Scribd
Upload
191 views

WTI Futures Curve Analysis Using PCA (Part I)

In this white paper, we will not delve into the theoretical economics behind the price changes or their spreads. Instead, we will examine the daily prices of the first four (4) contracts of WTI CL futures listed on NYMEX. Next, using exchange rules for WTI/CL contract trading, we will compute the number of days to the delivery month for each contract to construct the futures curve. Finally, we will carry out principal component analysis (PCA) in an attempt to uncover the core drivers behind the futures curve changes (i.e. level and general shape). For the data spreadsheet used in this paper, visit http://bitly.com/1aTSUUR

Uploaded by

NumXL Pro
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
191 views

WTI Futures Curve Analysis Using PCA (Part I)

In this white paper, we will not delve into the theoretical economics behind the price changes or their spreads. Instead, we will examine the daily prices of the first four (4) contracts of WTI CL futures listed on NYMEX. Next, using exchange rules for WTI/CL contract trading, we will compute the number of days to the delivery month for each contract to construct the futures curve. Finally, we will carry out principal component analysis (PCA) in an attempt to uncover the core drivers behind the futures curve changes (i.e. level and general shape). For the data spreadsheet used in this paper, visit http://bitly.com/1aTSUUR

Uploaded by

NumXL Pro
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

PCA

of WTI Futures (Part I)


Theoretically,crudeoilfuturepricesreflectthemarketparticipantsexpectationoffuturedemandand supply,aswellastheiroveralluncertainty. Thecrudeoilfuturemarketisaninterestingmarkettoanalyze.Lawsofcostofcarry,supplyand demandstillapply,butgeopoliticalriskweighsonrelativeprices. Historically,theoilfuturescurveisoftenfoundinbackwardation,whichmeanshigherpricesforshort termcontractsthanforlongtermcontracts.Thisisoftenexplainedbyatheoreticaltermcalled convenienceyield.Convenienceyieldisconceptuallysimilartodividendsinequity,whereitfavors physicalpossessionofthestockoverfuturedeliveryduetothedividendcashpayments.Inthecrudeoil market,convenienceyieldmaysignalmarketworryonfutureoilsupply(ordelivery),duetosome geopoliticalconcernsandthetendencytofavorholdingthecommoditynow.

Inthiswhitepaper,wewillnotdelveintothetheoreticaleconomicsbehindthepricechangesortheir spreads.Instead,wewillexaminethedailypricesofthefirstfour(4)contractsofWTICLfutureslisted onNYMEX.Next,usingexchangerulesforWTI/CLcontracttrading,wewillcomputethenumberofdays tothedeliverymonthforeachcontracttoconstructthefuturescurve.Finally,wewillcarryoutprincipal componentanalysis(PCA)inanattempttouncoverthecoredriversbehindthefuturescurvechanges (i.e.levelandgeneralshape).

Why should we care?


Theoilfuturemarketisverycomplexinitsdesign,and,inthispaper,wewillattempttouncoverand simplifytheunderlyingdriversreflectedinthedailyrelativepricesofdifferentcontractsforabetter understandingandbetterhedgingforaportfolioofsuchinstruments.

Background
Thegeneraldemandforpetroleumproductsishighlyseasonalandisgreatestduringthewintermonths, whencountriesintheNorthernHemisphereincreasetheiruseofdistillatedheatingoilsandresidual fuels.Supplyofcrudeoil,includingbothproductionandnetimports,alsoshowsasimilarseasonal variationbutwithasmallermagnitude. Duringthesummermonths,supplyexceedsdemandandpetroleuminventoriesnormallybuild;whereas duringthewinter,demandexceedssupplyandinventoriesaredrawndown.Asaresult,inventoriesalso demonstrateseasonality. Intheory,futurespricesarecomputedasfollows: PCAforWTI/CLFutures

Ft ,T St e

( rt ,T xt ,T qt ,T )(T t )

SpiderFinancialCorp,2013

Where

Ft ,T =futurepricesattime( t )fordeliveryat T St =WTIspotpricesfordeliveryatCushing,OK t =timenow T =futuredeliverytime rt ,T =nominal(perannum)interestrateattime( t )for T maturity


xt ,T nominal(perannum)marginalstoragecostattime( t )for T delivery
=

qt ,T =nominal(perannum)theoreticalconvenienceyieldattime( t )for T delivery

Now,letstakethelogarithmofeachside:

ln( Ft ,T ) ln( St ) (rt ,T xt ,T qt ,T )(T t )

Tocarryoutouranalysis,wewillusethelogarithmoffuturepricesandincludethelogoftheWTIspot pricesintothedataset. Next,wewillcomputethenetoftheinterestrate,storageandconvenienceyieldrates(i.e. t ,T ),which canbeexpressedasfollows:

t ,T rt ,T xt ,T qt ,T

ln( Ft ,T / St ) T t

Notethat t ,T theoreticallyconsistsofthreelooselycorrelatedfactors(interest,storageand convenienceannualyield),sowedexpectthatapplyingaPCAtypeofanalysisshouldyieldnomore thanthree(3)factors.

