T HO U G HT L E AD ER S HIP
Electronic Trading &
Algorithmic Intelligence
It’s impressive what science and technology can accomplish
these days. Taken together, the open-source software
revolution, affordable hardware and vast amounts of data have
brought to life static equations from science books, previously
accessible only to a numbered few. This has had a dramatic
impact on distinct fields, ranging from bio-technology to
education. Obviously, finance and trading are no exceptions.
W
ith the use of lines of code, our
Trading Product Development team
rides this wave in order to deliver a cutting-
edge trading performance to a number of
clients worldwide. Our development team
develops intelligent Intra-Day Trading
Algorithms that can consume vast quantities
of information and therefore react to market
changes on a nano-second time scale.
Besides opening up a new world of trading
strategies, these capabilities diminish
errors while at the same time dramatically
increasing the capacity to handle very
large volumes – once each instance of the
algorithm can handle multiple orders at
the same time. In terms of value, algos can
reduce all sorts of frictions and improve
performance vs. a benchmark such as the
volume-weighted average price (VWAP) by
systematically acting according to statistical
rules that are strictly followed during the
trading day.
The differentiation factor: Expertise in
Intelligent Algos; rather than relying on off-
the-shelf solutions, our algorithms consume
a number of proprietary quant-driven
Figure 1: How models “learn”? Optimizing parameters of a Neural-Network; a main model
behind algorithmic trading decisions.
16
THETRADETECH DAILY
trading signals tailored for clients’ specific
needs.
Deep-Learning: The real driving force:
What are exactly these models / signals
that enable leading players to enjoy such
a position compared to their competitors?
Deep-learning; A sub-class of A.I., deep-
learning architectures allow one to
explore and quantify a number of market
micro-structure patterns such as the
future direction of the order-book or
the probability of a higher than average
volume to be traded in the near future…
etc. Frameworks such as neural-networks,
also become more “intelligent” with each
passing day and do not forget what they
have “learned”. This allows excluding from
the equation psychological biases such as
fear and thus respond to a number of market
movements using what was learned from
previous similar scenarios in a probabilistic
fashion.
What does the future hold?
By exploiting our unique expertise and the
associated development of new quantitative
models through research, we intend to
provide our clients with intra-day execution
strategies that surpass expectations in
terms of performance. This means that we
have to provide significant improvements
and / or find unique liquidity pools at the
right time. Moreover, classical algorithms
such as volume tracking can now act more
opportunistically than before and over-the-
day orders can now have a bias regarding
future market trajectory and adapt these
biases automatically, if necessary.
Bottom line is, this cutting edge sub-
class of artificial intelligence – as well as
others - is here to stay and, if used properly,
can definitely help to quantify market
uncertainty and act optimally in the face of
uncertainty.
With so much available, we seem to
be limited only by our own knowledge
and creativity; core assets for the next
generation of this business. Combined
properly, these factors will produce
novel features that will deliver results,
independently of market conditions, for
years to come.