VaR results. This simplifies
significantly the validation of VaR results on a daily basis.
VaR results are stored in the Razor Risk database from where they are accessible for back-testing and historical reporting purposes. Razor Risk
supports overnight, intraday and near real-time calculation and reporting of VaR.
Razor Risk has been designed to provide excellent performance to enable the calculation of VaR in real-time. Razor Risk’s distributed processing
architecture enables full, sub-second historical and Monte Carlo simulation using an affordable technology platform, and these response times
include the application of any portfolio effects.
The distribution of analytics and portfolio aggregations across multiple servers enables Razor Risk to cache the trade data and all calculated
values across each server. Razor Risk leverages this when adding incremental trades, performing what-if analysis, and during pre-deal check
functionality.
Razor Risk analytics also leverage data caching for intermediate pricing data such as discount factors and hazard rates. On a given path and a
given aggregation node, the discount factors for a particular yield curve need only be calculated once, and these factors are cached and used to
price all trades corresponding to that curve.
5. HISTORICAL SIMULATION
Razor Risk’s HSVaR methodology supports the full revaluation of all transactions. Razor Risk’s unique balanced distributed processing capability
efficiently organises and caches the multiplicity of risk valuations over networked servers, delivering exceptional performance and scalability at
a fraction of the cost of alternative technologies. Benchmarks demonstrate Razor Risk’s outstanding supercomputing performance and linear
scalability.
VAR BREAKDOWN
Razor Risk supports the calculation and reporting of VaR at any aggregation level. These aggregation levels are user configurable. Razor Risk’s VaR
calculation revaluates each transaction for each scenario. Once this task has been completed, Razor Risk aggregates the individual transaction
results into the relevant portfolios. This approach ensures all the portfolio effects are captured for each portfolio but maintains the efficiencies
of only pricing each trade once - regardless of the number of portfolios in the system.
Risk can also be reported by geographical location. A location hierarchy can be defined in Razor Risk which breaks down locations into regions
and then into specific trading centres. The booking location for a trade is then determined, and VaR is aggregated for that trade based upon the
location hierarchy.
Similarly, risk can be broken down by profit centre by defining an internal unit hierarchy which mirrors the organisation’s internal structure and
aggregating risk to the appropriate profit centre is based upon the internal unit the trade was booked through. The same approach is used for
calculating risk from the trading and banking book.
Limits and limit utilisation can be measured for any portfolio aggregation defined within Razor Risk. Razor Risk can be configured to trigger an
excess event when a portfolio threshold is breached.
All Razor Risk VaR results can be reported in either tabular or graphical format.
FIGURE 1 - VAR DRILL-DOWN IN RAZOR RISK
6. PARTIAL RISK
Razor Risk’s partial risk and drill-down functionality provides the ability to decompose VaR into each of its components (Figure 2). This enables
VaR to be isolated to interest rate, foreign exchange, equity or volatility risk. Users can define which sources of risk should be isolated in the
partial risk calculation.
Razor Risk’s partial risk functionality enables VaR to be calculated for specific risk classes. Each market data input can be assigned to a risk class,
as this is also user configurable. When calculating the contribution towards VaR for a specific risk class, only historical data for the required
risk class is used, and the other risk classes are kept constant. Effects from combinations of risk classes can also be calculated and reported.
FIGURE 2 - PARTIAL RISK BY RISK TYPE
SPECIFIC RISK
Razor Risk provides two choices for the calculation of specific risk.
CREDIT MIGRATION MODEL
Razor Risk supports the modelling of credit migration behaviour of issuers or individual issues. Issues are first mapped to the appropriate credit spread
curve. The mapping is configurable but common approaches are to map the issue based on industry sector and credit rating. A mark to market (MTM)
factor can be calculated in Razor Risk from the current market price of each issue. The MTM factor can be used to calibrate the pricing of the individual
issue to the price predicted by the spread curve. Credit VaR (CVaR) is then calculated based upon the historical volatility and correlation of each curve.
