Duelyst Patch 1.73 S-Rank Changes
We have implemented changes to how ranking are calculated within S-Rank, after processing a lot of feedback about issues with the current system. Here’s a broad summary of the issues we considered as part of making changes:
- S-Rank players can be matched with Diamond players when queues are slow and when Diamond players are close to S-Rank.
- “Farming” Diamond Division later in the season during slow hours was perceived to be a strategy.
- Losing to Diamond Division players earlier in the season would deliver a significant hit to your standing, but have no effect on the Diamond player who could conceivably reach S-Rank in time over the season.
- Because of above, Players entering S-Rank early in the season could be at a higher risk than players entering S-Rank later in the season.
- Entering S-Rank with a strong 10 game streak could result in a very high position.
- This could encourage players to stop playing and hope to preserve their high spot due to rank loss anxiety.
- The window of wins required to “solidify” a player’s rating is low.
To re-iterate for the record, here are our objectives for S-Rank:
- It should be an approximate measure of skill. (Duelyst is not a fully deterministic game)
- It should not reward endless grinding over skilled play. (Quality of play over pure quantity of play)
- It should not matter when during the season you achieve S-Rank and when you have time to play, as long as we have enough data from your S-Rank games to estimate your skill level.
- The point in time you enter S-Rank during the season should not bestow an advantage or disadvantage.
- If you win an S-Rank game, your hidden rating will improve. Assuming the rankings have not been affected by other games ending or by new players entering S-Rank, your rank will likely improve as a result of your win.
Here’s what we’re doing:
- Diamond players will receive a more dynamic hidden rating so that playing against Diamond players for early S-Rankers at the beginning of the season does not have as deleterious of an effect.
- We are adding an additional parameter related to win count within S-Rank, which will contribute to determining your S-Rank ladder position. This parameter has severely diminishing returns after a certain number of wins.
- The hidden rating of Diamond players will impact (to a degree) how well they place when they enter S-Rank.
- We are adding a live real-time ladder for S-Rank. When you check your profile, it will update in realtime.
S-Rank will never be perfect for all types of players. We’ve considered a number of different models, and this is the one we feel that best fits both our objectives and a diverse set of players. We believe these changes are a step to get us closer to our objectives while softening some of the issues the community is feeling. We’re going to continue to tune the algorithm based on how well these changes perform over the October season.
NOTE: Let’s address a concern we did not mention above: if a player has a 30 win streak in S-Rank and reaches a high spot: we do not consider that pure luck. We’ve heard feedback that when someone does exceptionally well over a set of games, it must be luck, and we therefore must allow a grind that gives other players the opportunity to catch up. The “luck” argument posits that consistency over a huge number of games should be rewarded over exceptional performance in a medium set of games. We don’t think this approach is fair to the spirit of S-Rank as an approximation of skill that does not require an unreasonable time commitment.
Let us know what you think.
As always, dive over into the forums, discord, or reddit to let us know your thoughts.