Hello everyone! Thank you for trading this market.

I want to provide some motivation and context for setting up this market, as well as gather feedback on more concrete resolution criteria. At the end, I also include some personal thoughts around why the market may resolve YES and why the market may resolve NO.

Context and Motivation
I love the Monopoly Deal card game and often play with my family and friends. The number of games I've played is in the high-hundreds to low-thousands range. It's an interesting game with a nice balance of skill and luck to keep things interesting.

I often play the three-person version with my parents and am most familiar with this variation. The game can be played with 2 to 5 players, and the strategy and gameplay dramatically changes between the number of players.

I've seen a range of different strategies played. As I play, it often makes me wonder what the optimal strategy is. I looked on the internet to see if there was existing research around the strategies for this game. There is one paper which used BFS based algorithms to achieve a win-rate of 45% according to the abstract.

I'm starting to dip my toes into reinforcement learning, and I believe this would be an interesting practical application of that learning. It would also result in a bot for my parents to play with if I am unavailable. Hence, I intend to produce an agent that can play 3 person Monopoly Deal.

Proposed Resolution Criteria

At the end of 2025, I will catalogue and collect the best Monopoly Deal agents and then have them play a large number of games against skilled and experienced humans. I assume on the order of 50 games is enough to enable skill differences to show through the variations in luck. In order for this market to resolve positively, the bot must achieve a higher win-rate than any human in the tournament.

Some open questions include:

• How will the humans be selected for high skill, given the lack of organization around the game?

• How will the agents be comprehensively catalogued and identified?

• Will there just be two humans, or could multiple humans help to play the large number of games?

• In the case where there are multiple agents available, would we have a tournament style bracket where agents compete against each other to find the strongest agent?

I may end up running a tournament around this if enough interest and competing agents emerge.

Thoughts around YES versus NO

There are quite a few reasons why I believe this may resolve YES:

• Given then success in creating reinforcement learning-based agents for other games which are arguably more complex than Monopoly Deal, this definitely seems within the realm of possibility.

• I personally plan to work on this problem and think the probability of me succeeding is at least 60% and maybe higher depending on whether I am able to prioritize this work.

• The existing paper claiming a 45% win rate is already pretty good and may be higher than most humans.

However, I think that a NO is also a very concrete possibility:

• Monopoly Deal is heavily luck based. Even if there was an incredibly skilled agent, it may fail by sheer luck to achieve a sufficiently high win rate for a positive resolution.

• There is a decent chance (maybe ~20%) that factors in my life may result in me not making significant progress towards this problem personally.

• There doesn't seem to be a ton of research into this problem already, and so if I don't make progress on this problem I am not sure that someone else will.

Those are my thoughts. Thanks for participating in this market and let me know what you think!