Credit rating migration can also be modelled in Razor Risk to capture price changes and any losses resulting from migration of credit ratings. Razor Risk
simulates the rating migration behaviour of each curve based upon transition probabilities. If a rating migrates to default, then a recovery model can
be incorporated into the calculation to determine any losses. This approach provides a very sophisticated mechanism to calculate specific risk.
STANDARD MODEL
Razor Risk can be configured to support the standard model calculation of specific risk as specified in the ‘Basel II Market Risk Framework’,
2010. With the standard model, specific weightings are applied against positions in individual issues to calculate the risk. Razor Risk can be
configured to determine the appropriate weighting based upon the standard model requirements.
7. Drill-down and What-if Functionality
Razor Risk provides extensive reporting and drill-down functionality. The following screenshots provide examples of the information that is
available through either the Razor Risk client applications or external applications.
THE FOLLOWING SCREENSHOT ILLUSTRATES THE OUTPUT FOR A VAR DISTRIBUTION.
From this histogram, it is possible to select a single scenario of interest and to view the trade contributions for that scenario.
From this frame, the user may select an individual trade contribution of interest and view the details of the selected trade in the deal entry screen.
8. TRANSACTION LEVEL DRILL-DOWNS
Razor Risk provides extensive drill-down capability to the transaction level. The contribution of each trade at the transaction level can be viewed
in either tabular or graphical form. Figure 3 below provides an example of the transaction level drill-down supported by Razor Risk.
FIGURE 3 - RAZOR RISK TRADE CONTRIBUTIONS
It is also possible to drill-down into each individual scenario to view the market data changes associated with each scenario and the contribution
of each trade to the results of the scenario (Figure 4 and Figure 5).
FIGURE 4 - PERTURBED INPUT POINTS FOR SCENARIO 11 IN 1 DAY MCVAR
9. FIGURE 5 - FOR ANOTHER SCENARIO, THE GENERATED DISCOUNT FACTORS FOR SCENARIO 6 FOR 1 DAY MCVAR.
For any scenario based methodology, Razor Risk provides the ability to drill-down on the analysis. For example, for Monte Carlo-based CVaRthe
user can view the simulation of each individual risk factor (Figure 6).
FIGURE 6 – RISK FACTOR PROCESS
10. For MCVaR or HSVaR, the details of how the analysis derives the exposure can be evaluated, for an example see Figure 7.
FIGURE 7 – DISTRIBUTION ANALYSIS
GREEKS
Greeks are calculated within Razor Risk as part of the scenario analysis. The scenario analysis functionality enables sensitivity scenarios to be
defined for any input market data including spread curves for credit delta calculation. Razor Risk displays the scenario analysis results in a high
dimensional form for each portfolio depending on the hierarchy requirements and the sensitivity results. This produces a multi-dimensional
output set that can easily be aggregated by any selected dimension to report the required risk measure (Figure 8).
FIGURE 8 - RAZOR RISK SCENARIO REPORTING
11. FIGURE 8 - RAZOR RISK SCENARIO REPORTING (CONT’D)
RAZOR RISK WHAT-IF FUNCTIONALITY
Razor Risk provides extensive what-if capability, which is used in conjunction with the trade level drill-down. Razor Risk allows the user to view all data,
with and without the effect of the current what-if scenario and to switch between the two views or to view the before and after side-by-side. The effect
of the what-if scenario on limits can be presented in tabular form as absolute exposures or as the changes between exposures. It can also be shown
graphically, via before and after graphs, or as composite graphs showing the net effect of the scenario. The user can enter a single deal, construct a
structured deal using individual trade components, or enter a number of independent deals as part of a single scenario.
Multiple what-if scenarios may be evaluated in parallel.
Each what-if scenario may consist of changes to portfolio composition and/or changes to market data. This provides support for what-if
analysis to assess the impact on VaR from a range of possible changes in today’s market rates.
