30
Yin Luo, CFA Vice Chairman Quan.ta.ve Research, Economics, and Por8olio Strategy [email protected] Miguel Alvarez [email protected] Javed Jussa [email protected] Sheng Wang [email protected] Gaurav Rohal, CFA [email protected] QES Desk Phone: 1.646.582.9230 [email protected] Luo’s QES Quan?ta?ve Research, Economics, and PorDolio Strategy October 2017 DO NOT FORWARD – DO NOT DISTRIBUTE – DOCUMENT CAN ONLY BE PRINTED TWICE This report is limited solely for the use of clients of Wolfe Research. Please refer to the DISCLOSURE SECTION located at the end of this report for Analyst Cer.fica.ons and Other Disclosures. For Important Disclosures, please go to www.WolfeResearch.com/Disclosures or write to us at Wolfe Research, LLC, 420 Lexington Avenue, Suite 648, New York, NY 10170

Luo’s QES Miguel Alvarez Javed Jussa Quan?ta?ve …com.estimize.public.s3.amazonaws.com/l2q presentations/5_l2q_ldn...2 #1 Ranked Quant & Macro Research Team Yin Luo, CFA, CPA Vice

  • Upload
    vancong

  • View
    227

  • Download
    7

Embed Size (px)

Citation preview

YinLuo,CFAViceChairmanQuan.ta.veResearch,Economics,[email protected]@wolferesearch.comJavedJussaJJussa@[email protected],[email protected]:[email protected]

Luo’sQES

Quan?ta?veResearch,Economics,andPorDolioStrategy

October2017

DONOTFORWARD–DONOTDISTRIBUTE–DOCUMENTCANONLYBEPRINTEDTWICEThisreportislimitedsolelyfortheuseofclientsofWolfeResearch.PleaserefertotheDISCLOSURESECTIONlocatedattheendofthisreportforAnalystCer.fica.onsandOtherDisclosures.ForImportantDisclosures,pleasegotowww.WolfeResearch.com/DisclosuresorwritetousatWolfeResearch,LLC,420LexingtonAvenue,Suite648,NewYork,NY10170

2

#1RankedQuant&MacroResearchTeam

YinLuo,CFA,CPAViceChairmanQES

JavedJussaDirectorofQuan.ta.veResearch

MiguelAlvarezHeadofInvestmentSolu.ons,RiskandPor8olioConstruc.on

YinLuojoinedWolfeResearchinSeptember2016,asaViceChairmantoleadQESresearch.PriortoWolfeResearch,YinspentsevenyearsasaManagingDirectorandGlobalHeadofQuan.ta.veStrategyatDeutscheBank.BeforeDeutscheBank,hespentover12yearsininvestmentbankingandmanagementconsul.ng.

Javedisresponsibleforalphasignal,BigData,ESG,andsmall-capresearchandmanagingtheday-to-dayopera.onsoftheQESteam.PriortoWolfeResearch,JavedwastheUSHeadofQuan.ta.veStrategyatDeutscheBank.Javedalsohasseveralyearsofexperienceininvestmentbusinessandtechnologyconsul.ng.

Miguelisresponsibleforrisk,afribu.on,por8olioconstruc.on,andinvestmentsolu.ons.PriortoWolfeResearch,hewastheUSHeadofQuan.ta.veStrategyandInvestmentSolu.onsatDeutscheBank.MiguelalsoworkedattheEMinvestmentteamatBGI(nowBlackRock).HebeganhiscareerintheresearchgroupatBarrawherehewasresponsibleforriskmodelandpor8olioconstruc.onresearch.

ShengWangVP

Kar?kArora,PhDHeadofQESInfrastructure

GauravRohal,CFAVPHeadofClientServices

Shengistheteam’sexpertonmachinelearningandglobalstockselec.onmodels.BeforejoiningWolfeResearch,ShengwasinchargeoftheR&DofDeutscheBank’sglobalstockselec.onmodelsusingasuiteofsophis.catedmachinelearningtechniques.

Kar.kistheHeadofQESInfrastructure,responsiblefortheoveralltechnologyinfrastructureandconsul.ngservicestoclients.Kar.khasmanyyearsofexperienceinbothinvestmentresearchandtechnology,includingfiveyearsatDeutscheBank’sglobalquan.ta.veresearchinfrastructureteam.

GauravRohalisinchargeofourclientservices.BeforejoiningWolfeResearch,GauravspentfiveyearsatDeutscheBank’sQuan.ta.veResearchteamintheUSandAsia.Priortothat,GauravworkedataHFTproptradingdeskandataglobalinvestmentbankasaquantforexecu.onalgorithmsanddarkpools.

