Wednesday, April 23, 2014

World Boardgaming Championship Laurel Opportunities

Last Thursday Andrew came by after work to play some board games and escort Adam and I to his place in preparation for Lounge Day. Andrew arrived before Adam and we talked a bit while I was eating about WBC this year. Andrew is finally coming down which should be good times and he was looking at building a schedule despite the final list of games not existing at this point while they continue to scrounge for GMs. He asked me about what games I thought he had a reasonable shot at winning so he could schedule it and practice it.

I'm sure lots of people think this about their University crowd, but I do think the people I gamed with back in the day really were significantly above average when it came to game skill. This leads to things like Pounder and I making the finals in Puerto Rico in back to back years despite neither of us thinking we're any good at the game relative to the people we used to play with. So while Andrew loses a lot of games around here I think he's still going to be pretty good compared to the average person at WBC. But even then, there are some truly excellent players in a large number of games so identifying a decent game to work on seems like a reasonable thing to do.

I think the way to approach the question is to consider where a given game lies on the luck vs skill axis. Realistically it should be more of a 2d graph since there are definitely games with a lot of both luck and skill but I think simplifying it down to just one factor isn't too unreasonable. I feel like what I care about is how often an expert will lose to a pretty good player, and how often a pretty good player will lose to a relative rookie. Something like A Few Acres of Snow is going to end up way over on the skill side of the line even though when I play Nick Henning I feel like the result is going to be pretty random if we're not rusty. On the other hand something like Can't Stop is going to be entirely on the random side of the axis. I don't actually think you can be really good at Can't Stop. I think you can be bad for sure but if you stick a Can't Stop 'expert' at a table with 3 good players I bet they'll win 1 in 4 games.

But how to identify those games? We could go by gut feeling but I actually think a reasonable way to look at things would be to look at the list of people who have earned laurels historically and see how often the same people repeat over and over again. Taking Can't Stop, for example, the game has been played for 8 years at WBC. The top 6 earn laurels each year so there have been 48 people who have earned laurels at Can't Stop over the years. If we look at the laurel list for the game there are 47 unique people on the list. This means that only one person has managed to come in the top 6 more than once in 8 years! I think it's pretty clear that there's no consistency at all in the year to year performance of any given person, and that the one person who pulled it off twice simply got lucky. A Few Acres of Snow doesn't really have the data to look at since it's only been around for two years now but it did have the same top 3 both years. Can't Stop has 1 multi-laurelist in 8 years, AFAoS has 3 in 2 years. AFAoS also has a significantly smaller field as well which would need to be taken into account, but I think it's pretty clear that the two games are pretty far apart on the luck v skill axis.

I'm not really aware of an easy way to actually pull this info off of the BPA website (I may write a spider to go hunt it all down if I get really bored) but I don't know that we really need to. Identifying what type of game we're looking for and then using gut feeling to compare is probably good enough. Though if someone was actually just bored at work and was reading about the games anyway they could totally look into it...

So what type of game should Andrew be looking for? Well, the high skill games are certainly the best bet if one has the time to dedicate to really learning the game and high level opposition to practice against. But Andrew has a job and kids and probably doesn't have the right opponents on hand to pick up a game like Twilight Struggle and reach the top tier. A Few Acres of Snow is a possibility if it even gets a GM and comes back this year but I don't think any of the other high skill games are feasible.

Should it just be the high luck games instead? They tend to get a lot of players and the chances of winning are really more or less random. There's also not a lot to be done to get better at Can't Stop. Andrew's already as competent at the game as I think it's possible to be.

I think the proper area to be looking is the medium-high random Euro games. Something like a Vegas Showdown where there are a few viable paths to take and which one ends up winning is determined randomly by the order of the cards and tiles. The Theatre wins the game, assuming it gets to flip and be bought by the person who was playing for it. But that's not actually terribly consistent. I think it's still the right line of play, but you can't know if it'll win or not in time to make intelligent decisions about it. I feel like past me would argue with the statement that Vegas Showdown is really random at its core, but the laurel list makes a pretty good case I think. The game has been played for 6 years and has 29 different laurelists. (The webpage has 30, but it lists Mike Kaltman twice.) No one has placed in the top 6 more than twice. And it's not for lack of trying, because I know most of the people who have laurelled twice and I know they keep playing every year. Myself, Sceadeau, Randy, Eric, Cary, Mike, and an Andrew I don't know. Those are the sorts of names I expect to see high on the list, but when you compare this list to Twilight Struggle's list... It's pretty clear that one of these games has something keeping the top players from repeating year after year.

