[ M A R K E T
R E V I E W
|
M A C H I N E
L E A R N I N G ]
a few days with only the best ones
surviving.”
Opening to a bigger market
Holman says the development of
machine learning has come as in-
vestors and data scientists employ
less preconceived strategies with
their AI programmes.
“People who did this in the 90s
had a propensity to overfit data
that was hard to control,” says Hol-
man, who joined from Highbridge,
one of the leading quantitative
hedge funds in the world. “Because
of innovations in compute power
and technology this is much more
robust now. You can explore more
strategies and not overfit, which is
one of the definitions of successful
investing.”
Rebellion Research is anoth-
er forerunner in this field. The
company has been using machine
learning in some form or other
for the last ten years since it was
founded by four partners with
mathematical but little financial
background. The company’s ma-
chine learning programme looks
for historical macro patterns and
connections and makes investment
decisions based on this. While not
always correct, it has a 60% accura-
cy ratio, says Alexander Fleiss, one
of the four founders.
“The system is dispassionate,”
says Fleiss. “It looks for patterns
of relationships. We know we’re
wrong 30% to 40% of the time but
we are right the rest.”
But while sophisticated hedge
funds have been earliest movers
in this field, the next few years
could see the potential benefits of
AI opening up to a bigger market.
Quantopian is a Boston-head-
AI for the masses
The aim is for AI to one day be available to the masses.
Bloomberg, for one, has been developing AI technologies
for clients to use on its terminal. Applications range
from news text and financial filing analysis to produce
sentiment analysis to a question and answering service
which provides users answers to complicated financial
questions that require the backing of Bloomberg’s data
resources to answer them. Other applications include
predictive analytics - which provide automatic recom-
mendations of events of interest, or content to consume
based on statistical models.
“The interest in AI has changed,” says Gary Kazantsev,
head of the machine learning group at Bloomberg. “Five
to ten years ago people thought this was quite esoteric.
Now everyone around the world wants to have machine
learning incorporated into the business.”
quartered firm that provides
an online platform for users to
contribute investment algorithms
with the best being allocated
capital, ranging from $100,000 to
$3 million per algorithm. It is, in
effect, a form of crowd-sourcing
for algorithms. The company has
been funded by none other than
Point72 Asset Management run by
Steve Cohen, renowned as one of
the most technology-savvy hedge
Issue 52
TheTradeNews.com
43