“ WHAT ARE YOU REALLY GETTING?” ( SUMMARY DISCUSSION) SEATTLE LOS ANGELES Jeffrey J. MacLean 999 Third Avenue 2321 Rosecrans Avenue S enior Consultant Suite 4200 Suite 2250 Seattle, Washington 98104 El Segundo, California 90245 206.622.3700 telephone 310.297.1777 telephone 206.622.0548 facsimile 310.297.0878 facsimile
Overview of Discussion • It is always wise t o reflect upon best pract ices following significant capit al market s disrupt ions • Quant itat ive st rat egies have proliferat ed in last t went y years • They had difficult ly during t he recent market downt urn, most not ably August 2007 • Excess ret urns have been st eadily declining in recent periods • Analysis should be t heoret ical and philosophical in nat ure • Past ret urns alone t ell us lit t le t o not hing about t he fut ure, but help us underst and maj or risk fact ors • Fundament al forces driving success should be ident ified, underst ood and planned for accordingly • Barriers t o success should be heeded and avoided when appropriate • There is never an “ answer” t o any st rat egy • The goal is t o creat e a prism t hrough which t o dimension opport unities and st rat egies • Not creat e a black or whit e opinion on t hem • We humbly offer our philosophical views 2
General Types of Quantitative Models Quant s buy and sell securit ies based on deviat ions from t heoret ically expect ed pat t erns of behavior • Trading Models • Analysis of past ret urns t o predict t he fut ure • Generally short t erm orient ed • Moving averages, Oscillat ors, and Channel Breakout s t o name a few • Valuat ion Based S creening & Const ruct ion • Balance sheet and valuat ion analysis; public dat a only; no management int erviews • Hold syst emat ically t ilt ed port folio relat ive t o benchmark • Can be comput er/ model driven, or “ black box” • Avoid pit falls of human j udgment by eliminat ing it from decision; no mosaic t heory • Volat ilit y & Correlat ion Opt imizat ion • Focus is on st at ist ical met rics; st andard deviat ion, variance, co-variance, correlat ion, et c. • Oft en comput er/ model driven, or “ black box” • Goal is t o opt imize port folio for superior risk adj ust ed ret urns 3
Barrier: Efficient Market Hypothesis (EMH) All act ive managers are const rained by EMH in some form, but only quant s by all t hree • Weak EMH • Fut ure prices cannot be predict ed by past price pat t erns • Hist oric analysis offers no guidance t o fut ure pricing or behavior • No persist ent or repeat able pat t erns of behavior t o generat e “ alpha” • Bad for t rading st rat egies and models based on hist oric ret urns • S emi-S t rong EMH • Market s rapidly adj ust t o new public informat ion • Discovery “ alpha” is available, but finit e • Bad for managers t hat rely on public dat a; i.e., valuat ion-based const ruction • Allows “ alpha” for t hose using subj ect ive j udgment t o discern privat e informat ion; i.e., mosaic t heory • S t rong EMH • Market reflect s all informat ion, public and privat e • Bad for all act ive management ;“ alpha” cannot be found Not e: Theory pert ains t o “ alpha,” not Bet a-Alpha (you can beat an index wit h differing bet as) 4
Barrier: The Crowding Out Effect Number of Quant. Mgrs. as % eVestment US Large Cap Universe (Sept. 89 - June '09) 30% • There are only so many ways t o find “ alpha” through quantit at ive t ools 25% • S uch t ools are widely available 20% • People wit h quant it ative expert ise are common 15% • S ecret methods t o uncover alpha are unlikely t o remain 10% secret for long – model evolut ion is key 5% • Logically speaking, firms only employing quant itative met hodologies will compet e for finit e “ alpha” S ource: eVest ment Alliance; Wurt s & Associat es • Finit e alpha & more managers = less alpha t o go around Average Median Quarterly Gross Excess Return for US Large Cap Quant Mgrs to Preferred Benchmark (Sept. 89 - June 2009) 0.35% • Evidence loosely support s t heory of “ Crowding Out Effect ” 0.30% Average Quarterly Excess Return 0.30% • Conclusion: The number of managers employing 0.24% 0.25% quantit at ive st rat egies limit s pot ent ial “ alpha” 0.19% 0.20% 0.15% 0.10% 10% -15% 15% -20% 20% + % of Quant Mgrs in eVestment US Large Universe S ource: eVest ment Alliance; Wurt s & Associat es 5
Barrier: “ Goldilocks” Volatility Average Median Quarterly Excess Return for US Large Cap Quant Mgrs to Preferred Benchmark vs. Quarterly Change in VIX Index • Recall quant managers t rade on deviations from (March 1990 - June 2009) 0.40% t heoretically expect ed pat t erns of behavior 0.34% Average Quarterly Excess Return • Wit hout deviat ions t here is not hing t o t rade upon 0.30% • Imagine t wo ext remes: 0.22% 0.20% • No volat ilit y = no opport unit y 0.20% 0.17% • Infinite volatilit y = t heoretical relationships break down & no opport unity 0.10% 0.0-1.0 1.0-2.5 2.5-5.0 5.0+ • Hence quant s needs somet hing in bet ween Range of Quarterly Change in VIX S ource: eVest ment Alliance; Wurt s & Associat es • Evidence loosely support s concept of “ Goldilocks” volat ilit y at market level • Note some quant s require int ra-st ock volatilit y; or volat ilit y wit hin a given index • Conclusion: Implement ation of quantitative st rat egies rest upon cert ain volat ilit y assumpt ions t o produce “ alpha” 6
S ome Practical Barriers to S uccess Benchmark Const raints • Given EMH, volatilit y, and crowding out concerns, the more you Quantitative Fundamental const rain a manager’s st rategy, the more difficult it is achieve Median Strategies Strategies success Expense Ratio 0.50% 0.70% • Most t radit ional quant st rat egies are highly const rained R-Squared 94% 85% Active Share 22% 30% • Most alt ernat ive quant st rat egies are unconst rained Active Expense Ratio 2.2% 2.1% S our ce: eVest ment Al l iance, Wur t s & Associat es Fees • Act ually expensive for t he act ively managed port ion • Another barrier t o success given finite “ alpha” and benchmark const raint s for t radit ional managers Crowding Out Conundrum • Managers with highest likelihood for success have the most secret and difficult t o replicat e models, or are const ant ly changing t hem • S ecret models impede due diligence and formulation of forward looking forecast s • Must be aware of and willing t o accept t his conundrum 7
S ummary Maj or Risk Fact ors Associat ed Wit h Quant Managers • EMH port ends inability of quant s t o produce “ alpha” solely wit h public dat a, but not bet a-alpha • Quant derived “ alpha” is finit e; number of quant managers affect s aggregat e opportunity set • Cert ain volat ilit y condit ions are necessary for success • Benchmark const raint s limit abilit y t o overcome barriers t o success • Opaque models complicat e due diligence and forward looking analysis • Fees can be high on an act ively managed basis Increasing Chances of S uccess • Combat “ crowding out effect ” by allocating t o t hose wit h highly dynamic and evolving st rategies, and be willing t o accept associat ed lack of t ransparency • Relax benchmark const raint s t o allow opportunit y for beta-alpha and maximize efficiency of fees, or lower fees on highly const rained mandat es • Fairly evaluate performance in relation t o opportunit y set, benchmark const raint s, fees, and market volatility, not j ust past ret urns t o avoid poor manager t iming decisions • Above all, be aware risk fact ors and barriers t o success for quant itat ive st rat egies are different from fundamental 8
Recommend
More recommend