So I think ideally you're looking for a game that has some choices, and has some right decisions to make, but where something outside the choices works to determine who wins. A Euro where your choices have a strong correlation with the outcome of the game probably requires too much practice for someone with Andrew's schedule. Le Havre is what I'd call a low random Euro in this vein. 4 years at WBC, only 12 people have earned laurels.

There is one other category of game that's ripe for the picking... And that's the underattended game. If you can identify a game that other people aren't very interested in playing you may well be able to get into the top 6 just by showing up and being decent and/or lucky. In 2010 I remember winning a game of Factory Manager but decided to skip the semis because the Agricola semis were an hour later and I didn't want to risk Factory Manager not finishing in time. (It was scheduled for 2 hours so that seemed like a real worry, but then Agricola started late as we waited for (I think) Amun Re to finish and I was a really fast Factory Manager player so I probably could have pulled it off.) They only had 9 people show up for the semis. So 67% of the people who won either heat and bothered to show up for the semis earned laurels.

Trials are notorious for not getting many people (at least compared to Century events which have proven at least once before that they do get people) and keeping an eye on what they're up against could lead to a 'Caveman: Quest For Laurels' situation. Often people will put a lower weight on trials because they only have 1 prize instead of 3-6 but it's always top 6 for laurels.

But really there's not a great way to know in advance what events are going to be underattended or softer than normal. Maybe look for something like the Agricola semis which will have a lot of the really good Euro players and find something scheduled right beside it? Or look for new sponsor trials since those ones didn't even get voted in by the membership so there may really be little to no interest in the game.

That's all stuff to consider once the actual schedule comes out though. But for theorycrafting for now I think looking for Euros with a good amount of randomness and finding an online site with the game to get practice on would be a good start.


Nick Henning said...

So, question, are you(your friend) trying to maximize laurels at WBC? Or figure out which games you are most likely to do well in? Cause they are totally different things.

andrew said...

Hi Nick,

I'm more looking for games that I can do well at. Though I also am interested in going home and saying that I am the xth best player in the world at game y.

Robb Effinger said...

You should definitely consider innovation, Andrew. It's fairly high skill, some luck, under-attended, and you can practice online in short bursts. Of course, I do plan on stopping you from winning...

Sceadeau said...

So, even those of us who go and know the stiff competition have problems saying xth best in the world after coming home from WBC. :)

My recommendation is just to play the games you like playing. You play with some stiff competition, meaning you'll probably be better than the average bear.

Nick Henning said...

I think Sceadeau's advice is best here - playing the games you like (and practicing them) is going to do the best for you. With something like Twilight Struggle, if you really dig it, go for it, Id advise with a game like that to do some online reading about it.

Sthenno said...

I feel like these comments are missing the trees for the forest here. Sure, we can go to a competition and play games we like, but can we make a spreadsheet about it ahead of time? I guess with the schedule, but there's no real numbers to crunch.

Now, if we could get the number of Can't Stop competitors each year then we could really go to work. Assuming the people there are distributed in skill along a bell curve we could start comparing the probability of this outcome under various hypotheses about how much a standard deviation in skill comes out to in terms of win percentages. Could we do something with Bayesian analysis to figure out which win percent / standard deviation is the most likely? Man, a 95% confidence interval of how much win percentage a standard deviation of skill gets you at different games would be a thing of beauty.

Sceadeau said...

All of those numbers are available on their website.

My insight, as someone who has gone to this thing before, is that number crunching in this scenario won't really help you, unless you are looking for games you can get into that you will have the same chance as everyone else at.

Play the games that you have the statistical edge at because you are better at them than other players.

Tech (defined as information about a game that you've discovered that other people don't know) comes from playing the games...a lot. Also from playing a wide variety of people.

Your absolute best bet for winning comes from understanding the metagame to a level that gives you a minor tactical edge in your game. The metagame can be understood in a lot of ways, either reading about known strategies, or playing a ton of games online to understand the general tendencies.

See it, counter it, beat it.

Or, you know, go play Can't Stop. Someone has to win it.