You are on page 1of 3

20/9/2017 Wall Street seeks an edge from crowdsourcing

Trading Technology
Wall Street seeks an edge from crowdsourcing
Some traders are wary about contagion of herd behaviour

On Wall Street

Online livestreaming service used by Mike Roberts to allow viewers to tell him how to trade $50,000 of his money AFP

JULY 6, 2017 by Robin Wigglesworth

Last month a software developer called Mike Roberts set up a stream on Twitch, a popular online
livestreaming service where geeks watch other geeks play computer games, to allow viewers to tell
him how to trade $50,000 of his money.

The game is divided into five-minute voting rounds through the trading day, with random viewers
entering astute advice and comments such as Lol who bought XLE and ooo MOO is finally
breaking even. (XLE is an energy exchange-traded fund and MOO is a suitably named agriculture
ETF.) Every five minutes the top buy or sell commands will be executed and one share will be
bought or sold.

It might sound mad the retro look and thumping electro-pop soundtrack doesnt help but
Mike is on to something. Crowdsourcing is an increasingly hot topic in some corners of the finance
industry, where a clutch of ambitious start-ups are betting that collective wisdom can be the secret
to beating markets or analysing companies better.

Online platforms such as Quantopian, Numerai, QuantConnect and Quantiacs are hoping that a
horde of freelance programmers and computer scientists will upend the world of hedge funds an
https://www.ft.com/content/20fdbe88-61c3-11e7-91a7-502f7ee26895 1/3
20/9/2017 Wall Street seeks an edge from crowdsourcing

idea co-opted by WorldQuant, a hedge fund that manages money on behalf of Millennium while
Estimize crowdsources corporate earnings estimates that have proven to be more accurate than the
predictions of traditional Wall Street analysts.

Leigh Drogen, the chief executive of Estimize, reckons that finance is just at the first bat of the first
innings of this crowdsourcing trend, comparing it to how crowdsourced Wikipedia has upended
the business of traditional encyclopedia companies. The internet is a uniquely powerful tool to pool
the collective insight of thousands, even millions of people, cheaply and efficiently, he points out.

At its core it is an old idea. Over a century ago, British scientist Francis Galton was attending a
country fair in Plymouth, where someone had organised a competition to guess the weight of a fat
ox after it had been slaughtered, and he was struck at the results.

Almost eight hundred people entered into the competition ranging from expert cattle breeders,
butchers and farmers to bored fair attendees with little knowledge of cows and the median guess
for the weight of the ox after being slaughtered was 1,197 pounds. The right answer was 1,198
pounds in other words the crowd was astonishingly accurate.

This anecdote led James Surowieckis seminal book on the subject The Wisdom of Crowds but
also underscores the central challenges. Thats because crucially, each guess was truly independent,
rather than being influenced by the group. Surowieckis book also shows how an experiment by the
Swiss Federal Institute of Technology showed that the more information participants were given
about everyones guesses, the worse their predictions became.

This is something that Mr Drogen is aware of. While he stresses that Estimizes estimates are not
made by random people off the street it currently sources forecasts on corporate earnings and
economic data releases from more than 52,000 hedge funds, brokerages and freelance analysts
the fact that it includes experts and amateurs is crucial.

For example, Estimize does give more accurate forecasters a slightly higher weight in its overall
estimates, it found that giving them too great a weighting lessened the accuracy of its overall model.

Quantopian takes a similar tack. Earlier this year it began allocating a few million dollars to some of
the trading algorithms coded by the 130,000 freelance data scientists, mathematicians and
programmers on its online platform, for its own hedge fund. The money is from Steven Cohen, a
noted hedge fund manager and will be gradually scaled up once the fund has proven its mettle.

https://www.ft.com/content/20fdbe88-61c3-11e7-91a7-502f7ee26895 2/3
20/9/2017 Wall Street seeks an edge from crowdsourcing

But crucially Quantopian doesnt just simply select the algos with the best theoretical and actual
returns, but looks for ones that collectively result in the best, steadiest and most diverse return
streams. Jonathan Larkin, Quantopians chief investment officer, stresses that the diversity of its
contributors is vital.

Not everyone is convinced that a motley crew of random do-it-yourself quant can do any better
than existing quant hedge funds stuffed with PhDs, pointing out that the noise-to-signal ratio will
be sky-high for these amateurs. For regulatory reasons Quantopians overall, early results arent
disclosed. Mr Drogen says that the scepticism faced by Estimize and another of his start-ups,
crowdsourced stock-market picking platform ForceRank, is visceral, despite the results.

Thus far, the results for Mr Roberts StockStream are mixed at best. The game has been running
for a little over the month, and the Amazon developers $50,000 starting capital plus $500 of
free shares has been turned into $50,160, a slender loss of 0.7 per cent and roughly in line with
the broader S&P 500.

Still, there are plenty of people betting that crowdsourcing will be part of Wall Streets future.

Copyright The Financial Times Limited 2017. All rights reserved. You may share using our article tools. Please don't copy articles
from FT.com and redistribute by email or post to the web.

https://www.ft.com/content/20fdbe88-61c3-11e7-91a7-502f7ee26895 3/3

You might also like