Work and New Ideas @ Olsen

James has sent me an article with a good description of agent based models in economics. My proposed system theory differs from the agent based models in the following way:

Agent based models distinguish between ‘agents’ and complex structures based on interactions of agents. Separate rules apply to ‘agents’ and their interactions.

In the system theory approach I assume that every ‘object’ is a system. A system is made up of interactions of systems and itself interacts with other systems. Every system is subject to ‘relativistic effects’. For example: from the perspective of one system - an other system is indeterminate, i.e. only its current state or history of states are visible. This indeterminacy is responsible for the fact that ‘interactions between systems’ are typically not one time events, i.e. completed with one action, but result in a series of interactions, where the two systems reach a new equilibrium. the system approach thus simplifies: ‘agents’ and their interactions are systems governed by relativistic effects.

I gave everybody Editor rights, so you’ll be able to post new entries and edit existing ones. For new users which I did not yet give the rights, just write me an email

Hi folks,

I would like to discuss with you how we can start opening up our business to new concepts such as open source and mass collaboration. It has already been in the air for years but somehow this could never get in track. The question is, is it time now to go for it or not?

In the recent past we have been thinking that we need to still get some more momentum before we start such a strategy. But can’t the opening process actually give us the momentum we are still missing?

So let’s say we are not yet ready to open up the whole technology, since we need to have a stronger market position to sustain other companies possibly copying our business.

What we can do is then to provide the current research infrastructure through a web platform where anybody can play around, invent new trading models and run them in historical runs (through our servers). In this step we don’t open up the technology, but the ideas. And let people from all over the world play with them. As well as offering emulation runs, we can then plug in the personalized trading models into a customer own account. I.e. a customer can create a trading model and possibly we run it for him on his account.

This could be then a first step for creating a community around our business. People can get involved in creating their own trading models, either for the sake of satisfaction (like oanda’s fxgame), or for managing their own money. Plus we can even create a contest where we provide a money reward to the best performing models to attract even more attention.

That said, it looks like the resources we have to invest in are, at the beginning:

1. development time for creating the web client. The web client can either look like our already existing GUI or even a bit more sophisticated where we present configured objects in a graph way
2. a bunch of servers where to run the service. We in fact would have to run emulations on our servers. In this way the technology still stays at home and we get always a snapshot of what’s going on.

Yet another welcome message!