Data Preparation
Inthispaper,wewillusetheclosingmarksoftheimmediatefour(4)tradedNYMEXCLfuturecontracts oftheEIAwebsite.Furthermore,wealsousethespotpricesforWTIcrudeoilatCushing,OK(delivery locationforNYMEXCLcontracts)oftheEIAwebsiteaswell. Tocompileourdataset,weusethenumberofdaystothe1stdayofthedeliverymonthasourhorizon (i.e.theindependentvariableofthefuturecurve).WerefertothisasdaystodeliveryorDTD. Next,accordingtoNYMEXproductspecification,thetradingofacrudeoilfuturecontractterminates baseonthefollowingrule(s):

PCAforWTI/CLFutures

SpiderFinancialCorp,2013

Tradinginthecurrentdeliverymonthshallceaseonthethirdbusinessdaypriortothetwenty fifthcalendardayofthemonthproceedingthedeliverymonth.Ifthetwentyfifthcalendarday ofthemonthisanonbusinessday,tradingshallceaseonthethirdbusinessdaypriortothelast businessdayproceedingthetwentyfifthcalendarday.IntheeventthattheofficialExchange holidayschedulechangessubsequenttothelistingofCrudeOilfutures,theoriginallylisted expirationdateshallremainineffect.Intheeventthattheoriginallylistedexpirationdayis declaredaholiday,expirationwillmovetothebusinessdayimmediatelyprior Usingthelasttradingdayrules,wedeterminewhenthefrontcontractswitchestothefollowingmonth contract,and,thus,computethepropertradingdaystothe1stdayofthedeliverymonth.For computingthetradingdays,adjustingforweekendsandholidays,weusedtheNumXLcalendar functionswiththeUSDcalendar. Asaresult,foreachtradingday,weusethefour(4)contractstoconstructafuturecurve(futureprices versusnumberofdaystodeliver(DTD)). Next,oneachday,usingthefuturecurveabove,weinterpolate/extrapolate(cubicspline)thefuture pricesfordeliverytermsrangingfrom10daysto120days(12terms). Next,usingtheformulabelow,wetransformthefuturepricesintothenetoftheinterestrate,storage costandconvenienceyield(i.e. t ,T )

t ,T rt ,T xt ,T qt ,T

ln( Ft ,T / St ) T t

Forexample,onApril29,2013,theWTIfuturecurveexhibitsahumpshapedcurve:

PCAforWTI/CLFutures 3 SpiderFinancialCorp,2013

Onthesameday,theimplied(computed)netinterestrate,storageandconvenienceyield(NISC)for eachdeliveryterm,the t ,T exhibitsthefollowingshape(graphbelow).

Althoughthefuturepricesbetween50100DTDremainsflat,theunderlyingnetofinterest,storageand convenienceyieldchangesduetothechangeintimetodelivery. Finally,wecomputetwelve(12)timeseriesforthenetinterest,storageandconvenienceyield(NISC)for deliverytermsrangingfrom10to120days.

Analysis
LetsfirstexaminethecorrelationbetweenthetwelveNISCinputtimeseries.

PCAforWTI/CLFutures 4 SpiderFinancialCorp,2013

Theshorttermdeliveries(<30days)oftheNISCcorrelateweaklywithlongertermsfutures.Notethat thisphenomenonisnotfoundintherawfutureprices.

Now,letsrunPCAanalysis.Beforewelaunchthewizard,insertarowabovetheinputdataforthemask variableandsetallitsvaluesto1.Thiswillhelpustoexcludeinputvariableswithoutredoingthe analysis.

LaunchthePCAWizard,specifyinputvariablesandcomputethePCAstatistics.

PCAshowsthatthefirsttwoprincipalcomponents(akadrivers)accountfor98.7%oftheoverall variation,andthefirstthreeprincipalcomponentscapture99.9%. Letsexaminetheloadingsofthosedriversinanattempttofindapractical/physicalproxyforthem.For thefirstprincipalcomponent:

PCAforWTI/CLFutures 5 SpiderFinancialCorp,2013

ThefirstPCloadings(akatermstructure)exhibitapatternsimilartotheyieldcurve:Contagoinshort term,andflatforlongerterm.Wemaythinkofthefirstcomponentasaproxyfortheinterestrate. Thesecondprincipalcomponents(akadriver)loadingexhibitsthefollowingpattern:

ThispatternissimilartothePC1,withtheexceptionofthekinkfor1020days,andthenegativevalues upto50days.Thismaybeassumedasaproxyfortheconvenienceyield;shorttermtenorshave negativevaluescausingthefuturepricestoriseandpossiblycreatingabackwardation.Forlongerterm tenors,thevalueispositive,reducingthefuturepriceandstrengtheningthebackwardation. Thethirdprincipalcomponentisrelativelyhardertoexplain:

PCAforWTI/CLFutures 6 SpiderFinancialCorp,2013

Canthisbethestoragecostperyear?Unlikely,astheloadinggoesnegativebetween20and70daysto delivery.Fortunately,itsvarianceandcontributiontotheoverallvariationarerelativelysmall.

Conclusion
Insum,wefoundthatthenetinterest,storageandconvenienceyield(NISC)ofWTIfuturesareprimarily drivenbytwouncorrelateddrivers.Thefirstdriverexhibitsatermstructuresimilartotheyieldcurve andtheseconddriverwashypothesizedasaproxytotheconvenienceyield. Wait a minute! Youmaywonder:canIleverageainterestrateinstrument(e.g.Eurodollar,swaps,etc.)tohedgethe interestrateexposureinmyWTIfuturesportfolio? Inafollowuppaper,wewillexaminetheLIBORyieldcurvedataintoouranalysisandfinetuneourrisk driversfurther,isolatingthestorageandconvenienceyieldfromtheinterestrate. Why do we care? AportfolioofWTIfuturecontractscanbehedged(97.8%effective)fornonspotpricechangesusing onlytwo(2)differentfuturecontracts. Whataboutspotchanges? Whatisthehedgingration? Howoftendowerebalancethehedge?

Inafollowuppaper,welldiscussthehedginginrelationtoPCAinfurtherdetails. Why do we stop here? Thereisalotofmaterialheretoswallow,soweoptedtopauseatthisstagetogiveyouopportunityto digestandgetcomfortablewithourearlierdiscussion,andbetterprepareyouforamoreadvanced handlingofthetopic.

PCAforWTI/CLFutures

SpiderFinancialCorp,2013

You might also like