[ M A R K E T
R E V I E W
I
n 2015, David Siegel, the
co-founder of quantitative
hedge fund giant Two Sigma,
told a conference that there would
come a time “when no human
investment manager will be able to
beat the computer.” What sounded
like science fiction is gradually
becoming reality as computers
move into the investment world.
Artificial intelligence (AI) has
already beaten humans at many
other games—chess, Go and poker
to name but a few—but its use in
investing has been slower to mate-
rialise. This, however, looks set to
change.
Machine learning is a catch-all
42
TheTrade
Summer 2017
|
M A C H I N E
L E A R N I N G ]
name for a range of algorithms that
can identify repeatable patterns
and relationships within observed
data. The development of machine
of its actively managed funds into
algorithmic funds with a focus on
AI and quant-based investing. The
move has seen seven of 53 stock
“Our algos are biologically inspired.”
JEFF HOLMAN,
CHIEF INVESTMENT OFFICER, SENTIENT
learning is being driven by the
computing and data revolutions
of the past decade. Computing
power has broadly doubled every
two years since the 1970s making it
easier for new entrants to tap into
the computing power needed to
develop AI technologies. Mean-
while, it is estimated that 90% of
data in existence today was created
in the last two years. A company
like AHL, the quantitative invest-
ment arm of hedge fund manager
Man, for example, receives around
1.5 billion data ticks every day.
Harnessing and storing data has
become easier—in 1981 a gigabyte
of storage cost $300,000, the price
is below 10 cents today.
And asset managers are sitting up
and taking notice.
In March BlackRock - the largest
fund manager in the world - an-
nounced it was moving a number
pickers stepping down, and will
impact some $30 billion of the
company’s equity funds. It could
mark a watershed in the progress
of fund management as more
follow its lead into machine-based
investment.
Sentient, founded in San Francis-
co in 2008 by scientists and engi-
neers, has spent six years building
out the different components to
capture data for investment pur-
poses. Using dark cycle technology
to tap into hundreds of thousands
of unused computers from 4000
facilities in the world it claims to
have one of the largest compute
facilities in the world dedicated to
AI. Its machine learning algorithm
uses evolutionary processing —a
branch of AI inspired by bio-natu-
ral selection—from which it has de-
veloped AI techniques and applied
them to stock pattern prediction.
“Our algos are biologically
inspired,” says Jeff Holman, chief
investment officer at Sentient.
“They are evolutionary algos used
to develop strategies via natural
selection. They undergo mutations
and combine or breed with each
other. The compute power means
you can test billions of strategies in