ZhaoJinAssociate

JasonZhong,PhDAssociate,GlobalMacro

•  MumbaiResearch&TechnologyTeam•  SydneyTechnologyTeam

ZhaoJinispartofthesystema.cequityresearchteam.PriortoWolfeResearch,ZhaoworkedasanAVPatBankofAmericaMerrillLynch,developingtheriskcalcula.onpla8ormforthemortgageteam.Zhaoalsospent.meatSungard,focusingonpor8oliomanagementsonware.

Jasonspecializesinmacroeconomicsandglobalmacroresearch.BeforejoiningWolfeResearch,JasonspentfiveyearsintheDepartmentofQuan.ta.veHealthSciencesintheClevelandClinic,developingtheframeworkofrisk-adjustedoutcomesrepor.ng.

•  #1inQuan.ta.veResearch(II-America,II-Europe,II-Asia)

•  ToprankedinPor8olioStrategyandAccoun.ng&TaxPolicy

•  Theprobabilityoftheleadingdigitbeinga1isnot1/9(11.1%)butrather30%,basedontheBenford’slaw.Surprisingly,wefindalmosteverysinglefinancialstatementlineitemfitstheBenford’slawperfectly.

3Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

NumericalPaOerns

Salesleadingdigitdistribu?on ConformitytoBenfordlawforstandardaccoun?ngitems

NumberofemployeesinUSpubliccompanies Theore?calorexpecteddistribu?on

0

100

200

300

400

500

600

700

1 2 3 4 5 6 7 8 9

Num

bero

fEmployees:Distributionof

FirstD

igit

Thou

sand

s

FirstDigit

0%

5%

10%

15%

20%

25%

30%

1 2 3 4 5 6 7 8 9

Theo

reticalDistributionofFirstD

igit

FirstDigit

Benford'sLaw(TheLaw ofFirstDigit)

0

100

200

300

400

500

600

1 2 3 4 5 6 7 8 9

Sales:DistributionofFirstD

igit

Thou

sand

s

FirstDigit

0%

20%

40%

60%

80%

100%

SALES COGS TotalAssets CashFlowfrom

Operations

120StandardAccounting

Items

Distrib

utionofFirstD

igit

1

2

3

4

5

6

7

8

9

4Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

TheMADScien?st

VisuallyconformingtotheBenford’sdistribu?on:Disney Visuallynon-conformingtoBenford’sdistribu?on:Enron

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

1 2 3 4 5 6 7 8 9

Freq

uencyDistrib

utionofFirstD

igit

FirstDigit

Disney Theoretical

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

35.0%

1 2 3 4 5 6 7 8 9

Freq

uencyDistrib

utionofFirstD

igit

FirstDigit

Enron Theoretical

Higher numbersaremorefrequentthanthenatural

distribution

•  Therearetwosta.s.cs–theMAD(MeanAbsoluteDevia.on)andtheKolmogorov–Smirnov(KS)sta.s.c.•  MADsimplycomputesthecumula.veabsolutedevia.onbetweenthecompany’sandtheactualdistribu.on

(AD)versusthetheore.calorexpecteddistribu.on(ED):MAD= (∑1↑𝑘▒|𝐴𝐷−𝐸𝐷| )/k 

•  TheKSsta.s.csistypicallyusedtocomparethesimilarityoftwodistribu.ons.Itcomputesthemaximum

absolutedifferencebetweentheactualandexpecteddistribu.on:𝐾𝑆=max{…,…|(𝐴𝐷↓1 + 𝐴𝐷↓2 )−(𝐸𝐷↓1 + 𝐸𝐷↓2 )|,…,|(𝐴𝐷↓1 +…𝐴𝐷↓9 )−(𝐸𝐷↓1 +…𝐸𝐷↓9 )|}

•  WecomputetheMADandKSfactorsforallcompaniesinourinvestmentuniverse,usingabout120accoun.ngitemsfromthebalancesheet,incomestatementandcashflowstatement.Thenon-conformersaredefinedasthebofom5%ofcompanies.

5Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

Irregulari?esUnderperform

Coverageofnon-conformingfirms Sharpera?ocomparison

Cumula?veperformancebasedontheKSmodel Cumula?veperformancebasedontheMADfactor

0

5

10

15

20

25

Wealth

($)

KSEquallyWeighted Russell3000EquallyWeighted

Non-conformingcompaniessharplyunderperformthe

market

0

5

10

15

20

25

Wealth

($)

MADEquallyWeighted Russell3000EquallyWeighted

Non-conformingcompaniessharplyunderperformthe

market

Non-conformingcompaniessharplyunderperformthe

market

0

500

1,000

1,500

2,000

2,500

3,000

3,500 Non-Conformers Russell3000

0.00

0.20

0.40

0.60

0.80

EquallyWeighted CapWeighted

Sharpe

Ratio

KS Russell

6Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

GlobalPerformance

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

Wealth

($)

Index(marketcapwgt) Shortportfolio(McapweightedKStop5%)

0.0

2.0

4.0

6.0

8.0

10.0

12.0

Wealth

($)

Index(equalwgt) Shortportfolio(equalweightedKStop5%)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Wealth

($)

Index(equalwgt) Shortportfolio(equalweightedKStop5%)

0.0

2.0

4.0

6.0

8.0

10.0

12.0

Wealth

($)

Index(equalwgt) Shortportfolio(equalweightedKStop5%)

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0Wealth

($)

Index(marketcapwgt) Shortportfolio(McapweightedKStop5%)

0.0

2.0

4.0

6.0

8.0

10.0

12.0

Wealth

($)

Index(marketcapwgt) Shortportfolio(McapweightedKStop5%)

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

Wealth

($)

Index(marketcapwgt) Shortportfolio(McapweightedKStop5%)

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0Wealth

($)

Index(equalwgt) Shortportfolio(equalweightedKStop5%)

Europe UK Japan

AsiaexJapan Canada Australia/NewZealand

LATAM EMEA

•  PubliccompaniesintheUSfilealmost5000documentseveryday.Thevastmajorityoftheavailableinforma.onisinunstructuredformats,e.g.,text,audio,video,andimage.

•  Theunderwhelmingperformanceoftradi.onalfactorsandtherapiddevelopmentincompu.ngpowerandmachinelearningmeansprocessingunstructuredinforma.ontogenerateusefulnumericalsignalsbecomesincreasinglyimportant.

7Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,EDGAR,WolfeResearchLuo’sQES

WebScrapingUnstructuredTextualInforma?on

NumberofEDGARfilings(daily) Numberoffilings(formtype)

Numberof10-KfilingsaroundtheyearNumberof10-Qfilingsaroundtheyear

•  Transformingunstructuredinforma.onintonumericdatainreal.merequiresasuiteofintegratedsystem,fromwebscraping,datacollec.on,distributedparallelcompu.ng,advancedNaturalLanguageProcessing(NLP),tomachinelearningtechniques.

•  Cloudcompu.ngproviderssuchasAmazon,Microson,andGooglehavedemocra.zedaccesstothedistributedcompu.ng.•  Weu.lizeHadoopframeworkforperforminglargescaledatamining.EDGARwebsiteprovidesmasterindexfilesthatareusedto

iden.fyrelevantfilingdocumentsandloca.on.Weparse,storeandsinthroughthesefilingsforrelevantqualita.veinforma.on.Thefocusishowtobestquan.fydescrip.vetextualdocumentsintoinvestmentintelligence.

8Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

Map-ReduceframeworkfortextminingtheSECEDGARwebsite

Map-Reduceframeworkdataflowschema

•  Interes.ngly,weseeaplungeintheNettoneofthe“RiskFactors”sec.onduringtheheightofthe2008FinancialCrisis.•  Thiscoincideswiththeintroduc.onofmassivenumberofnewtextualdescrip.onsinthispar.cularsec.onfromOctober2008

toApril2009.Thesenewwordshavepredominantlynega.vetones,expressingtheconcernsabouttherecessionanditsimpactonthecompanyfinancials.

•  Thesharpchangeinthelanguageduringthefinancialcrisisquicklyrevertedbacktobusinessasusualinthesubsequentyears,albeitwithhigherwordcount.

9Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

Sen?mentandtoneanalysis

Wordcountandnetsen?mentforthe“RiskFactors”sec?onofthe10-Qfilings

-0.11

-0.11

-0.10

-0.10

-0.09

-0.09

-0.08

0

50

100

150

200

250

300

350

Netse

ntim

ent(med

ian)

Wordcoun

t(med

ian)

WordCount NetSentiment(rightaxis)

Massivechangeinlanguagewithseverely negativesentiment,duringtheheightoffinancialcrisis.

10Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

Historicalperformanceofthenega?vesen?mentfactorsusing10-Qfilings

Quin?lereturnsofthenega?vesen?mentfactor Sharpera?oofthenega?vesen?mentfactor

Thenega?vesen?mentfactor(Consolidated),rankICThenega?vesen?mentfactor(MarketRisk),rankIC

11Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

Historicalperformanceofthedistancefactorusingthe10-Kfilings

Quin?lereturnsoftheJaccarddistancefactor Sharpera?ooftheJaccarddistancefactor

Sharpera?ooftheCosinedistancefactorQuin?lereturnsoftheCosinedistancefactor

•  Inregulatoryfilings,companiescanpresenttheirperformanceandbusinessusingeithertextualdescrip.onornumericaldata.Arguably,whenthenumbersareweak,firmsmightafempttodistractinvestors’afen.on.

•  Inordertocapturethisphenomenon,wecomputethepercentageofnumericdataembeddedineachsec.onofthefilings.•  Formostsec.ons,firmswithhigherpercentagesofnumericcontentshavehighersubsequentreturns.

12Sources:BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

NumericversusTextualPropor?on

Sharpera?oofthenumericpercentagemeasure MD&Asec?on,non-sectorneutralrankIC

MD&Asec?on,sectorneutralrankICMarketRisksec?on,non-sectorneutralrankIC

•  Trafficdetec.onviasatellitesisacomplicatedprocess.Thechartbelowprovidesanoverviewofthevariousstagesinvolvedinanalyzingtrafficdataforretailers.

•  Firstofall,geographicimageryiscapturedfromglobalsatellites,aerial/airplanephotography,anddrones.Imagesareprocessedandessen.allydigi.zedforfeatureextrac.on.Basedontheimagery,variousmodelsaswellasgeoloca.ondatabasesareusedtoisolateroadsandparkinglots.Next,toolsandsonwareprogramsareusedtodetectvehicleswithinparkinglotsandroadways.Lastly,vehiclefeaturesuchassize,color,type(e.g.,car,truck,passengervan)canbeextracted.

13Sources:WolfeResearchLuo’sQES

Satellite101

Satelliteimageprocessing

•  HighEarthorbitorgeosynchronousorbit(GEO)satelliteshoveratanal.tudeabove35,000kilometers.Manyweathersatellitesfallintothiscategory.MediumEarthorbitorMEOtendtostayintherangeof2,000to35,000kilometers.Themostcommonuseforsatellitesinthisregionisfornaviga.onsuchasGPS.LowEarthorbitorLEOsatelliteshoveratarangeof200to2,000kilometers.Thesesatellitesareusefulforobserva.on,scien.ficexplora.onaswellasmilitaryandreconnaissancepurposes.

•  TheheightoftheorbitordistancebetweenthesatelliteandtheEarth’ssurfacedeterminehowquicklythesatelliterotatesaroundtheearth.Asasatellitegetsclosertotheearth,thegravita.onfieldgetsstronger,causingthesatellitetomovequickeraroundtheearth.Therefore,LEOsspinaroundtheearthatamuchgreatervelocitythanGEOs.ALEOcanorbittheearthinanhourwhereasitmaytakeaGEOmorethan24hours.

•  Ahighearthorbitsatellitethatisapproximately36,000kilometersfromtheearth’ssurface(or42,000kilometersfromthecenteroftheearth)matchestherota.onalspeedoftheearth.Thisiswhythesesatellitesarereferredtoasgeosynchronousorgeosta.onary.Thesesatellitesareusefulforcommunica.onbecausegroundsta.onsatelliteslocatedontheearthdonotneedtorotatetotrackthem.Incontrast,LEOsatellitestendtobeusefulforobserva.on.ImagestakenbyLEO’stendtohavehigherresolu.ons.

14Sources:hfps://www.e-educa.on.psu.edu/geog480/node/444

Al?tude&Rota?on

TypesofSatellites

•  ApolarorbitsatellitepassesaboveorneartheNorthandSouthPolesduringitsrota.on.ThesesatellitesaremostlyLEO’sapproximately700kmabovetheearth’ssurface.Ittakesabout90minutesforthesatellitetocompleteoneorbitaroundtheEarth.Polarsatellitestypicallycoveralongandwideregionsincetheyrotatequicklyaroundtheearth.Theyaretypicallyusedforremotesensingandtrafficimagery.Thedisadvantageofapolarorbitalsatelliteisthatthesameregionontheearth’ssurfacecan’tbesensedcon.nuously,becausethesatelliterotatesaroundthepoles,whiletheearthrotatesarounditsaxis

•  Sun-synchronoussatellitestendtopassoveranygivenla.tudeatalmostthesamesolar.meeachday.Simplyput,thesetypesofsatellitestendtopassoveranygivenpointontheplanet’ssurfaceatthesamelocal.me.Suchanorbitplacesasatelliteinconstantsunlight(onthesunnysideoftheearth)andisusefulforimagingandremotesensing.The.meistypicallybetweenmid-morningandmid-anernoononthesunsideoftheorbit.Theycancapturetheearth’scanvasatroughlythesame.meduringeachpass,sothatligh.ngremainsuniform.Thisenablesimagestobecomparableover.me.

