量化投资解决方案研究

March 2026 Index 1. Team structure and R&D process p.2 2. The team 3. Selected publications p.25 —Equities—PrivateMarkets—Credit—Foreignexchange—InterestRates—Commodities—PortfolioConstruction—HedgeFunds—OptionsMarkets—RiskModelling,AssetAllocationandThematic—IntradayandShort-TermFrequencies—PerformanceMonitors—AcademicReviews 4. Team awards 1. QIS Research: Team Structure Caio Natividade,Team Head and Signal ResearchVivek Anand,Portfolio Construction and Risk ResearchAjay Hiralal-Shara,Quantitative DevelopmentWai Ip,Artificial Intelligence & Data Science ResearchPankil Parikh,Quantitative DevelopmentFrancesco Pozzetti,Options ResearchLuiz Silva,Systematic Global Macro ResearchGianpaolo Tomasi,Quantitative Equity ResearchAbi Zareei,Quantitative Equity Research 1. QIS Research: R&D process (2) 𝑆𝑅≈𝐼𝐶×𝑏𝑟𝑒𝑎𝑑𝑡ℎ New frontiers –New manifestations of old drivers–New drivers (i.e. COVID)–Alpha aggregation methods Tier 3 markets–CTA: 500 cross-asset: ETFs, options, All 3 axis are equally important in the alpha landscape. 1. QIS Research: Equities (1)Multi-style, multi-frequency, multi-data 1. QIS Research: Equities (2) ▪Researchon the topic of‘traditional style factors’sinceearly2010,leadingtogenerationsoffactorportfolios(LowBeta,Momentum,RF2QualityandRF2Value).▪‘Traditional’(RF2)ValueandQuality:sector-regionneutral,highlydiversified(6subfactorsforValue,5forQuality),equally-weighted(2017+).▪NewversionsofValueandQuality:accountforintangiblesanddonotpenalizegrowth-relatedcapex(‘Owner’Valueand‘Owner’Quality)(2020+).▪Momentum:built on idiosyncratic returns,long-short butwithCTA-inspiredmarketdirectionality.▪LowBeta:market-neutralityachievedbydeleveragingtheshortleg(2017+),and‘timed’usingvol-targeting(2024+).▪Reversion:PCA-basedmarketneutralreversion(2021+). Approachrefinedovertime:4publicationsfromDBQuantResearch(2012,2013,2014and2020). 6‘Kitchensink’approach:113globalstock-specificfeaturesspanningvarious styles(e.g.Value,Growth,Profitability,Momentum,BalanceSheet)arefirstcomputedtocreateanalphalibrary. ThesefeaturesarethenprovidedasinputtoanML-basedalgorithm,leadingintoacompositescore. Thesupervised learning exercise uses different trainingwindowstoincreasediversification. Capturesdiversifiedfactormomentum,aparticularfeatureofequitymarkets. Productimplementation available in long-short and long-onlyformat. 1. QIS Research: Equities (3) 1. QIS Research: Fixed Income MarketsThe cookbook approach: driversfactorsportfolios performance 𝑆𝑅≈𝐼𝐶×𝑏𝑟𝑒𝑎𝑑𝑡ℎ The Credit cookbook The Rates cookbook –Guide to style investing in investment gradeand high yield Credit using CDS instruments.–In-depth coverage of market drivers and howthey translate into style investing.–A diversified style approach: Momentum,Value, Low Beta, Quality and Low Duration.–Portfolio construction technology thatbalances between transfer coefficients andturnover cost.–Applications for smart beta and market neutralportfolios. –Accurate bucketing of interest rate marketsaccording to clusters, which map intoregions.–In-depth coverage of market drivers andhow they translate into style investing.–A diversified style approach: Momentum,Value, Low Beta and Carry.–Portfolio Construction based ondiversification and confidence sizing.–Applications for smart beta and marketneutral portfolios. 1. QIS Research: FX and Commodity MarketsThe cookbook approach: driversfactorsportfoliosperformance 𝑆𝑅≈𝐼𝐶×𝑏𝑟𝑒𝑎𝑑𝑡ℎ The Commodities cookbook The FX cookbook –In-depth coverage of market drivers and theircyclical and structural variations.–In-depth modelling of sector idiosyncrasies inthe asset class.–Introduced curve momentum as animplementable strategy. Explores the riskpremia rationale, linking it to supply & demanddynamics and the underlying macro variables. –In-depth coverage of market drivers andhow they translate into style investing.–A diversified style approach: Momentum,Value, Carry, Monetary Policy andPositioning.–Region-specific signal fine tuning due tostructural idiosyncrasies.–Applications for absolute returns-marketneutral and market directional.–Applications for currency overlayprogrammes. Predictive power per driver–panel regressions 1. QIS Research: Intraday CTAPattern hunting in intraday CTA markets 1. QIS Research: Options MarketsHedged vs unhedged, income vs protection 1. QIS Research: Portfolio Construction (1)Building convex and market neutral portfolios ▪Thoughtprocessinitiatedin2016andfinetunedovertheyears,withreferenceportfoliosavailableonBloomberg.▪Designedtoperformduringtailmarketswhilebeingcapitalefficientand cost efficient.Focus:convexity,bleed andbasis.▪Highlydirectional,time series based,volatile(capitalefficient)strategies.Seek as many defensive inputs aspossibletobenefitfromnon-linearcorrelations.▪Robustnesschecksforfalsepositivesandmacroregimedependencies.▪Designedfortail,henceestimateandcalibrateparametersusingnon-Gaussianmetrics.▪Modelunknownsthroughbootstrapsandscenarioanalysisba