ACPC: new events, and the old event proposal document

[Neil Burch]

Hello everyone,

In the round table discussion, a few people mentioned a form that described the kind of questions that need to be answered for a new event.  This can be found at http://www.computerpokercompetition.org/downloads/documents/event_proposal.pdf

This isn't some sort of official document.  It's just intended to make sure that any proposal answers the kind of questions that determine both the success and the utility of an event.

One event proposed at the workshop was stopping the 3P limit event and replacing it with 6P no-limit.  I think most of the questions in the "new event" document were already brought up.  Here's my paraphrase of the questions, and some of the issues that occur to me.

Note that I'm not championing this event -- it's interesting, but not my thing.  That'll be up to someone else.

- What are the rules?

6P NL is pretty well defined, but what are the blinds: 50/100 or 1/2?  What are the stacks?   1/2 chip blinds with 100 chip stacks does have the advantage of better matching human conventions.

What do we do for bankroll if this is very popular?  Looking at all opponent combinations could blow up if the event is popular.

- Are there interested competitors?

We can always just try and see, but getting some people who will commit to entering ahead of time (maybe 4 teams?) seems like a good idea.  If there really is interest, this shouldn't be a large hurdle.  By contrast, if we can't even get 4 teams to commit to entering, maybe it's not yet time for 6P NL.

t would be nice to know that there are likely to be at least 6 teams already interested.

- What new science does this event open up?

My recollection of the discussion is that there's a consensus that three player limit seems stagnant, and this a proposed replacement

- What kind of resources are needed?

Storage is probably uninteresting, but time is a potential concern.  A simple limit of 7 seconds of thinking time per hand would mean up to 42 seconds in a six player game.  This probably wouldn't happen much in a ring game -- 4/6 players can fold for free -- but it would be good to have some sort of estimate of how long a match would take.  CPU time is the single biggest limiting factor for the competition.

- Everything to do with statistical significance.

What's an estimated per hand standard deviation of winnings?  Some sort of duplicate technique could still be a thing, but we almost certainly don't want to do all possible 6 player table permutations.  If some sort of duplicate method is chosen, how much does it help?  What kind of win rates might we see?  I would guess that the win rates are lower, because of the 0 chip blinds for 4 of the 6 players: are they low enough to be a problem?