•  Thereareseveraladvantagestosun-synchronoussatellitesinanearpolarorbit.Thelowal.tudepermitshighresolu.on,whichispreferredforimagingandremotesensing.Thepolarorbitallowsforalargemosaicswathofdailyimagingcoverage.Mostearthobservingsatellitemissionsusesun-synchronoussatellitesinlownearpolarorbits.Thesetypesofsatellitesareveryusefulforcardetec.on.

15Sources:hfp://tornado.sfsu.edu/geosciences/classes/m415_715/Monteverdi/Satellite/PolarOrbiter/Polar_Orbits.htm

Orbit

Sun-synchronousorbit

•  Spa.alresolu.onreferstothepixelsizeofsatelliteimagerycoveringtheearth’ssurface.Forexample,aspa.alresolu.onof30mmeansthesmallestunitthatmapstoasinglepixelwithinanimageisapproximately30mx30m.Itisapparentthatahigherresolu.onenablesbeferimageprocessingandfeaturedetec.on.

16Sources:hfp://www.eorc.jaxa.jp/ALOS-2/en/img_up/alos2_1st/pal2_1s.mg_20140619-21.htm

Resolu?on

Varyingimageresolu?on

•  Satelliteimagerycanbeobtainedaspanchroma.c(greyscale),naturalcolor(RGB)aswellasmul.spectralbands.Mul.spectralbandscancapturelightbeyondthevisiblelightfrequencies,allowingforextrac.onofaddi.onalinforma.onthathumaneyefailstocapture.Forthepurposesofthisreport,wefocusongreyscaleandnaturalcolors(RGB)wavelengths

•  Forthepurposesofvehicledetec.on,greyscaleisasufficientandthepreferredcolorband.Formanyapplica.onsofimageprocessing,colorinforma.ondoesnothelpusiden.fyimportantedgesorotherfeatures.Byusinggreyscale,wecan,ineffect,reducethesignaltonoisera.o.Conver.ngtogreyscalealsoreducesthedimensionalityofimageprocessing

•  𝐺𝑟𝑒𝑦𝑠𝑐𝑎𝑙𝑒=0.299𝑟+0.58𝑔+0.114𝑏

17Sources:hfps://www.e-educa.on.psu.edu/geog480/node/444

ImagePreprocessing

ColorBands

Sources:hfps://www.e-educa.on.psu.edu/geog480/node/444

RGBtogreyscaleconversion

•  Mul.plethresholdingallowsformul.plebreakpoints.Itreliesoncolortodis.nguishanobjectfromthebackground,whichcanbeproblema.c.Forexample,adarkcoloredvehiclemaybemisclassifiedasthebackground.

•  Bayesianmodelscananalyzethevariousfeaturesofvehiclessuchaschanginggradients,windows,shadows,curvatureetc.Amodelistrainedonwhethertheexistenceofthesefeaturescancorrectlyclassifyavehicle.Sincemachinelearningsmodelsrelyonfeaturesratherthanjustthevehiclecolor,theycanbefarmoreaccuratethantradi.onalmethods.

18Sources:DOI:10.4236/jfs.2014.42015

Mul?pleThresholdingandBayesianNetworks

Thresholding+Cleansed Final

NaturalColor Thresholding

•  Featureextrac.onismoreadvancedandinteres.ngthansimplyrecognizingaparkinglot.Algorithmscanpoten.allyiden.fyqueuesbypixelforma.onshavingsufficientlength,boundedwidthandlowcurvature.Queuesalsoshowarepe..vepafernalongacenterline,bothincontrastandwidth.

•  Differen.a.ngbetweencarsandtruckscanalsobeimplementedsystema.cally.Thebasiclogicistocomputethepixelwidth,length,andareaofeachvehicle.Therearearangeoftechniquesavailabletodetectcertainvehiclefeatures.

