Team I participated did not win, but I think we improved our method quite well. Initially, we did not schedule features (with a price tag), but tried to prioritize individual user stories (one feature consisted of multiple user stories) by ordering them according to decreasing ratio of value and cost (effort). Value of each story was, in the first technique we used, determined by feature price equally distributed to stories involved. This seemed reasonable, but resulted in multiple features in progress, and incomplete features gave no income.
We improved the method to use the following rules
- Make a guaranteed release as soon as a feature is complete. Do not let the variation in velocity to cause a release to be skipped.
- Prioritize full features (all related stories together) in decreasing order of business value (price) divided by effort.
- You should take account things like customer satisfaction in business value.
I insisted on having a set of rules for playing the game. That probably made us lose some money in the first rounds. Also, we did not have rules for customer satisfaction or strategic moves like abandoning some customers in favor of others. Intuition might have worked better this time. Also, we did not schedule small items to fill the remainder of sprints with useful work, which would have yielded a small speedup for some features occasionally.
http://confluence.agilefinland.com/display/af/Agile+Dinner+in+Tampere+20090514+-+Business+Value+Game+Workshop