Forums / Community / Matchmaking Feedback & Discussion

[Locked] Matchmaking Feedback Update – Feb 13

OP ZaedynFel

  1. 1
  2. ...
  3. 2
  4. 3
  5. 4
  6. 5
  7. ...
  8. 8
Sp00kyM0nk wrote:
Anything in the works for a skill "buffer" zone? IE: should Platinum players be able to play againt Silvers?
I'm curious about that to. Sometimes I face plat level players while only a mid to high silver player. I even played against a plat player while I was a bronze player once.
AgentTEX41 wrote:
Why rotate breakout and piss off half the audience every time you change it when we can have a ranked with new rules and social with classic rules
I am good with idea also.
Surrender, Leaderboards, and Playlist-based XP ranks are all great things I'd like to see done, but I totally understand waiting until H6 for those changes. If I could pick one for h5 it would be surrender.

Reducing the size of Onyx: Yes please. The idea of top 2% being onyx feels pretty right to me, makes onyx roughly the equivalent of a 50 in H3 which I think is a good place for it.

Seasonal emblems/unlocks as a reward for high-level players: I love this idea, but I would caution against making these unlocks too difficult to acquire for lower level players. I would suggest 100% chance for Champs to receive most recent unlock, 50% for Onyx, 20% for Diamond, and anyone who doesn't get the most recent one would get the most recent unlock they don't yet have. That way you are at most a month behind, and the completionists among us don't have to worry about having an unlock they can't get.

Party restrictions: This seems to me like a fair compromise. I would prefer pure party restrictions, as I'm sure many would, but I understand we don't quite have the population to pull them off. My question is how do you handle To2s/To3s? I imagine that advantage is a little more difficult to quantify, but if you don't take it into account, rank-focused players might just switch to the next largest team size that isn't seen as a significant advantage.
ZaedynFel wrote:
ZaedynFel wrote:
Just to clarify - you're saying Halo 3's per playlist rankings (as in 1-50) were gated on skill, not the progression ranking which was as simple as just winning a game, getting XP.
Ah, no, what I mean was that the military ranks you were granted based on XP were gated on your 1-50 skill.

For example, even if you had enough XP to go from Lieutenant to Captain, it wouldn't let you unless you had at least a 20 for skill.

You could go deep within a rank instead with your XP, but that was sort of a badge of shame. Like, being a First Lieutenant meant you would never be Captain and your skill was definitely below 20.

That's what I was asking. I personally would prefer not gating on skill within the progression system. Maybe I would consider giving out XP faster to higher-skilled players so they get to XP ranks faster.
Ah i see. i was confused by that bit as well. Progressions XP should not be gated by skill in any way. I like to keep "Skill" and "Progression" systems separate.
I don't see any need to give out progression XP faster for higher-skilled players. They are going to be winning a larger percentage of their games already, plus they tend to put more value on the skill rank designation. Not that you need design input :) but i would do it like this:
  • 1 win = 1 point
  • No skill requirement for any progression rank
  • Rewards are SYMBOLS not numbers. Symbols "feel" more meaningful. The current SR system just feels.... meh.
  • Use this for enticing players with double XP weekends and the like.
Rajit123 wrote:
ZaedynFel wrote:
Leavers and SurrenderI have a couple of follow up points here based on some of your comments and questions.
  • Can you tell the difference between network issues and a normal quit? Yes, but we can’t tell the difference between network issues and someone pulling the plug. If you know all you have to do to avoid losing CSR and getting banned is to pull the plug instead of the usual quit, what would you do? Because of this, we have to penalize equally.
  • Can you have a less harsh CSR penalty for surrendering? No, because then everyone would collude to surrender if it looked like they were going to lose. This would make figuring out when exactly we allow a surrender much harder. We have to treat a surrender the same as a normal loss. Any variation on this can lead to exploits. For example, if you lost less points for surrendering because 2 people left the match, then it would be advantageous to convince two teammates to leave so you could surrender and take a smaller hit to your CSR.
(2 / 2)
something i want to point out regarding the difference between network disconnect and pulling the plug, i tested this out back when it was very common for me.
A point on interest being that the message that i get in response to each of the 2 types of disconnects is very different, with the wording of each clearly showing that the game/console recognized the source of the problem (xbox has no internet as compared to lost connection to game server)