•  Insummary,theprocessoftakingsatelliteimageryanddecipheringtrafficpafernsisinteres.ngyetcomplex.Itcanalsobeexpensiveandlaborintensive.Thankfully,datavendorsspecializinginsatelliteimagerycanprovidethetransformedandstructureddata.Inthenextsec.on,wetakeadeepdiveintotheRSMetricsdatasetandseehowitcanbeusedinstockselec.on.

19

Sources:hfp://www.walrusvision.com/wordpress/otsu-thresholder-algorithm-works/

FeatureExtrac?on

Queuefeature Largevehiclefeatureextrac?onusingMeanShifclustering

Sources:hfp://content.iospress.com/ar.cles/journal-of-intelligent-and-fuzzy-systems/ifs2201

•  Fill_Rate=Estimate of number of cars for a company at a point in time/Estimate number of available car spaces for a company at a point in time 

20Sources:RSMetrics

FillRate

Stripcenters Powercentersoroutletmalls

Standaloneretailloca?onwithsquarelot Standaloneretailloca?onwithunconven?onallot

•  Asexpected,Saturdays’havethehighestfillrates,asconsumersventureoutontheweekendshopping.Interes.ngly,Sunday’sfillratesarelowerthanwehadan.cipated,possiblyduetolateopeninghoursonSundays.

•  Thegeographicbreakdownsoffillratescoincidewiththereligiouspar.cipa.onratesorreligiositybystate(seeFigure39).Weseetheweakestreligiouspar.cipa.onratesintheWestcoastwherefillratesarethehighestonSundays.Incontrast,theSouthhashighestreligiouspar.cipa.onrates,resul.nginthelowestfillratesonSundays.

21Sources:Alba-Flores,R.[2005].“Evalua.onoftheUseofHigh-Resolu.onSatelliteImageryinTransporta.onApplica.ons”,FinalReport

FillRatebyDaysofTheWeek

Averagefillratebyregionanddayoftheweek(DOW)

Religiositybystate

•  Theretailbusinessisseasonal.Itisdifficulttogaugethehealthofretailcompaniesbasedondataoverasinglemonth.Rather,performanceshouldbemeasuredonasustained,persistentandconsistentmanner.

22Sources:RSMetrics,BloombergFinanceLLP,FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

PersistentTrafficGrowth

Quin?lereturnperformance Quan?leSharpeperformance

Coverage Long/ShortPerformance

23

M&ATargets

AverageCumula?veExcessReturns MedianCumula?veExcessReturns

FactorExposureBeforeandAfertheM&AAnnouncement AdjustedM&AFrequencybySector

Sources:FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

Valuefactors•  Moreexpensivebasedon

dividendyield,earningsyield,cashflowyield,tangiblebook-to-market,EBITDA/EV;but

•  Cheaperbasedonprice-to-sales,book-to-market,andrevenue/TEV

Qualityfactors•  Theyarelessprofitableonmost

metrics,withtheexcep.onofgrossprofitmargin.Aposi.vegrossmarginandanega.venetmarginmeanstargetfirmsarepar.cularlyinefficientingeneralmanagementandadministra.veac.vi.es.

•  Theyhavelowercorporategovernancestandardbutpossiblehigherdividendpayingsustainability,asreflectedbythelowerpayoutra.o.

•  Theyhaveslightlylowerfinancialleverageandlowerbankruptcyrisk

24

Sources:FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

Tradi?onalStock-Sec?onFactors

A) Average M&A IC B) Risk Adjusted M&A IC

A) Average M&A IC B) Risk Adjusted M&A IC

Qua

lityFactors

ValueFactors

Sen.mentfactors•  Takeovercompaniesaredislikedbysell-side

analystsbyallcommonmeasures.

Growthfactors•  Allthegrowthrelatedfactorsshownega.ve

M&AIC.•  Onemajorreasonthatanacquirerwantsto

buyatargetistoturnaroundaslowgrowthcompany,inordertogeneratemergersynergyandoutsizedprofit

TechnicalFactors•  Theyhavepoorpricemomentum.•  Theyhavelowerliquidity,e.g.,float

turnover,Amihudilliquidity.•  Theyaremorevola.le.•  Themostinteres.ngaspectisthattarget

companiestendtorallysharplyimmediatelybeforeM&Aannouncements(posi.veone-monthreturn),coupledbyhigherabnormaltradingvolume,andexcessKurtosis.ThisispossibleduetothefactthatM&Atransac.onsaresome.mesan.cipatedbycertainmarketpar.cipants.

