Feedback

This looks like a good concept, but needs more in-depth thinking out yet. What you've written in the abstract to date isn't very suitable for the ultimate objective of summarising an idea etc. (Specific details of Ultimate Frisbee aren't necessary to appreciate your idea.) Making models of different general and specific tournament formats seems to be the essential thrust of your idea. Don't forget the need for a top-level idea, such as evaluating the usefulness of EM in comparing such models etc. You'll need quite a sophisticated model to justify the 30:70 split, so good to think of ways of enriching the model concept

I'm thinking of your model in terms of a sort of management tool. Perhaps eddi can play a useful role here in storing details of teams etc. What I'm not so clear about is what the nature of the dependency envisaged here is: e.g. is team performance predicted on basis of past team performance? or are you contemplating modelling of actual players / even play in any sense?

Watch spelling - typos include:

loose -> lose
peoples -> people's
otherside -> other side
playout -> play out
tournamnet / tornament -> tournament
scenareos -> scenarios

More specific references are needed! You might consult footballTurner2000 and perhaps parts of the cricket model in ~wmb/public/projects/games, though it's true that these may be too much concerned with playing games rather than management issues.