Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Workshop 4 (B) - Algos For Humans Dev Gill Marex Spectron - Original.1550055327

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

HOW CAN WE AS HUMANS DEPLOY ALGOS?

Dev Gill, Marex Sugar

1
WHAT ARE ALGORITHMS?

– Algorithms are computer programs that follow a set of rules.


– When it comes to order generation and execution, they can be divided into two categories:

1. Using a computer model to decide what to trade and when. These can be referred to
as “alpha-generating” (in that they attempt to beat a benchmark) or as “black-box” (in
that they are run in a closed environment by a client, who do not want others knowing
the methodology behind them). Market-Making or liquidity-providing algorithms would fit
in this category.

2. Knowing that you want to trade, using a computer model to decide how to execute it to
achieve a desired result, for example a benchmark price (settlement, VWAP etc) or to
limit market impact. These are more accurately known as “execution strategies”.
– Typically, clients of Marex will use type (1), whilst Marex provides clients with access to
type (2).
– Note that the term “algorithm” can apply both to the order (as in: I have sent you an
algorithm) and also to the decision making process (as in: This order was worked via an
algorithm). We often use “algo” for short.

2
“EXECUTION STRATEGY” ALGORITHM USE

– “Execution strategy” algos are more commonly offered by brokers and used by either the
trading desk at the client or the broker here at Marex.
– In this case, the client placing the order already knows what they want to do (for example,
buy 1,000 Copper 3M). The purpose of the algo in this case is to achieve an execution as
close as possible to the client’s desires.
– Depending on the client or order in question, the desire may take different forms. For
example:

– Be as close to the volume-weighted average price over a specified time period in the
market.

– Complete the order by being 10% of the overall market volume.

– Execute the order at a price around the market levels when I place the order, but without
moving the price away from me (“having an impact”).
– These different goals inform the design of our algo trading suite.

3
VWAP

– VWAP attempts to achieve an overall fill price as close to the volume-weighted


average price over that period as possible.
– To do this, VWAP uses a historical volume curve to predict where volume will
trade over the period of the order, and sends child orders in proportion to that
volume.

4
TWAP – SCHEDULED BASED

– TWAP attempts to participate evenly across a given time period. As such, it can
also be known as a “time-slice” algorithm.
– The execution path of the order from the previous slide would look like this:

5
PEG
– We want to buy 100 SBK9 using Peg, with a display size of 1, a reference point of “Bid” and an offset of 1. The market
looks as it does on the previous slide, before we place the order.

– After we place the order, the first child order is submitted at the bid price plus 1 tick (our offset).

– Let’s assume someone now sells 19 lots at market, taking our bid (12.75) and the 2 lots at (12.74). If nothing else
changes, then our next child order would go in at the bid plus 1 tick again. This cycle repeats until the order is filled:

6
MAREX ALGORITHMIC TRADING SUITE

– Enhanced algorithmic execution capabilities, available via FIX and Marex Sales Teams
• VWAP - Volume Weighted Average Price
• TWAP – Time Weighted Average Price
• POV – Percentage of Volume
• POV AXIS – Percentage of Volume AXIS
• TIME SLICED
• Marex VWAP
• Marex TWAP
• MAREX Peg

– Strategies available on CME, LME, ICE, EUREX across all products

You might also like