25Sources:FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

Tradi?onalStock-Sec?onFactors,Cont’dA) Average M&A IC B) Risk Adjusted M&A IC

A) Average M&A IC B) Risk Adjusted M&A IC

TechnicalFactors A) Average M&A IC B) Risk Adjusted M&A IC

GrowthFactors

Sen?

men

tFactors

•  Form3andForm4areSECfilingsthatrelatetoinsidertrading.Everydirector,officerorownerofmorethan10%ofaclassofequitysecurityregisteredmustfilewiththeSECastatementofownershipregardingsuchasecurity.Theini.alfilingisonForm3andchangesarereportedonForm4.

•  Forunderperformingcompaniesbutexpec.ngtobeacquired,insidersmayengagemoreac.vetransac.onoftheirownstocks.

•  Thereareheavierthannormalinsidertransac.onsfromtwoyearspriortotheannouncementdateun.lsixmonthsrightbeforethedealsareannounced.

•  Thenumberofinsidertradesisactuallymuchlighteronequarterbeforetheannouncementdate,possiblybecauseeitherlock-outperiodpreventsinsidersfromtradingrightbeforetransac.ons,orinsiderswanttoavoidbeingperceivedastradingonprivateinforma.on.

26Sources:FTSERussell,S&PCapitalIQ,ThomsonReuters,EDGAR,WolfeResearchLuo’sQES

EDGARFilingSignals

NumberofFilingsforForm4andForm3

AverageM&AICforFilingMonthsPrior

A) Form 4 B) Form 3

0

500

1000

1500

2000

2500

0

100

200

300

400

500

600

A) Form 4 B) Form 3

•  TheSchedule(SC)13Disaformthatmustbefiledwhenapersonorgroupacquiresmorethan5%ofanyclassofacompany'sshares.Thisinforma.onmustbedisclosedwithin10daysofthetransac.on.

•  SC13GissimilartoSC13Dusedtoreportaparty'sownershipofstockthatisover5%ofthecompany.SC13Gisshorterandrequireslessinforma.onfromthefilingparty.TobeabletofileSC13GinsteadofSC13D,thepartymustownbetween5and20%inthecompany.Thepartyacquiringthestakeinthecompanymustonlybeapassiveinvestoranddoesnotintendtoexertcontrol.

27Sources:FTSERussell,S&PCapitalIQ,ThomsonReuters,EDGAR,WolfeResearchLuo’sQES

SC13DandSC13G

NumberofFilingsforSC13D,SC13D/A,SC13GandSC13G/AA) Form SC 13D B) Form SC 13D/A

C) Form SC 13G B) Form SC 13G/A

RiskAdjustedM&AIC–SC13D,SC13D/A,SC13G,SC13G/A

AverageM&AICforDifferentLagsA) Form SC 13D B) Form SC 13D/A

C) Form SC 13G B) Form SC 13G/A

•  Aswemovefroma80-factormodeltoa235-factormodel,predic.vepowerimproves.The80factorsareasubsetofthe235signals.•  Wecanfurtherboostperformancebyaddingnon-tradi.onalfactors(i.e.,EDGARfilingbased,eventcountsignals,andoursectorM&A

momentumfactor).

28Sources:FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

AHorseraceofMachineLearningAlgorithmsinRareEventPredic?on

IncreasingtheNumberofTradi?onalQuantFactor

AddingNon-tradi?onalFactorsfurtherBoostsPerformance

A) Average M&A IC B) Risk adjusted M&A IC

A) Average M&A IC B) Risk Adjusted M&A IC

29Sources:FTSERussell,S&PCapitalIQ,ThomsonReuters,WolfeResearchLuo’sQES

AvoidShor?ngPoten?alTakeoverTargets

RemovingHigh-Takeover-ProbabilityStocks(basedonSMAP)BoostsPerformance

Cumula?vePerformance,ROEFactorPorDolio ImprovingthePerformanceofLEAPusingtheSMAPModel

A) Annualized Return B) Sharpe Ratio

C) Annualized Volatility D) Max Drawdown

•  Weseeacrosstheboardperformanceimprovementbyremovingthehighesttakeoverprobabilitystocksfromtheshortside.

•  Bynotshor.ngpoten.alM&Atargetsalsoreducestheriskforallcommonfactors.

•  RemovinghightakeoverprobabilitystocksfromtheshortsidealsohelpshighefficacyalphamodelssuchastheLEAP.

