Tuesday, June 24, 2014

Lessons Learned from Tournament Results

National season is upon us and qualifying doolists throughout North America are scouring deck lists to get a hint of the matches they will face.  Do I need to prepare for Sylvans?  What’s the hot tech?  Most of all, is my deck good enough to take on the best?

As a connoisseur of deck lists, let me give you my take on what one can and can’t learn from studying tournament results. 
  What Deck Lists Can Teach Us

Know thy enemy.  As I have pointed out before, there is little variation between builds of the same archetype.  Konami is in the business of creating sets that work together.  In this way, they are like the Lego Company.  Sets with prefabbed instructions have pushed out the generic blocks. You can try to get creative by adding other blocks to the Death Star, but it’s going to look goofy. Yugioh archetypes are similar. 

Lest you think this is all a corporate plot against creativity, most successful decks are the result of the Yugiverse’s collective conscious. We test, play, post, rinse, and repeat.  The web makes the process efficient and rather homogeneous.  Don’t believe me? Try changing a meta deck by more than 10 cards.  You will, almost inevitably, make the deck worse.  I’ve tried it.  For the past four weeks, I have tested a variety of ideas in my Geargia deck.  They all loose. 

As a builder, uniformity is dull. As a player, uniformity means I can know what my opponent plays.  By spending time among the deck lists, you can find out the number of battle traps, on-summon traps, spells, and effect monsters played.  Take advantage of it.

Gain a little inspiration.  Though the majority of the deck lists are pretty uniform, one can find interesting tech choices scattered about.  Be honest, how many of you thought about Mystical Rephpanel until you saw it on a deck list?  Sure, we all wish we were Yugisavants with the entire catalog of 7,000 cards at our fingertips; but, there is no shame in using your fingertips to Google deck lists. 

  What Deck Lists Can't Teach Us

Find the best deck.  Unfortunately, deck lists do a poor job of telling us what the best deck is. To get some idea of the interaction between a deck’s win percentage and tournament results, I set up a poor man’s Monte Carlo simulation.  In this model, I created six archetypes: Psychopomp, Aumakua, Veles, Morrigan, Evangelion, and Margot Minions*.  The model was created so that each archetype had an assigned win percentage against the other decks.  I ran this model twice, using two different sets of percentages.  These percentages are given in the table below.  The winning deck is listed in the column on the left.  For example, Psychopomp will beat Veles 56.6% of the time in simulation 1 and 66.7% of the time in simulation 2.
Real life Monte Carlo simulations run these scenarios hundreds of times.  Each time, a variable is changed so that the modeler gets an idea of importance of that variable to the model.  But time, money, and expertise limited my little experiment to 100 players, playing five tournaments of six rounds each.  Each “tournament” was run using the Swiss format with three points going to the winner.  The model accounts for draws.  On average, 3.8% of the matches ended in a draw, which seems like a reasonably realistic number.  Each deck also had an equal number of participants**.  The table below gives the percentages of decks finishing in the top 16.  

 While the analysis is somewhat crude, I do think there are conclusions that can be drawn:

  1. Looking at a handful results can be misleading.  If you only looked at tournaments 1 and 2 in simulation 1, you may have concluded that the Veles archetype is the one to beat.  It’s not.  Even with the higher win percentages of simulation 2, there is a lot of variation between tournaments.  Bigger tournaments with more rounds may lower this variability. Still, be cautious when looking at small samples.
  2. Tournaments often have a few bad decks in the top 16.  Every once and a while, the Yugioh community gets excited about a rogue deck finishing at the top.  Believe me, there are a lot of rogue decks out there, but their tops mean little.  It’s nice to see Evangelion take a few top 16 spots, but it’s still a terrible deck. 

* These names have real meanings.  Have fun looking them up!
**This is a key assumption.  I will look at the influence of a deck's popularity in the future

No comments:

Post a Comment