maybe im wrong, but this observation on my part makes me believe that some degree of differentiation between connection loss and pulling the plug does exist/can be created, preventing innocent and unfortunate players from being penalised with a ban/loss of CSR.
Being able to differentiate locally doesn't necessarily mean that the server can always tell the difference. There are also ways to spoof different types of connection issues. For example you can bridge your console connection through your PC, then use your PC to manipulate that connection in any myriad of ways. There are too many uncontrollable variables for them to do anything with disconnects other than to treat them like quits.
Thanks for the info OP.
Quote:
If your opponent can only beat you 1 in 30 games, then you should only get 1 point for beating them, assuming your current CSR is in the right place. Conversely, if you beat a team that usually beats you 29 out of 30 times, you should get 29 points, and they should lose 29 points.
In order for it to actually be accurate, it needs to account for:

  • HOW MUCH the win is by... a 50-49 game shouldn't affect ranks as much as a 50-20 game
  • If a teammate quits or is AFK within the FIRST HALF of the game
Edit: Also, be very careful about how the changes apply to FFA
FFA ranks already tend to be more volatile/fast-changing than other playlists, and game finishes more inconsistent.
It needs to be tested completely separately from the rest of ranks.

And since this is related indirectly, some people want 6-man and some people want 8-man. I know it's not an 'even' number but 7-man FFA is really where it should be at.
I know this isn't a playlist feedback thread, but the number of people in FFA greatly affects how the ranks work out, so it matters.
Edit 2: And top 3 should count as a 'win' on your service record. It's discouraging for to play FFA when you know it will result in a 10-40% win rate in the long run, despite the fact that you're still beating half the players in the match just like team games. (now, admittedly, I'm getting a bit off-track, sry.. but yeah, it's still all related particularly since we're discussing matchmaking 'rewards' and service record displays)
Quote:
Ranking Rewards
We’ve talked previously about both promoting Team Arena as the main playlist for Ranking, and also making Champion more rewarding. I’m currently looking into what we have right now that we could give out as exclusive Ranking system rewards. My current thinking, which is still very rough, is to do something like this, only for Team Arena:
  • Have a season reward based on your final CSR Rank.
  • Top Champions always get the current season reward
  • Lower ranks have a much smaller chance to get the current reward
  • Lower ranks have a much higher chance to get the previous season rewards
  • Example: Diamond players have a 10% chance of getting last season’s reward, and a 40% chance of getting the reward from two seasons ago, but almost no chance of getting the current season reward.
This sounds like a great way to deal with it tbh.

Though one thing that definitely needs to be done (though I'm not sure if it's possible for H5, it at least needs to be in H6)
is to put a list of past champion ranks clearly displayed on the MAIN IN-GAME Arena ranks page.
There's room to add another column to the right-hand side of the page, or maybe another "card" (like the KDA/Assault/Slayer etc. boxes) specifically for past highest ranks.
ZaedynFel wrote:
Party Restrictions.....
Example:
  • Fireteam A has 4 of the best players in the whole system. They each have a solo skill of 4000, so the team average is 4000.
  • Today, Halo 5 would try to find any group with an average of 4000. Including 4 solo players with a skill of 4000 each.
  • But we believe that group of solo 4000-players would get destroyed by Fireteam A.
  • Instead of telling the matchmaker Fireteam A is 4000, tell the matchmaker Fireteam A is 4500. Remember 4175 would be a 1 in 100 chance, so 4500 is a huge gap.
  • Now the matchmaker will not consider a 4500 vs. a 4000. It will look for 4 solo 4500 players or another full fireteam where the individual players are 4000.
  • But there ARE no 4500 solo players, they don't exist. The only way for an even match would be another 4000-rated full Fireteam with that +500 boost (exactly like party restrictions).
  • So the system makes the 4500 team wait a long time because there aren't any 4500s
  • We could allow the Fireteam to eventually match the 4000-rated solos.
The use of average in the match making algorithm still bothers me. The "bronze but really onyx" accounts can manipulate this system way too easy. One or two of them in a T04 will skew the average significantly rendering your party up CSR penalty pointless. I'm not saying they shouldn't be able to play together, but it should be matching (or at least trying to match) from the top rank, not the average.
ZaedynFel wrote:
CSR RanksIn those cases where top Onyx and Champion players cannot find a fair match and are matched against easily defeated opponents, they will not get many points. This is needed to maintain the skill integrity of the top of the ladder. The amount of points you get for defeating a team needs to be proportional to how difficult your opponent is. If your opponent can only beat you 1 in 30 games, then you should only get 1 point for beating them, assuming your current CSR is in the right place. Conversely, if you beat a team that usually beats you 29 out of 30 times, you should get 29 points, and they should lose 29 points. This keeps the system accurate.