A) Sharpe Ratio B) Max Drawdown

30

DISCLOSURE SECTION AnalystCer?fica?on:TheanalystofWolfeResearchprimarilyresponsibleforthisresearchreportwhosenameappearsfirstonthefrontpageofthisresearchreportherebycer.fiesthat(i)therecommenda.onsandopinionsexpressedinthisresearchreportaccuratelyreflecttheresearchanalysts’personalviewsaboutthesubjectsecuri.esorissuersand(ii)nopartoftheresearchanalysts’compensa.onwas,isorwillbedirectlyorindirectlyrelatedtothespecificrecommenda.onsorviewscontainedinthisreport.OtherDisclosures:WolfeResearch,LLCdoesnotassignra.ngsofBuy,HoldorSelltothestocksitcovers.Outperform,PeerPerformandUnderperformarenottherespec.veequivalentsofBuy,HoldandSellbutrepresentrela.veweigh.ngsasdefinedabove.Tosa.sfyregulatoryrequirements,OutperformhasbeendesignatedtocorrespondwithBuy,PeerPerformhasbeendesignatedtocorrespondwithHoldandUnderperformhasbeendesignatedtocorrespondwithSell.WolfeResearchSecuri.esandWolfeResearch,LLChaveadoptedtheuseofWolfeResearchasbrandnames.WolfeResearchSecuri.es,amemberofFINRA(www.finra.org)isthebroker-dealeraffiliateofWolfeResearch,LLCandisresponsibleforthecontentsofthismaterial.AnyanalystspublishingthesereportsareduallyemployedbyWolfeResearch,LLCandWolfeResearchSecuri.es.Thecontentofthisreportistobeusedsolelyforinforma.onalpurposesandshouldnotberegardedasanoffer,orasolicita.onofanoffer,tobuyorsellasecurity,financialinstrumentorservicediscussedherein.Opinionsinthiscommunica.oncons.tutethecurrentjudgmentoftheauthorasofthedateand.meofthisreportandaresubjecttochangewithoutno.ce.Informa.onhereinisbelievedtobereliablebutWolfeResearchanditsaffiliates,includingbutnotlimitedtoWolfeResearchSecuri.es,makesnorepresenta.onthatitiscompleteoraccurate.Theinforma.onprovidedinthiscommunica.onisnotdesignedtoreplacearecipient'sowndecision-makingprocessesforassessingaproposedtransac.onorinvestmentinvolvingafinancialinstrumentdiscussedherein.Recipientsareencouragedtoseekfinancialadvicefromtheirfinancialadvisorregardingtheappropriatenessofinves.nginasecurityorfinancialinstrumentreferredtointhisreportandshouldunderstandthatstatementsregardingthefutureperformanceofthefinancialinstrumentsorthesecuri.esreferencedhereinmaynotberealized.Pastperformanceisnotindica.veoffutureresults.Thisreportisnotintendedfordistribu.onto,oruseby,anypersonoren.tyinanyloca.onwheresuchdistribu.onorusewouldbecontrarytoapplicablelaw,orwhichwouldsubjectWolfeResearch,LLCoranyaffiliatetoanyregistra.onrequirementwithinsuchloca.on.Foraddi.onalimportantdisclosures,pleaseseewww.wolferesearch.com\disclosures.TheviewsexpressedinWolfeResearch,LLCresearchreportswithregardstosectorsand/orspecificcompaniesmayfrom.meto.mebeinconsistentwiththeviewsimpliedbyinclusionofthosesectorsandcompaniesinotherWolfeResearch,LLCanalysts’researchreportsandmodelingscreens.WolfeResearchcommunicateswithclientsacrossavarietyofmediumsoftheclients’choosingincludingemails,voiceblastsandelectronicpublica.ontoourproprietarywebsite.Copyright©WolfeResearch,LLC2017.Allrightsreserved.Allmaterialpresentedinthisdocument,unlessspecificallyindicatedotherwise,isundercopyrighttoWolfeResearch,LLC.Noneofthematerial,noritscontent,noranycopyofit,maybealteredinanyway,ortransmifedtoordistributedtoanyotherparty,withoutthepriorexpresswrifenpermissionofWolfeResearch,LLC.ThisreportislimitedforthesoleuseofclientsofWolfeResearch.Authorizedusershavereceivedanencryp.ondecoderwhichlegislatesandmonitorstheaccesstoWolfeResearch,LLCcontent.Anydistribu.onofthecontentproducedbyWolfeResearch,LLCwillviolatetheunderstandingofthetermsofourrela.onship.