The rise of DIY quants could spell the end of easy prof­its from com­puter-driven trad­ing

With cheap data and soft­ware, ma­chine-based in­vest­ing is open to any­one with the math chops “De­moc­ra­ti­za­tion of tools doesn’t nec­es­sar­ily mean… de­moc­ra­ti­za­tion of good judg­ment”

Bloomberg Businessweek (Asia) - - CONTENTS - −Dani Burger

Clay pot­tery dec­o­rates the halls of a thatched, sin­gle-story adobe home in the desert town of Las Cruces, N.M. Out back, where scrub brush stretches into the arid plain be­tween the moun­tains and the Rio Grande, is a 50-foot­tall wire­less In­ter­net tower.

Roger Hunter is in the kitchen, grind­ing hand-roasted coffee beans. The 66-year-old for­mer math pro­fes­sor turned in­vestor apol­o­gizes if he ap­pears a bit out of sorts. As chief tech­nol­ogy of­fi­cer of a two-man startup called QTS Cap­i­tal Man­age­ment— his part­ner lives in Canada—Hunter pulled an all-nighter fix­ing a sys­tems glitch.

It’s a long way from Wall Street, but for Hunter and in­vestors like him, be­ing apart from the fran­tic ac­tiv­ity of New York is an ad­van­tage. “This is par­tic­u­larly true when de­vel­op­ing code and ex­plor­ing new strate­gies,” he says. His strate­gies are all au­to­mated and might in­clude trades on any­thing from cur­ren­cies to hog fu­tures to op­tions on mar­ket volatil­ity.

With lit­tle more than open source soft­ware and an In­ter­net con­nec­tion, Hunter is one of a new breed of traders break­ing into quan­ti­ta­tive in­vest­ing. Quants, as they’re known, crunch a dizzy­ing amount of data from across global mar­kets and write pro­grams to trade on the pat­terns they spot in as­set prices. Pow­er­house firms such as AQR

Cap­i­tal Man­age­ment and Citadel have used fast com­put­ers and closely guarded al­go­rithms to try to beat the mar­ket, and an­a­lysts es­ti­mate the 40-year-old in­dus­try runs more than $1 tril­lion in as­sets. Now low-cost, high­pow­ered tech­nol­ogy is raz­ing vir­tu­ally ev­ery bar­rier to en­try.

The rise of DIY quants comes as the pro­lif­er­a­tion of ma­chine-based strate­gies has made it harder for tra­di­tional play­ers to suc­ceed. In Jan­uary, Martin Tay­lor of Nevsky Cap­i­tal closed his 15-year-old hedge fund, lament­ing the dis­tort­ing in­flu­ence of com­puter traders. But es­tab­lished quants, too, are feel­ing the heat. “Tech­no­log­i­cal edge is harder to come by be­cause the more egal­i­tar­ian th­ese tools have be­come, the more dif­fi­cult it is to come up with some­thing truly new,” says An­drew Lo, a fi­nance pro­fes­sor at MIT and

chair­man of Al­phaSim­plex Group, a quant re­search firm.

In other words, the growth in traders us­ing quant strate­gies also tends to di­min­ish easy prof­its, says David McLean, a fi­nance pro­fes­sor at DePaul Univer­sity. He cites re­search show­ing that three years af­ter an aca­demic pa­per on an au­to­mated strat­egy is pub­lished, re­turns based on that strat­egy fall by more than half as more traders catch on. Mean­while, as more quant traders vie for an edge, some en­vi­sion a world where so many al­go­rithms are un­leashed on elec­tronic mar­kets that sud­den shocks—such as Au­gust’s melt­down in U.S. stocks—be­come more fre­quent.

While it’s dif­fi­cult to know pre­cisely how many quant star­tups there are,

Quan­topian, a Bos­ton-based firm that pro­vides coders the tools and soft­ware they need to build quan­ti­ta­tive trad­ing pro­grams, has seen its user base climb to 66,000 from 1,570 in 2011, the year it started.

