Toward Better Matchmaking
(Long post, got a bit carried away since I’ve worked with matchmaking systems in my day job. Posting here in the hope that the devs at least give it some consideration. Hopefully other players find this stuff as interesting as I do!)
Territory Wars is my favorite part of this game so I'm excited to get more content like that with the upcoming Grand Arena mode! However, like many other players, the description given in the latest Road Ahead is a little concerning:
This isn’t very detailed so there’s still a lot we don’t know about how exactly this matchmaking system would work. All we really know is (1) players are placed into an 8 person tournament based on GP, and (2) the tournament is then played using a round-robin format. Unfortunately, we already know matchmaking based on GP is inherently flawed because GP doesn’t do a good job of reflecting the “strength” of players’ rosters. Here are some of the problems:
Characters vs. Ships: My GP is ~2.8m and about 1.2m of that is from ships. Someone with much stronger characters but a weaker roster of ships would crush me in a 1v1 TW-style battle (assuming the map has a similar ratio of character to ship territories). This could be alleviated by treating ship power separately from total GP in their matchmaking algorithm, but we will have to wait and see.
Wide vs. Deep: Some players maintain a “wide” roster with many characters at G7/8 and 7 stars because this helps them score higher in TB and can be required to meet some guild’s GP requirements. This is a disadvantage in Grand Arena because a player with a ”deep” roster (characters are either highly geared characters or untouched G1s). A matchmaking system that is purely based on GP does not distinguish between these types of players and puts the “wide” players at a disadvantage.
Mods: Equipped mods raise GP so players can intentionally tank their GP before matchmaking by removing mods, then put them back on during the setup phase. I’m actually not too concerned about this because it costs credits and seems like a pain, but I’m sure some players will do it anyway. This could also be fixed by locking in players’ mods when matchmaking begins.
So what is the alternative? Something like Elo ratings would be a good start. Elo ratings and similar competitive ranking systems have been used in everything from chess to Counterstrike. CG could match players based on GP during the first Grand Arena, then use their performance to match them in all subsequent events. Players who rack up lots of wins would have their rating increase and then be matched with similarly rated players in the next event. Those who end up with lopsided matches and lose would move down the ladder and face similarly weaker opponents. This could even include a GP-based limit to prevent highly rated players from facing opponents with impossibly stronger rosters just because they are good at Grand Arena.
This benefits the players, but what about CG’s bottom line? Is it worth the hassle?
Cost: I obviously don’t know how things work behind the scenes at CG, but this shouldn’t take much dev time to implement. Elo ratings are well established, simple to compute, and boil down to a simple 4 digit integer (so it shouldn’t be a burden on the servers to store these values). Although this could be used to create a “ladder” system with higher reward tiers in top tier matchups, it is still compatible with the tournament-based reward system that it looks like they will be using for Grand Arena.
Benefits:
Pure GP matchmaking actually *punishes* some players who spend money on the game! “Collector”-style players who want to have all the characters at 7 stars/Level 85 and buy all the new marquees can be easily beaten by F2P players with similar GP who instead focus their limited resources on a deep roster of only the strongest squads. Better matchmaking *rewards* collectors by placing them in fairer matchups where they have a shot at winning top tier rewards.
Highly competitive players would have an even stronger incentive to buy and gear up the latest and greatest characters to keep up with their opponents and continue winning the highest tier of Grand Arena rewards. CG could even display player ratings on their profile to create a “prestige” incentive to stay highly rated.
Conclusion
In short, CG’s approach to matchmaking focuses on GP which is flawed for a number of reasons. It would be better, and more profitable, to implement a rating system. This avoids demoralizing players with lopsided matchups (as we’ve seen in some cases with TW matchmaking) and incentivizes spending. Win-win!
Territory Wars is my favorite part of this game so I'm excited to get more content like that with the upcoming Grand Arena mode! However, like many other players, the description given in the latest Road Ahead is a little concerning:
At the beginning of the Grand Arena Event, you’ll be matched with 7 other players from across the game who have a Galactic Power score very similar to your own.
Each round, you’ll be matched with other players who have the same win/loss record as you, and you’ll never fight the same opponent twice in a Grand Arena event.
This isn’t very detailed so there’s still a lot we don’t know about how exactly this matchmaking system would work. All we really know is (1) players are placed into an 8 person tournament based on GP, and (2) the tournament is then played using a round-robin format. Unfortunately, we already know matchmaking based on GP is inherently flawed because GP doesn’t do a good job of reflecting the “strength” of players’ rosters. Here are some of the problems:
So what is the alternative? Something like Elo ratings would be a good start. Elo ratings and similar competitive ranking systems have been used in everything from chess to Counterstrike. CG could match players based on GP during the first Grand Arena, then use their performance to match them in all subsequent events. Players who rack up lots of wins would have their rating increase and then be matched with similarly rated players in the next event. Those who end up with lopsided matches and lose would move down the ladder and face similarly weaker opponents. This could even include a GP-based limit to prevent highly rated players from facing opponents with impossibly stronger rosters just because they are good at Grand Arena.
This benefits the players, but what about CG’s bottom line? Is it worth the hassle?
Cost: I obviously don’t know how things work behind the scenes at CG, but this shouldn’t take much dev time to implement. Elo ratings are well established, simple to compute, and boil down to a simple 4 digit integer (so it shouldn’t be a burden on the servers to store these values). Although this could be used to create a “ladder” system with higher reward tiers in top tier matchups, it is still compatible with the tournament-based reward system that it looks like they will be using for Grand Arena.
Benefits:
Pure GP matchmaking actually *punishes* some players who spend money on the game! “Collector”-style players who want to have all the characters at 7 stars/Level 85 and buy all the new marquees can be easily beaten by F2P players with similar GP who instead focus their limited resources on a deep roster of only the strongest squads. Better matchmaking *rewards* collectors by placing them in fairer matchups where they have a shot at winning top tier rewards.
Highly competitive players would have an even stronger incentive to buy and gear up the latest and greatest characters to keep up with their opponents and continue winning the highest tier of Grand Arena rewards. CG could even display player ratings on their profile to create a “prestige” incentive to stay highly rated.
Conclusion
In short, CG’s approach to matchmaking focuses on GP which is flawed for a number of reasons. It would be better, and more profitable, to implement a rating system. This avoids demoralizing players with lopsided matchups (as we’ve seen in some cases with TW matchmaking) and incentivizes spending. Win-win!
