the details
“ The result is: markets are not totally efficient, not totally random. There is a small amount of bias information”
Systematic Trading
Systematic trading has many advantages over human decision making.
The biggest problem is the tendency to over-optimise; to curve fit historic data to generate acceptable returns.
The reality is: financial markets are very close to being random. There is a very small bias which arises from the fact that everyone is trying to do the same thing. Speculative traders, systems, retail traders, market makers, long-term traders, day traders, strategics; all try to buy low and sell high. Longs tend to wait in a rising market, sell in a falling market, and vice versa.
The result is: markets are not totally efficient, not totally random. There is a small amount of bias information.
Systems will outperform humans, but with limitations. You cannot ‘create’ more information than exists. Curve fitting pushes unachievable profits into the historic performance, leaving the system to underperform going forward. The winning strategy is to find more efficient ways to extract the information and apply robust risk management to protect the system from ‘unknowns’.
Swarm Intelligence
Definition: the collective behavior of decentralized, self-organized systems, natural or artificial. It could be considered as a sub-group of Artificial Intelligence.
SI systems are typically made up of a population of simple agents interacting locally with one another and with their environment. The resulting effect is that the ‘colony’ exhibits significantly more complex intelligent behavior than the individual agents are capable of.
Ants build vast, complex underground structures that function efficiently (including sophisticated temperature control and air flow); even though as individual agents they have almost no intelligence.
Fire ants build towers to survive flooding that are continually being re-constructed to control temperature and shaped correctly to repel water and hold the maximum agents.
“ Ants build vast, complex underground structures that function efficiently even though as individual agents they have almost no intelligence”
Properties of swarm
Made up of sufficient agents to generate effect
Communication between agents must be efficient
Agents behave consistently and interchangeably
Agents
must be simple; and or have ‘limited scope’
Agents have simple behaviour and limited rules
No individual agent or group of agents controls the colony
Agent rules must drive to keep the colony together
The colony has no memory of individual agents
Power of swarm

In a 16 agent colony there are 16 individuals but 65534 connections of agents.
The ant is the equivalent of a single neuron in a brain.
By itself, or even with millions of unconnected individual ants, it cannot achieve much. But if the agents are connected to form a Swarm, the number of connections is orders of magnitude larger than the number of individuals. The colony becomes a brain because of the huge number of connections.
That is why an ant colony is so much more intelligent than an ant.

Swarm XVI - The Basics
Systematic Trading Software
Minimal curve fitting.
Parameters set in advance, as far as possible.
Long backtest, with 50% as ‘blind’ test.
Slow trading and sampling. Reduces costs.
Liquid markets – increased capacity and reduced slippage.
As many agents as is feasible. (Practical considerations).
Extract the small amount of information available efficiently.
Apply Swarm Intelligence.
Use multi-level Risk Management.
Swarm XVI – The Parameters


Risk Management
1
Level 1
The program has a ‘fast’ internal volatility tracker that normalises the risk for each of the markets. This has the effect of reducing risk if one or more markets increases in volatility. It is a basic VaR modulator. This is re-calculated weekly, to reduce constant adjustments.
2
Level 2
There is a systematic RM overlay that tracks the output across the combined portfolio. It does this over twenty-two time frames. There are automatic de-gear levels if the program reaches pre-set boundary levels. This is calculated daily after close.
3
Level 3
There is a discretionary intra-day procedure. Selected risk managers monitor the performance on a live ‘tick-by tick’ basis and the watch world events. They have the authority to reduce risk at any point, under any circumstances. The purpose is to protect the equity, where there is a significant increase in risk or a new situation that the program may not have encountered before.
4
Level 4
Break Stops @ 7.5% daily and @ 20% of AUM peak-to-trough monthly drawdown. If triggered all positions are closed and client notified and a report made on losses. Permission is then required to re-open.
About Us
Swarm Technology has developed a way to harness the natural power of Swarm Intelligence to extract information more efficiently than traditional systematic methods. This is embedded in the Swarm Matrix™ trading program