• Pojo Home Page
  
• Message Board
  
• Contact Staff
  

June 3, 2005

The Great Yu-Gi-Oh! Simulation Study
Recruiting and Randomization

So now you know all about the study, here’s where we’re at right now.

At the time I am writing this article, 42 duelists have volunteered for the study.  About 1/3 have reported the key confounding demographic information: Age and Years of Experience.  Randomization for 40 duelists ran on the second of June, and the extra two duelists who joined on June third were randomized on that day.

Randomization is a way for us to adjust for potential bias in the results.  Bias occurs when there is a systematic problem with the study.  For example, if I put all of the least experienced people in the control group, I might see a lower winning percentage in that group simply because they do not know the game as well as the people in the experimental group.  Once I have complete demographic information, I will compare the two groups on the two measures, to see if there is any significant difference in their average ages or experience.  If there is, then we will have to worry about that in the analysis later on.

Randomization was carried out by generating 40 group indicators, 20 for each group.  By randomly sorting  the indicators, and merging them onto the recruiting file, I effected a random assignment of group to each duellist in the study.  In addition, the first round pairings were posted, and the study went live at 8:00 pm on the second of June.

To properly assess the effect of large deck size, we are beginning with something simple: taking an already existing deck and playing it against itself, where one of the versions of the deck has an additional card, with the same additional card in all of the experiment group duelists’ decks. The deck comes from the Texas Shonen Jump Champion, Ryan Hayakama.  Here is the deck list:

Monsters: 17
1 Tribe-Infecting Virus
1 Sinister Serpent
1 Berserk Gorilla
2 D.D. Assailant
1 Blade Knight
1 D.D. Warrior Lady
2 Magician of Faith
1 Airknight Parshath
1 Black Luster Soldier – Envoy of the Beginning
1 Jinzo
1 Breaker the Magical Warrior
2 Kycoo the Ghost Destroyer
1 Spirit Reaper
1 Sangan

Spells: 16
2 Scapegoat
2 Nobleman of Crossout
1 Metamorphosis
1 Mystical Space Typhoon
1 Pot of Greed
1 Snatch Steal
1 Lightning Vortex
1 Heavy Storm
1 Creature Swap
1 Smashing Ground
1 Graceful Charity
1 Premature Burial
1 Enemy Controller
1 Delinquent Duo

Traps: 7
1 Call of the Haunted
1 Bottomless Trap Hole
1 Dust Tornado
1 Torrential Tribute
1 Sakuretsu Armor
1 Ring of Destruction
1 Mirror Force

The additional card in the experimental deck is Reinforcement of the Army.

To simplify the situation, we are allowing an infinite fusion deck, to allow for the fact that anyone who would play Metamorphosis would have access to the desired fusions, although some argue that the only use will be for Thousand Eyes Restrict. Here, we decide to err on the side of caution, since it is irrelevant to the study itself.

So now, the duels are going on as we speak.  Duelists are testing these constructions hundreds of times, to see if there will be any effect for the extra card.  At the end of this phase, we will look at additional cards in the deck (extra monsters, magic or trap), and try to discover empirically if there truly is an optimal deck size.  We also hope to replicate the experiment using some alternative deck types.

The results of this study could change the way we play the game, and I, for one, am looking forward to the results.  Stay tuned, constant reader, for the answers are on the way.

Send correspondence to ygoprofessor@bellsouth.net. I don’t recommend you send me decks to rate or fix. I do welcome discussion on the experimental design and data analysis plan, if you are interested. I don’t really want any hate mail, but if you feel you must, I can’t stop you.  Just put “HATE MAIL” in the subject heading when you send it so I’ll know where to look for it. Thanks for reading. Until next time – same Bat-time, same Bat-channel.