camron.systems.australia section 24 of 27

© 2007 « home « next »

do not copy - protected by copyscape

algorithmic trading

otherwise known as robots

All work other than citation work is copyright. Unless permitted under the Copyright Act 1968, no part may be reproduced, nor any exclusive right exercised, without the permission of Camron Systems Australia 2008.

purpose
In keeping with our philosophy of informing lone-traders of current trends you will find

an avalanche of material about algorithms in general.

some quality articles explaining what they do.

nothing about how they work.

The "majors" do not disclose their methods, which are constantly evolving, for fear of reverse engineering.

The SFE-SPI200 futures market is dominated by algorithmic trading systems.

The ability to recognise algorithmic activity is an essential skill for retail SPI200 traders.
During slow periods of intermitent activity, where the market maker simulates action with offset orders, filling the order book with spoof (tease) orders, a retail trader can execute an order, get filled, and find themselves under water in a short space of time. In these modern electronic times the need for "small" retail traders to learn the art of tape-reading has increased, not decreased. It's essential.

The skill is the recognition of the market-maker on both sides of the order-book. Before the electronic market, "locals" performed the function of market-makers. They've been replaced by a single computer system which can populate, de-populate and re-populate the order book in a nano-second. Re-arranging the order-book instantly.

lone-wanderer
The lone-trader must be able to detect when the market-maker is the sole occupant of both sides of the order book ("market depth"). During this time the market-maker will know when they have filled a lone-wanderer. During the night session when the spread is wide, 10 levels can be withdrawn from the order-book, pulling the rug out from underneath the lone-trader, testing their resolve.

If we were developing an electronic-algorithmic trading system, that's how it would be done.
See the screenshot below. We would know the total depth. We would know how many were ours.
If someone bought one off us at the front of the ask queue, we would pull our buy orders out, and sell 1 lot, 10 levels down, to see if the buyer had a stop entered.

screenshot is available in the preview package

electronic systems
A mechanical system is an automated set of rules operated to make a profit.

There are three type of algorithm systems
• A buy-side algorithm is an automated set of rules used for accumulation.
• A sell-side algorithm is an automated set of rules used for distribution.
• A dual-sided algorithm used by market-maker to either create or seek supply and demand.

Typically, a market maker operates both buy and sell processes, side-by-side, that are aware of one-another.
see market making explained and monopoly power of the market-maker in setting the open price.

pinging
Pinging is searching for liquidity.
Orders are sent out, held for several seconds and then withdrawn.
In one reported example, in 60 seconds, 6000 orders were issued and withdrawn.
Only one successful fill was achieved.

robotics
Is there a solution?
Yes. It can be mastered.
With a long involvement in technical programming, robotics, and numerical-control systems we have a distinct advantage in this area in that the fashion is catching up with us. Or a rusty skill is coming back into fashion. We can see them. We can see it happen. There are imperfections in algorithmic systems. Our software solutions, detailed on other pages, contain the kernel of these skills. Algorithmic trading systems are a modern application of old-style numerical-control-systems. Learn about them. Not hard.

tell-tales

The behaviour of algorithms is obvious once you know what you are looking for. The presence of an opposing robot is crystal clear to (a) another robot, and (b) institutional traders who are trading against robots. The tell-tales are there. Bigger and better than before.

If you think like a robot what would you expect when coming up against an opposing robot.
If you think like a market-maker what would you do when up against an opposing robot.

Algorithms are used by "majors" in

index rebalancing
arbitraging
market making

manifesting themselves in

sweep orders
iceberg orders
market depth dancing

some worthwhile reference material
We've done some of the research for you.

Algorithmic Trading - Machines are taking Over
Financial Times Mandate - Algorithmic Trading
When the robots get it wrong - damage - and - opportunity

update october 2008

Over the past 4 years, the 5 major "prime broking houses" poured hundreds of millions of $ into the development of algorithmic systems and establishment of huge centralised computerised trading rooms. 3 have gone and the remaining 2 have become "banks" which means their "prop-shop" activities will be severely curtailed.

Their algorithms didnt help them. If anything, their unfailing belief in and reliance on algorithmic systems probably contributed to their demise.

solutions to complex problems

• a window on algorithm activity and how to detect it.
• detect when algorithms are in "buy mode"
• detect when algorithms are in "sell mode"
• sophisticated yet simple solutions not provided by anyone else.

research papers on these topics available on subscription (p.o.a)

watch this space

coming
pay-per-view e-paper on detecting and responding to algorithms
pay-per-view e-paper on detecting and responding to icebergs orders