We also understand the concerns over the size of Onyx. The original intention I had for Onyx was to be only the top 2% or so players. It looks like it may be larger than intended, so I may cut back in the near future. This may mean that most of you that are currently Onyx would drop to Diamond, so I’d like feedback on how you would feel about that. On the one hand, it makes getting to Onyx more rewarding and a clearer indication of Halo mastery. On the other, it feels bad to lose it. One possible option is to only make this more exclusive change for Team Arena.
I'm very excited to see the results of the re-tuning, and strongly support adjusting the CSR gained/lost more steeply based on the skill gap between teams. That's the inherent beauty of an ELO-based system, which really wasn't shining through enough in recent seasons. It seemed like my rank was mostly a function of my initial placement (which still has major accuracy issues) and number of games played. With this adjustment, your rank should much more accurately reflect the contribution you're making to help your team win against good competitors.

I would recommend you guys allow that change (and the other big ones you're talking about) to settle for a season or two, and see how it affects the ranking distribution, before making an additional big change that reduces the number of onyx players. My rationale: H5 is a game with a HUGE skill gap, and thus the ranking distribution needs to have the flexibility and space to accommodate a wide range of skills. If you artificially "shift or narrow the CSR bell curve leftward" to restrict the number of players who reach Onyx, that's going to have cascading effects down through all the lower ranks. And it may mean that you lose any room at the low end to separate beginner players of widely varying skills (for example if all the current bronze/silver/gold/plat ranks get squished into bronze/silver range).

Just my $0.02...

Thanks for soliciting feedback!
ZaedynFel wrote:
Party Restrictions.....
The use of average in the match making algorithm still bothers me. The "bronze but really onyx" accounts can manipulate this system way too easy. One or two of them in a T04 will skew the average significantly rendering your party up CSR penalty pointless. I'm not saying they shouldn't be able to play together, but it should be matching (or at least trying to match) from the top rank, not the average.
This has always bothered me as well. One solution would be to treat the lower-ranked fireteam member(s) as higher ranks, for the purposes of CSR calculations, for example:

  • Set some max CSR differential that is considered legitimate for a To4 in MM (let's say 500)
  • Assume some To4 has CSRs: 2200, 2100, 2000, 1000.
  • Matchmaking treats the 1000 CSR player as a 1700 (i.e. 2200-500)
  • All matchmaking algorithms and post-game CSR adjustment algorithms act as though the CSR 1000 player had CSR 1700.
This wouldn't eliminate smurfing or boosting issues, but it would decrease their impact.
Can we get that party restriction system for some of the social playlists preferably BTB for example? I know it's social, but 98% of the time when I match full parties it's just a blowout game in favor of the full party. It mainly just results in massive quitting on the solo team's side (JIP people quit as well) and the people that stay end up getting slaughtered. It's not fun and it's not competitive.
yes!! Please rotate old breakout in the playlist please
As a 2v2 and FFA guy, I'm really hoping every non-Arena playlist doesn't take a back seat in the future.
This post has been edited by a moderator. Please do not call out individuals or make inappropriate comments.

*Original post. Click at your own discretion.
Spoiler:
Show
Breakout was in the beta and was a launch playlist. It needs to remain as such.
That's not really a good reason to keep something good or bad, but there's obviously a population according to Menke so it doesn't look like it's getting removed any time soon.