EACH SCENARIO CAN BE CONSTRUCTED FROM:
Inserting new deals, either by manual entry or by copying and amending an existing deal
Amending existing deals
Cancelling existing deals
The effect of the what-if scenario is applied to all portfolios in the system and is not constrained to any one portfolio. All calculations are performed
in real-time and include the full processing of the historical/Monte Carlo simulation giving rise to the full range of portfolio effects. Individual what-
if scenarios can be named and saved and restored at any time by the user. These scenarios can also be shared between users of the system so that
several users can collaborate in the generation and analysis of a single scenario. The user can load multiple scenarios and switch between them at will,
as well as copy and paste data between scenarios. All drill-down and what-if results can be exported from Razor Risk to external applications as either
Extensible Markup Language (XML) or can simply be copied into Microsoft Excel.
12. INCREMENTAL PROCESSING
Razor Risk’s incremental processing enables the impact of any deal amendments to be quickly modelled in the system without having to perform time
consuming reruns of VaR results. The simulated valuations calculated by the pricing servers are stored within each portfolio aggregation server across
all paths and all aggregation nodes. When evaluating incremental trades, only the new trade(s) are priced and the results are re-aggregated with the
existing portfolio, and the new exposure is calculated. Razor Risk does not have to re-calculate the entire portfolio when evaluating the effects of new
trades on an existing portfolio. This approach significantly reduces the manpower and time required to produce accurate daily VaR results.
Scenario Analysis
SCENARIO ANALYSIS FUNCTIONALITY
Full scenario analysis and stress testing functionality is provided as part of Razor Risk’s scenario analysis functionality. Razor Risk’s scenario analysis
provides full access to Razor Risk’s market data set. Stress tests are defined by applying changes to the market data set and applying these changes
against the portfolio (Figure 9). These stress tests are then applied as part of the turn-of-day process, or run intra-day as required. There is no limit on
the number and complexity of stress tests that can be defined. Razor Risk’s pricing models also output additional trading risk analytics, such as the
Greeks, PV01, modified duration, Macaulay duration, convexity, clean price and accrued interest.
Market Risk Workflow
Razor Risk provides extensive support to enable risk managers to quickly isolate sources of VaR or other market risk discrepancies to reduce the amount
of time and effort that is required to produce accurate, validated market risk results. Razor Risk enables VaR results from any portfolio aggregation to be
compared against a previously defined baseline to assist in isolating any changes in VaR. Baseline changes are shown in a graphical form and support
full drill-down to determine the source of the VaR changes (Figure 9 and Figure 10).
FIGURE 9 - TODAY’S PARTIAL RISK BREAKDOWN
FIGURE 10 - COMPARISON TO YESTERDAY’S BASELINE
13. STRESS TESTING GRANULARITY
Stress tests can be run at any portfolio level, in the same manner as VaR or Credit Risk calculations. Full drill-down into the stress test results is
supported down to the transaction level.
SHOCKS CAN BE APPLIED AS EITHER:
Factor
Factor by which to multiply the current market
Absolute
Additive/subtractive value to add to the current market
Outright
A specific value to be set (current market has no impact).
Additional analyses can also be performed on the results of the stress test.
FIGURE 11 - RAZOR RISK SCENARIO EDIT FRAME
A canned set of stress tests can be defined and applied to the current market data set in line with the Group of Thirty (G30) standards. The frequency at
which these stress tests can be run can also be specified (e.g., daily, weekly, etc.). Historical stress tests based on specific historical events can also be
set up (e.g., the ERM crisis, stock market crashes, etc.).
User-defined stress tests are fully-supported in Razor Risk. As well as stressing any of the input market data, it is also possible to change parameters of
the Monte Carlo simulation to determine the impact on portfolio risk (e.g., modelling changes to mean reversion or changes to volatility or correlation
of input rates).
All stress tests in Razor Risk are defined and stored in the Razor Risk database. To conduct a stress test, the name of the stress test and the portfolio(s)
against which it should be run are specified. Stress tests can be run either as part of turn-of-day process or intra-day, as required. The stress tests are
run in the same manner as any other type of risk analysis in Razor Risk. Stress tests can be automatically run at the same time as a VaR or credit risk
analysis so there is no additional operational effort associated with performing stress tests in Razor Risk. Different stress tests can also be scheduled
to run on a periodic basis (e.g., regulatory stress tests once a week internal stress tests on a daily basis, etc.).
Limits can be set on stress tests in exactly the same manner that limits can be set on VaR or credit risk results. Razor Risk’s Limit Management module
is used to define the limits and to monitor limit utilisation. If utilisation exceeds the available limit then an excess is raised.
All stress test results can be exported from Razor Risk to external applications as either XML, or cut and paste into Microsoft Excel.
14. SENSITIVITY ANALYSIS
There is extensive support in Razor Risk for sensitivity analysis across all classes of market risk factors. Some examples are set out in Figures 12 through
16 as follows]:
FIGURE 12 — PV01
FIGURE 13 – FX SENSITIVITY
16. FIGURE 16 – PORTFOLIO GREEKS – SUPPORT ALSO EXISTS FOR PORTFOLIO VEGA AND RHO
Extendibility
The analytics of Razor Risk’s Market Risk module are fully extendible so a client can incorporate their own analytics.
AREAS OF EXTENDIBILITY INCLUDE:
Incorporation of a client or a third party’s own pricing models
Extension of product coverage by providing new pricing models and additional screens
Inclusion of proprietary term structure generation routines
Aggregation of results externally to razor risk and aggregated into the overall calculation.
Liquidity and Funding Risk
Razor Risk’s liquidity risk functionality enables the risk for all daily portfolio cash flows, including coupons, to be measured and reported. This
information can be broken down by currency, and limits can be set on daily cash flow amounts. Razor Risk can also easily be extended by the client
to incorporate a proprietary liquidity model. Some of our clients have adopted this approach to incorporate their own proprietary liquidity model into
Razor Risk.
17. FIGURE 17 – BACK-TESTING IN RAZOR RISK
FIGURE 18 – BACK-TESTING IN RAZOR RISK – EXCESS EVENTS
Back-Testing
Back-testing is performed in Razor Risk in line with the ‘Sound Practices for Back Testing Counterparty Credit Risk Models’ Dec 2010 and ‘Basel II
Market Risk Framework’ Dec 2010. Razor Risk calculates theoretical P/L which is then compared against the relevant VaR results on a daily basis.
Any exceptions are recorded and highlighted graphically by displaying a different colour. Razor Risk also supports back-testing against actual
P/L. The actual P/L results are fed into Razor Risk from an external system, and Razor Risk compares the VaR results against the actual P/L. Figure
17 below shows the graphical and tabular representation of back test results for a particular portfolio in Razor Risk. The theoretical and actual
P/L are shown plotted against the 99% 1 day VaR for the portfolio.
The magnitude and frequency of excesses is reported in Razor Risk, and back-testing can be performed for any portfolio in the system such as
business centre or desk. Expected tail gains and losses are calculated as part of the VaR analysis. This information is plotted in Razor Risk and
can be easily extracted into Microsoft Excel.
18. FIGURE 19 – BACK-TESTING IN RAZOR RISK – FULL SET OF RESULTS
19. About TMX Technology Solutions Inc.
TMX Technology Solutions Inc. is a leading provider of risk management technology and consulting solutions
to financial institutions worldwide. We provide successful solutions to proactively measure and manage risk
in the Americas, Europe and Asia.
With offices in Sydney, Toronto, New York and London, TMX Technology Solutions has a highly skilled team
of specialists who provide risk management technology and consulting services across the financial markets
and risk management sectors. We operate on a global risk consultancy structure, drawing upon the expertise
of all employees in implementing best practices for our clients’ individual needs. This methodology supports
an efficient, low cost, minimal risk implementation, allowing our clients to maximise optimal risk and
reward. TMX Technology Solutions has a 100 per cent successful implementation record for Razor Risk.