`by Chris Bright­man, Research Affil­i­ates

Oliver Wen­dell Holmes’ 1858 poem “The Deacon’s Mas­ter­piece”¹ describes a per­fected one-horse “shay,” a highly engi­neered car­riage designed so that the fail­ure of a sin­gle part could not cause an untimely break­down. By elim­i­nat­ing the weak­est links, the car­riage per­forms flaw­lessly, at first. But the shay does not have a happy end­ing. It sud­denly dis­in­te­grates with all the parts fail­ing at once, leav­ing its rider dazed atop a pile of rub­ble. Holmes—the father of the emi­nent U.S. Supreme Court Jus­tice— mocked the pseudo-scientific efforts of the overe­d­u­cated Dea­cons of his day to engi­neer imprac­ti­cal structures.

In our domain, the Dea­cons are quants (finan­cial engi­neers) and their Mas­ter­piece is an overly com­plex quan­ti­ta­tive invest­ment strat­egy. The sec­ond week in August marks the four-year anniver­sary of the quant melt­down of 2007. While the events of 2008, includ­ing nation­al­iza­tion of Fan­nie Mae and Fred­die Mac, the fail­ure of Lehman, the bailout of AIG, cre­ation of TARP, etc., have rightly received more scrutiny, August 2007 fore­shad­owed the global finan­cial cri­sis and deserves more atten­tion by today’s investors. Ana­lyz­ing the under­ly­ing causes of the quant melt­down helps reveal the per­ils of com­plex quan­ti­ta­tive strate­gies and high­lights the dif­fer­ence between trans­par­ent and rules-based alter­na­tive beta strate­gies such as the Fun­da­men­tal Index® method­ol­ogy and newer opti­mized approaches.

The Quant Melt­down
Dur­ing the week of August 6, 2007, many large and pre­vi­ously suc­cess­ful hedge funds were forced to de-lever their port­fo­lios and liq­ui­date com­monly held secu­ri­ties, result­ing in
simul­ta­ne­ous draw­downs of 30%, 50%, or worse. To make mat­ters worse, these invest­ments had been sold as risk-controlled and uncor­re­lated to the mar­ket. Khan­dani and Lo con­cluded that a “… deadly feed­back loop of coör­di­nated forced liq­ui­da­tions lead­ing to dete­ri­o­ra­tion of col­lat­eral value took hold dur­ing the sec­ond week of August 2007, ulti­mately result­ing in the col­lapse of a num­ber of quan­ti­ta­tive equity market-neutral man­agers, and double-digit losses for many oth­ers.”² Quan­ti­ta­tively man­aged enhanced index funds expe­ri­enced sim­i­lar simul­ta­ne­ous trau­mas, though the mag­ni­tude of losses was lower due to the lack of leverage.

None could have fore­cast the pre­cise tim­ing of the sud­den liq­ui­da­tion of a large trad­ing desk that cat­alyzed the quant melt­down.³ But should we have been sur­prised that those funds failed cat­a­stroph­i­cally? After all, the quant funds of 2007 shared the same struc­tural flaws as the highly engi­neered finan­cial trad­ing strate­gies that caused the stock mar­ket crash in 1987 and the implo­sion of Long-Term Cap­i­tal Man­age­ment in 1998.⁴

Inside the Black Box
To help avoid future melt­downs in our port­fo­lios, we need to look inside the black box of quant strate­gies. Sim­ply put, quants use advanced sta­tis­ti­cal meth­ods and high fre­quency data to cre­ate com­plex finan­cial mod­els. With expe­ri­ence, skill, and some luck, a few of these mod­els suc­cess­fully fore­cast future secu­rity price changes. In the short term, these strate­gies pro­vide con­sis­tent trad­ing prof­its and gather assets into asso­ci­ated funds. Con­sis­tent prof­its can hide inher­ent risks, how­ever. Most com­plex quant strate­gies have proven to be unsta­ble. Mar­kets evolve in response to the cre­ation and adop­tion of these strate­gies. At first, the iden­ti­fied pre­dictabil­ity in secu­rity price move­ments is rein­forced as funds using the quant model, along with sim­i­lar funds using sim­i­lar mod­els, begin buy­ing and sell­ing the same secu­ri­ties. Early suc­cess and clever mar­ket­ing attracts large flows into the funds, which, in turn, dri­ves the prices of secu­ri­ties held by these funds to unsus­tain­able extremes. The result is a brit­tle price struc­ture await­ing the inevitable crisis.

Lever­age cre­ates an even more toxic brew. In the years lead­ing up to the quant melt­down in August 2007, the same mod­els used to man­age enhanced index funds (with rel­a­tively low track­ing errors and high infor­ma­tion ratios) were increas­ingly employed to cre­ate lev­ered absolute return-oriented long/short funds. To facil­i­tate the use of lever­age, risk mod­els were used to min­i­mize coun­try, sec­tor, and other com­mon fac­tor risks. With all the risk seem­ingly wrung out of the strat­egy, ever more cap­i­tal and lever­age were applied.

Para­dox­i­cally, quan­ti­ta­tive risk man­age­ment was part of the prob­lem. While risk mod­els are use­ful tools for mea­sur­ing risk, using mod­els to tightly con­trol risk is mis­guided and dan­ger­ous. Because no model is, or ever can be, a com­plete descrip­tion of the com­plex dynamic sys­tem that is a mar­ket, all risk mod­els fail to cap­ture some risk. By elim­i­nat­ing all of the risks mea­sured by their mod­els, the quants trans­ferred the risk in their funds into the areas their mod­els could not mea­sure and they did not understand.

