A quick note/musing on the Worlds simulation Adam posted about below: With no disrespect to Lukas Blohon, it’s obvious that in no realistic model is he 16% to win the tournament. So why did the simulation come out that way, and does it mean that the expected win percentages that our model is assuming are pretty far off?

Well, the short answer is no. When we look at all matches in our dataset between “veteran” players (three events or ten matches played), players whose rating is 195-205 points higher than their opponent have a 58.89% win percentage, which is very close to what the model expects. The problem is that at any given moment in time, the rating of any player who has just won a tournament or had a couple of deep runs is inflated a bit above their equilibrium point. When we take a snapshot of the current ratings, and then run a simulation forward only sixteen matches, the effect of that inflation is exaggerated.

There would probably be a more “polls-plus” way to try to simulate the outcome of a given tournament, adjusting for recent big swings in rating, but it would take (a) a larger dataset (ours is still relatively small for the moment), and (b) more time. For now, if you’re using our site to help make your MTG Worlds fantasy draft picks (as I have been!), take the actual match/tournament history data seriously, and the win probabilities as entertainment. ^_^