Open source cod­ing lan­guages such as R and Python, which are build­ing blocks for crit­i­cal num­ber crunch­ing, are posted for free on on­line li­braries. Bou­tique ser­vices such as Es­timize pro­vide crowd­sourced earn­ings es­ti­mates. “There’s so much data, so much open sourced soft­ware and com­put­ing power avail­able,” says Emanuel Der­man, di­rec­tor of Columbia Univer­sity’s fi­nan­cial en­gi­neer­ing pro­gram and the for­mer head of quan­ti­ta­tive risk strat­egy at Gold­man Sachs.

Back in Man­hat­tan, the elite hedge funds still rule, hav­ing built up the sta­tus and the rep­u­ta­tion that come with years of out­size re­turns. Point72

As­set Man­age­ment’s Mid­town-based quant busi­ness, with its large glass con­fer­ence rooms and white walls adorned with founder Steven Co­hen’s per­sonal art col­lec­tion, looks and feels noth­ing like a startup.

And ac­cord­ing to Ross Garon, the head of Point72’s quant shop, big firms have lit­tle to fear from smaller com­peti­tors. They still have the best tech­nol­ogy and bright­est minds (not to men­tion the most money). “The de­moc­ra­ti­za­tion of tools doesn’t nec­es­sar­ily mean there’s the de­moc­ra­ti­za­tion of good judg­ment of what to re­search,” he says.

De­spite those dis­ad­van­tages, Hunter and his part­ner, Ernie Chan, who works from Ni­a­gara-on-the-Lake in south­ern On­tario, have held their own. QTS re­turned 12 per­cent last year, out­strip­ping the U.S. stock mar­ket and the av­er­age for hedge funds glob­ally. It runs $22 mil­lion for in­di­vid­u­als and one large fam­ily fund.

To keep costs low, QTS uses a ser­vice called Al­go­Seek to get ac­cess to price data for fu­tures, pulling in an “as­tro­nom­i­cal” amount of in­for­ma­tion for $500 a month. Hunter him­self wrote the code that QTS’s op­tions trade on. The firm em­ploys part-time con­trac­tors and uses tools such as Ama­zon Web Ser­vices to aug­ment its com­put­ing ca­pac­ity when a lap­top won’t do the trick.

Bil­lion-dol­lar-plus firms might not worry about two guys run­ning a few mil­lion dol­lars. But “there’s a threat they’re miss­ing,” says Dan Dunn, who over­sees prod­uct man­age­ment and mem­ber­ship at Quan­topian. “The re­al­ity is there are bril­liant peo­ple all over the world who they have never seen or heard of un­til they show up and eat their lunch,” Dunn says.

Then again, all those bril­liant minds some­times trip over one an­other. JPMor­gan Chase’s Marko Kolanovic pointed to the role of crowded quant trades in the events of Au­gust, when U.S. stocks plum­meted 11 per­cent in six days. Many blamed China and the Fed­eral Re­serve. Kolanovic told clients au­to­matic sell­ing by “price-in­sen­si­tive” quants made ev­ery­thing worse.

Where Kolanovic sees dan­ger, Hunter senses op­por­tu­nity. On any given day, he tests 10 dif­fer­ent mod­els while ex­e­cut­ing eight strate­gies for clients. He and Chan are con­sid­er­ing de­vel­op­ing code to profit from dis­tor­tions that show up in man­aged fu­tures when too many quants trade the same strate­gies. “We’ve thought about try­ing to take ad­van­tage of it, cer­tainly if the al­go­rithm is clearly af­fect­ing the mar­ket,” Hunter says.

The bot­tom line Com­puter-driven trad­ing strate­gies are eas­ier than ever to ex­e­cute, but that may chip away at their prof­itabil­ity.

Newspapers in English

Newspapers from Australia

© PressReader. All rights reserved.