Quant strate­gies pro­duce remark­able prof­its in the early stages. But inevitably, the process becomes unsta­ble and often ends with vio­lent illiq­uid­ity events, such as the stock mar­ket crash of 1987, the Long-Term Cap­i­tal Management-induced cri­sis in Sep­tem­ber 1998, and the quant melt­down in August 2007. The largest losses in those episodes were suf­fered by the most recent investors who were attracted by daz­zling early per­for­mance records. Instead of con­sis­tent prof­its, the later investors were stuck with shock­ing losses real­ized dur­ing fund liq­ui­da­tion as investors fled from the implod­ing strategies.

As Harry Markowitz stated in the mid­dle of the cri­sis, “…the lay­ers of finan­cially engi­neered prod­ucts… com­bined with the high lev­els of lever­age, proved to be too much of a good thing.”⁵

Fun­da­men­tal not Quant Only four years after the last quant melt­down, over-engineered quan­ti­ta­tive invest­ment strate­gies are back. The lat­est incar­na­tion is com­plexly opti­mized alter­na­tive betas. Such strate­gies attempt to engi­neer indices with the low­est pos­si­ble volatil­ity, the high­est pos­si­ble Sharpe ratio, or the max­i­mum pos­si­ble diver­si­fi­ca­tion. The more com­plex the engi­neer­ing, the bet­ter the model per­forms in the back­test. As investors begin to adopt such nar­row indices, early per­for­mance may be reward­ing. Fund inflows will cre­ate buy­ing and sell­ing pres­sure on the same nar­row set of secu­ri­ties. This pat­tern will cre­ate a brit­tle price struc­ture resem­bling the Deacon’s Mas­ter­piece and will set the stage for the next wreck.

Rec­og­niz­ing the trou­ble with quants, should we eschew quan­ti­ta­tive study of secu­rity price move­ments and aban­don risk mod­els? Of course not! Advanced sta­tis­ti­cal meth­ods are invalu­able tools to help us under­stand secu­ri­ties mar­kets. Like­wise, risk mod­els help us mea­sure, mon­i­tor, and decom­pose the risks in our port­fo­lios. For exam­ple, with regard to the Fun­da­men­tal Index method­ol­ogy, we use quan­ti­ta­tive meth­ods to demon­strate how and why com­pa­nies with low mar­ket prices rel­a­tive to fun­da­men­tal mea­sures of com­pany size pro­vide higher returns than com­pa­nies with high mar­ket prices rel­a­tive to fun­da­men­tals. We use risk mod­els to exam­ine whether and how value priced com­pa­nies have dif­fer­ent risk char­ac­ter­is­tics than other companies.

The Fun­da­men­tal Index method­ol­ogy is far less com­plex and there­fore less risky than a highly engi­neered quant model. Fun­da­men­tal weights are sim­ple, log­i­cal, and sta­ble. Fun­da­men­tal Index port­fo­lios are trans­par­ently con­structed and broadly diver­si­fied. The Fun­da­men­tal Index strat­egy uses the time-tested tech­nique of sys­tem­atic rebal­anc­ing to cap­ture the long-term return pre­mium offered by the market’s excess volatility.

The fol­low­ing pas­sage from Holmes’ poem descries the end of the one-horse shay. But it could eas­ily be a fit­ting nar­ra­tive to the quant strate­gies dur­ing that fate­ful week in August 2007.

“…it went to pieces all at once, —
All at once, and noth­ing first, —
Just as bub­bles do when they burst.
End of the won­der­ful one-hoss shay.
Logic is logic. That’s all I say.”

The per­for­mance of Fun­da­men­tal Index strate­gies may break down occa­sion­ally over the long wind­ing road to invest­ment suc­cess, just as tra­di­tional index funds can cre­ate some nasty sur­prises. How­ever, these set­backs are just that and even­tu­ally the Fun­da­men­tal Index strategy’s sim­ple and sta­ble rebal­anc­ing process puts the port­fo­lio back on track. That’s our logic. What do you say?

End­notes
1. Oliver Wen­dell Holmes, 1890, The Deacon’s Mas­ter­piece or The Won­der­ful “One-Hoss Shay”: A Log­i­cal Story, New York: Houghton, Mif­flin and Com­pany. Illus­tra­tions by Howard Pyle.
2. Amir E. Khan­dani and Andrew W. Lo, 2007, “What Hap­pened to the Quants in August 2007?” Jour­nal of Invest­ment Man­age­ment, vol. 5, Fourth Quar­ter
3. Khan­dani and Lo, 2007.
4. Richard Book­staber, 2007, A Demon of Our Own Design: Mar­kets, Hedge Funds, and the Per­ils of Finan­cial Inno­va­tion, New York: Wiley.
5. Harry Markowitz, 2008 “The Father of Port­fo­lio The­ory on the Cri­sis,” Wall Street Jour­nal, Novem­ber 3. http://online.wsj.com/article/SB122567428153591981.html?mod=djemEditorialPage

Copy­right © Research Affil­i­ates

Read more from the author/contributor here.

Comments

Comments are closed.