8 replies to this topic

[Renthrak]

The idea is fairly simple. Rather than having the matchmaker take individual player ELOs and team averages as the basis for finding players, group players into brackets for a range of ELO and match players within the same bracket.

ELO ranges from 0 to 2800. An example of brackets would be:

#1: 0 to 350
#2: 351 to 700
#3: 701 to 1050
#4: 1051 to 1400
#5: 1401 to 1750
#6: 1751 to 2100
#7: 2101 to 2450
#8: 2451 to 2800

So, a player with an ELO of 1550 would be in Bracket #5. For matchmaking purposes, the server only needs to know that he is in Bracket #5, and will start trying to match with other Bracket #5 players. If there aren't enough players in Bracket #5 to fill the team, the matchmaker expands the range to include Bracket #4 and Bracket #6.

For grouped players, the matchmaker should use the Bracket of the highest ranked player in the group to avoid the possibility of exploits. This would allow veteran players to take less experienced pilots with them to gain experience against stronger opponents, while preventing powerful teams from taking a single player with extremely low ELO to skew the team average and match them with weaker teams.

For individual player progression, this would change things up a bit. A player with an ELO of 1752 would regularly be matched against a player with an ELO of 2099, so they would regularly face more skilled players. For a player with an ELO of 2050, however, they would usually be matched against players with a lower ELO. As soon as a player transitions from one Bracket to another, however, the situation would immediately reverse. Rising to a higher Bracket would then match the player against more skilled opponents on a regular basis, and falling to a lower Bracket would match the player against less skilled opponents more often.

Calculating changes in ELO at the end of the match would be unaffected. Only the initial matchmaking would use the Brackets. This means that it becomes more difficult to increase your ELO at the high end of your Bracket, and more difficult to reduce ELO at the low end of the bracket. To avoid becoming completely stuck, I would suggest altering the scoring so that the ELO always changes by +/- 0.1 or something. Due to this, the starting ELO should not be at the half-way point. Instead, I think beginning in Bracket #2 or #3 would be preferable.

This would change the experience for players, but there is also a major benefit for the server: Matchmaking speed should be improved, possibly to a significant degree. The reason being that instead of having to take individual player ELOs, find players in a range around that score, average the team scores and try to match another group doing the same, the matchmaker would only need to consider the Bracket number. This should reduce the overhead for the matchmaker even with weight matching still in effect. With less time spent trying to match ELOs, the weight matching would have more influence, which should allow the tonnage deviation to be narrowed without creating unreasonable wait times.

Edited by Renthrak, 08 June 2013 - 12:14 PM.

[A Man In A Can]

Completely agree with this idea. It's a win-win all around.

Edited by CYBRN4CR, 10 June 2013 - 04:57 AM.

[Mechteric]

8 is probably too many divisions, I think 4 would do the trick, if not 3 minimum.

[Seddrik]

Dunno. I dont like Elo as it is, but not sure the OP post would be much better. My suggestion is to have ONLY a new person status for 7-10 days, where they get primarily matched with only new players where possible, then willing in from there. After that I'd prefer weight balancing only.

I want real battles, not fixed battles. Seems like after winning 3-5 times in a row, I always get shoved on worse and worse teams till I am forced to lose due to teammates.

Edited by Seddrik, 10 June 2013 - 07:15 AM.

[Lord of All]

Considering the matchmaking system already uses the ELO rating all your really asking is to be shown what bracket everyone is in instead of what there rating is. But I still voted yes and no to the public ELO disclosure. Go figure.

[Skoaljaw]

+1

[Levi Porphyrogenitus]

Elo brackets could be implemented several different ways and every single one of them would improve MM.

They could do 1:1 bracketing, where they pick one person for each team from the same bracket, and the overall team is bracket-independent.

They could do team:team bracketing, where everyone on both teams are drawn from the same bracket.

They could do lance:lance bracketing, where they pick four players from a bracket for each team, but both lances on a team don't need to be from the same bracket.

And so on and so forth.

[Renthrak]

Lord of All, on 10 June 2013 - 07:57 AM, said:

Considering the matchmaking system already uses the ELO rating all your really asking is to be shown what bracket everyone is in instead of what there rating is. But I still voted yes and no to the public ELO disclosure. Go figure.

Not quite. Instead of having the matchmaker take your exact ELO and match that every time you search for a game, your Bracket number would be stored instead. Basically, it's taking all of the ELO math out of the process of finding a game. All the MM has to do is compare your Bracket number to other players to find eligible people, then check the weight to pick balanced teams. Only in the event of insufficient players in your Bracket would the MM consider the next higher and lower Brackets.

Basically, the Brackets would determine who you should be matched with ahead of time.

Note that there is no need to reveal the Bracket number to the players. If players want a general idea of their skill ranking, however, revealing only the Bracket number would still hide your ELO score.

Edited by Renthrak, 10 June 2013 - 02:06 PM.

[Blood78]

What ever it is, they need to make it public. I'm tired of you suck, I'm better than you debates but most of all. I want transparency on MWO's matchmaking system which we can't really verify without publically visible ELO or ladder rank of all the players.

http://mwomercs.com/...o-team-listens/