Edit: I'll revise and say it's probably a good reason if it's a core playlist that your game is based around like slayer for example.
LUKEPOWA wrote:
Breakout was in the beta and was a launch playlist. It needs to remain as such.
That's not really a good reason to keep something good or bad, but there's obviously a population according to Menke so it doesn't look like it's getting removed any time soon.
sorry for some of us this is the reason we want to play this game.
For Arena, have you considered team ranking and an individual ranking. As you say some teams hold advantages, a good indicator as to actual skill is to monitor and report team and individual ranking.
great idea! that would offer encouragement and reward for me and countless others to stay in otherwise horrid games. we get indie stats that might actually indicate something apart from team wins minus quits and reduce said quits at the same time... i love it.

spinning off... track and show player's value in past 2-3-4 man FTs apart from random play in some capacity.

also... as i'm sure you're aware.... bronze-silver-gold are literally infested with sandbagging smurfs that just wreck those ranks. smurf detection and deterrence would be on top of my wishlist. if you took the h5 games i've played to this point (somewhere around 3000 arena matches) and removed just half the occurrences of the following entirely-too-common situations then we're getting somewhere entirely worthwhile (as in i would literally never leave a game):

unranked tankers (either squad of course),
rank backtanking after wrecking for a level (using a variety of means but all in common with this guy.... maybe ban him a little more eagerly and energetically than the others),
pairing me in consecutive games with a player that for whatever reason was horrid and finished dead last in the prior game (either that or give me a consolation gold pack...lol)

final thought: 3 betrayals in a game = 1 hour ban the first time, 8 hours the next, 24, 48, etc... (...and a gold pack for the other guys... #what!)
Awesome update Josh. Great to see the direction you're considering. :D
When did the original Rule that CSR can't fall below a rank Tier 1 from losing - get broken?
I've seen players be Diamond 1 in one match, then they lose and are in the next game as Platinum 6. This should not happen, according to the originally described rules of CSR - you can only lose tiers within a rank.
LUKEPOWA wrote:
Breakout was in the beta and was a launch playlist. It needs to remain as such.
That's not really a good reason to keep something good or bad, but there's obviously a population according to Menke so it doesn't look like it's getting removed any time soon.

Edit: I'll revise and say it's probably a good reason if it's a core playlist that your game is based around like slayer for example.
To me i would think of it as a core playlist. As i said before, in the beta and a launch playlist.
I have played halo for so long, the other playlists have been played out for me. Breakout was fresh and new.
I think i like it so much because i am a huge CTF fan.
To touch on other points:
Breakout shouldn't be a playlist, period. In its time as a playlist in Halo 5:- Released as a ranked playlist in October 2015, population quickly died because it was stale and the meta couldn't ever evolve (just like Walshy told them during the development process as seen on The Sprint).
- Eventually new maps were added to the playlist and it was turned into a social playlist because it likely didn't have enough players and resulted in slow matchmaking times
- Someone decided that instead of reducing this playlist to a rotational one, it needed more resources dedicated to it with a full rework completely removing the uniqueness of Breakout, managing to piss off the current Breakout fans and for some reason making it ranked again. Breakout should have taken a short time out, brought back in as a rotational social playlist constantly swapping with other playlists like Assault (4v4, please!), Triple Team, Super Fiesta, etc. I bring up what Quinn DelHoyo said when Monitor's Bounty was released - "Halo 5's ranked playlists will be approached and handled more sacredly, and with a higher degree of scrutiny." Breakout existing as a ranked playlist completely goes against what he said in December. I saw that some underpopulated social playlists got the axe recently (and will be returning as rotational playlists) - this should be the case here. Breakout should return as a rotational playlist which features both old and new Breakout.
Please don't ever listen to moa about breakout. nothing he says makes any sense at all. Breakout was in the beta and was a launch playlist. It needs to remain as such. Moa has a big boner for not liking it, good don't play it and move on with your life. There still remains a good sized population that wants to play it. Also F walshy.
If there was a good sized population wanting to play it it wouldn't have been the lowest populated playlist Holiday 2015 and it wouldn't have been moved to social and reworked entirely.
  1. 1
  2. ...
  3. 2
  4. 3
  5. 4
  6. 5
  7. ...
  8. 8