In the past, I've often seen PvP projects developed by teams that typically use ELO as their primary matchmaking solution.📝 This always seemed a bit as a risky choice because ELO is certainly a great system, but it was created for other purposes. 🌟First of all, good things about it (in games): — It can work and cover a couple of MM basic requirements — For games with 1vs1 PvP, with some custom adjustments, ELO can work really good But in my opinion when MM systems are going to design & development phases, it should be built based on detailed requirements from the game perspective. Otherwise it may be hard to adapt games and ELO to each other, to make it work really well. 🔖What else is important to consider when there is a task to make ELO-based MM (or similar system): — If the game is about a bigger format than 1vs1, it should be handled in some way by additional logic (e.g. based on weighted averages). — Some calibration mechanics can be useful. Because some players can be very good from the start or learn to be that way very fast. — Rating “degradation” through inactivity time could be handy. Some players may leave the game with a high rating and return to it when they forgot how to play at the same level. — Some kind of “Seasons” with rating downscaling at the start (in proportional way) may be useful. — Custom rules for onboarding of new players. For example, average rating gain at the start could be higher than zero. It will help to differentiate layers with players of different skill levels. — If the game is based not only on the skill, but on the power progression as well, it also should be handled (e.g. in the form of the gearscore, affecting the ELO rating in direct or indirect form). ⚒️At the end of design, it is helpful to ask yourself: is the final system transparent and flexible enough? Is it easy to operate? After reflection about these questions some simplification needs could appear in your head. Or some other rating systems may be taken into consideration.
Elo-based MM is just a collective image for any skill-based MM, honoring Elo's contribution to the subject. In reality even Chess.com (the very primary user of Elo) uses Glicko system, which is significantly more advanced (and still pretty old). Proof. https://support.chess.com/en/articles/8566476-how-do-ratings-work-on-chess-com#:~:text=Chess.com%20uses%20the%20Glicko%20rating%20system
What do you think about layering players based on their victory per game ratio? In games with team fights ELO might be hard to balance. There should also be the question of roles in a given team. In GW2 I often find myself losing because the team lacks certain key roles (like enough burst damage to pressure nodes, or class with mobility to rotate on the map and decapture nodes)
You should check out the posts by Charlie Olson There are way better means of skill based matchmaking than ELO - its hardly 'great'.
Placement Matches for every season should be on the list as well, and work similar to how new players are treated in the system, by being less strict, the longer (either based on time, number of patches, or the like) the player hasn’t played. Perhaps the rating also needs a “consistency certainty” value, to judge how sure the game is about the rating. Obviously, this should be low for new players – but still used in the overall calculations. Besides that, it’s certainly a per game thing, in regards to the first point. 🤗
🎮 Gaming Entrepreneur | Co-founder & Game Lead en Evil Zeppelin
1moThe main objective of using the ELO System is to give a reward based on the level of the players. Obviously the ELO System also use to classify and order players (using points earned/lost for rewards) and with this you can establish an MM System (matching the players with similar points). But if your objective is MM, using a system ranking, for example based on leagues, would be “sufficient”. Precisely the ELO System works and exists if MM does not exist, that is, a system that determines your opponent based on your player level. The ELO System can work in a PvP system that does not exactly use an MM system, such as in an open world MMO where players of all levels can converge. Basically the ELO System harms the player with the highest level and benefits the player with the lowest level. This way, a higher Level Player does not take advantage of a lower Level Player. It doesn't bring you any benefit.