Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
HowGPUsEnableXVAPricingandRiskCalculationsforRiskAggregation
JamesMesney |PrincipalSolutionsEngineer|[email protected]
SettingtheScene
WhatisXVA?
3
X-Value Adjustment (XVA) refers to “Valuation Adjustments” in relation to derivative instruments held by banks.
The “X” in XVA means “C”forcredit,“D”fordebt,“F”forfunding,“K”forcapital…
“Doing”XVAisRiskModelling.
It’sallaboutcomputingpotentialRISK,nowandinthefuture.
OBJECTIVE:INSULATETHEBANKFROMRISKWHEREVERPOSSIBLEANDALLOCATETHERIGHTCAPITAL
WhyisXVANeeded?
4
Pre 2007 - Trades cleared at fair valuations • Costs for capital and collateral were irrelevant to
banks’ investment decisions
• Increasingly complex investment portfolios emerge of variable value
ElevatedCapitalisation+CollateralisedTrades=EXPENSE&REDUCEDPROFITABILITY!
HowcanbanksoperateefficientlyAND playbytherules?
ANSWER=LOTSOFDATA+CLEVERFORECASTINGMODELS,LOTS OFCOMPUTE
4
Post 2007 – MAJOR REFORM!! Counterparty Credit Risk and Basel III Accord
XVAChallenges
• ToComputeRisk• Continuallyandcomprehensivelymeasuretradingactivity,currencymovements
• Consistentlyandtimely– attradingspeed…Batchprocessingisnolongersatisfactory
• 100,000sormillionsoftradesperday• 10,000sofcounterparties… in20+currencies!
• Calculations– ComplexandTimeCritical• ComputeintensiveworkloadsarewellsuitedfortheGPU
• REALTIMEadjustmentcalculationsareverycomputationallyintensive
• MonteCarloSimulation
55
LifeAfterMoore’sLaw40YearsofMicroprocessorTrendData
1980 1990 2000 2010 2020
102
103
104
105
106
107
Single-threaded perf
1.5X per year
1.1X per yearTransistors(thousands)
Original data up to the year 2010 collected and plotted by M. Horowitz, F. Labonte, O. Shacham, K. Olukotun, L. Hammond, and C. Batten New plot and data collected for 2010-2015 by K. Rupp
SpecINT
Original data up to the year 2010 collected and plotted by M. Horowitz, F. Labonte, O. Shacham, K. Olukotun, L. Hammond, and C. Batten New plot and data collected for 2010-2015 by K. Rupp
1980 1990 2000 2010 2020
102
103
104
105
106
107
Single-threaded perf
1.5X per year
1.1X per year
GPU-Computing perf1.5X per year 1000X
By 2025
TheRiseofGPUComputing
SpecINT
TheCPUBottleneck
8
Withlimitedprocessinggainsonthehorizon,CPUsarefurtherandfurtherbehindthegrowthindata
8
GPUAccelerationOvercomesProcessingBottleneck
4,000+coresperdeviceversus~16coresper
typicalCPU
Highperformancecomputingtrendtousing
GPU’stosolvemassiveprocessingchallenges GPUaccelerationbrings
highperformancecomputetocommodityhardware
Parallelprocessingisidealforscanningentiredataset&bruteforce compute
9
DeployingXVA
XVAArchitecture
11
ETL/STREAMPROCESSING
ONDEMANDSCALEOUT+
Server1
SQL
NativeAPIs
PARALLELINGEST Export
CustomConnectors
In-DatabaseProcessing
SQLBIDMach
MLLibs
BIDASHBOARDS
BI/VISUALISATION
CUSTOMAPPS
KINETICA‘REVEAL’
BATCH&STR
EAMINGDA
TA
PythonJavaC++CUDA
Server'n'
LiveTradingdataCounterparties
OptionsCurrencies
FuturesMarket
ForeignExchangeBloomberg
Reuters
+DOWNSTREAMAPPS
XVAModel
Kinetica’sXVADeploymentatMajorMultinationalBank
12
Largefinancialinstitutionmovescounterpartyriskanalysisfromovernighttoreal-time.
• DatacollectedandfedtoXVAlibrarywhichcomputesriskmetricsforeachtrade
• XVAoutputsstoredinKineticadatabase
• Flexiblereal-timemonitoringbytraders,auditorsandmanagement
• Dataretainedforhistoricanalyses,machine-learning…
FinancialServices:AnIndustryInTransition
13
Threeformidableforces—aweakglobaleconomy,digitization,andregulation—threatento
significantlylowerprofitsbyasmuchas$90Bfortheglobalbankingindustryoverthenextthreeyears.
Source:McKinsey&Co,PwC
FinancialServicesenterprisesmustreinventtheirbusinessbytransformingthecore–
Resilience,Reorientation,Renewal
Kinetica for:Resilience– ManageRevenue,Costs,Capital,andRisksReorient– Customer-centricity,Digitization,OpenBank
Renewal- NewMarkets,Products,Customers
Kinetica:ADistributed,In-Memory,GPUDatabase
14
GPU-accelerateddatabaseoperations
Naturallanguageprocessingbasedfull-textsearch
NativeGISandIP-addressobject
support
Realtimedatahandlerstoingeststructuredand
unstructureddata
Deepintegrationwithopensourceandcommercialframeworksandapplications:TensorFlow,Hadoop,Spark,NiFi,Storm,Kafka,Tableau,
KibanaandCaravel
Predictablescaleoutfordataingestionand
querying
Notypicaltuning,indexing,andtweaking
Distributedvisualizationpipelinebuiltin
KineticaEnablingBroadEnterpriseSolutions
RETAIL/CPGOmni-Channel
CustomerExperienceSupplyChainOptimization
TargetedMarketing
UTILITIESSmartMeters
SmartGridOptimizationInfrastructureModernization
CROSSINDUSTRYReal-TimeAnalyticsConvergeAI&BI
Location-BasedAnalyticsIoT Analytics
FINANCIALSERVICESRiskModelingFinancialCrimesCompliance
CustomerExperience
HEALTHCAREDrugDevelopmentPrecisionMedicine
Patient360
MEDIA/ENTERTAINMENTSentimentAnalytics
RecommendationEnginesAdTargeting
COMMUNICATIONSCustomerChurn
NetworkOptimizationContentTargeting
TRAVELPriceOptimization
CustomerExperienceEquipmentMaintenance
16
Kinetica:UniqueStrengths&Capabilities
16
SuperchargeBI
TakingadvantageoftheparallelnatureoftheGPUKineticadeliverslow-latency,highperformanceanalyticsonlargeandsteamingdatasets
Simultaneouslyingest,explore,analyze,andvisualizedatawithinmillisecondstomakecriticaldecisions.
User-definedfunctions(UDFs)allowfordistributedcustomcompute
directlyfromwithinthedatabase.
Easiertoworkwithlargegeospatialdatasets.
Fast,DistributedDatabaseEngine
InDatabaseAnalytics
NativeGeospatial&VisualizationPipeline
JamesMesney |PrincipalSolutionsEngineer|[email protected]
Thank You!ComegetyourKineticat-shirtandcopyofthenewO’ReillybookatboothG.01!