Provide tournament data in a parsable format for data analysts to use.
From their own requests, two major projects:
1) card popularity: for that I need decklists, that did well in the event. My current definition of 'doing well' is basically at least 2 wins during swiss. I also cut at below 30% winrate, but that only makes a difference for events with 7 or more swiss rounds which rarely happens.
Plan: Provide additional information for each player:
decklist submission
Discord tags
WOTC ID.
Optional commander input or parsed from moxfield
2) match up data: for that I need pod compositions and result of pods. I can already work with a subset of information so if you only have commanders and winner, I can already use that. If you also include names of players, I can feed that into ELO scores. Bonus points if the names match the names on topdeck. Otherwise it's impossible to connect. More bonus points if starting players is known. That way I can use it to better estimate seating order bias. For most events on topdeck I can't guarantee seating order so data there is less reliable.
Plan:
For each player, parse the commander (or input it manually) (same as above)
Starting player information - depends on acceptable implementation of seat normalization.
Provide tournament data in a parsable format for data analysts to use.
From their own requests, two major projects:
1) card popularity: for that I need decklists, that did well in the event. My current definition of 'doing well' is basically at least 2 wins during swiss. I also cut at below 30% winrate, but that only makes a difference for events with 7 or more swiss rounds which rarely happens.
Plan: Provide additional information for each player:
2) match up data: for that I need pod compositions and result of pods. I can already work with a subset of information so if you only have commanders and winner, I can already use that. If you also include names of players, I can feed that into ELO scores. Bonus points if the names match the names on topdeck. Otherwise it's impossible to connect. More bonus points if starting players is known. That way I can use it to better estimate seating order bias. For most events on topdeck I can't guarantee seating order so data there is less reliable.
Plan: