

Tier System Too Limited
#1
Posted 27 October 2015 - 08:43 AM
To my understanding, this is due to me being teamed with any player from T4 up to T1, this includes the tryhards and also the total new players.
Now, if I'm seeing this correctly, we should have at least double the amount of Tiers.
If we consider >50% T1 to be the new T1 and <50% T5 to be the new T10, I'd be placed at Tier 4 (<50% T2).
Now, the MM when looking for a team, would start to search for Tiers the same as mine (T4) and when not finding, increasing the range, but since the Tiers would be much better separated, an old T4 player would now be either T8 or T7, so the MM would leave the T8 for last when searching for a team.
This would at the very least, leave the least balanced match for latest to be picked.
In this example, the MM would look for T4, then T5 (>50% old T3) and T3 (>50% old T2), then T6 (<50% old T3) and T2 (<50% old T1), then T7 (>50% old T4) and T1 (>50% old T1)... and so on.
So with this, the chances of finding better matches to be made would increase.
What do you guys think?
#2
Posted 27 October 2015 - 08:45 AM
#3
Posted 27 October 2015 - 08:48 AM
Lugh, on 27 October 2015 - 08:45 AM, said:
The waiting times are not good, why would that be a problem? Wait times and unbalanced matches are mostly due to low active (and available - looking for match) pop at the moment of the search. All I propose is the idea of refining the Tier system for the (potential) benefit of the match quality.
#4
Posted 27 October 2015 - 09:03 AM
FlipOver, on 27 October 2015 - 08:48 AM, said:
It's true that it wouldn't really affect wait times much, just segregate people a little more precisely, but I doubt it would improve your match quality much either. As stated above, the issue is really player population, there just aren't enough other players of a skill level similar to yours so the matchmaker opens the release valve and starts looking further afield. In 1 out of 20 matches, yes, matchmaker would find 23 other people of your exact skill level (say tier 4/10 in the new system) and you'd have a very competitive match but 19 out of 20 times it will do the same thing it does now and throw you with the 23 people closest to you to get the match started... with varying results. Really doesn't solve anything.
#5
Posted 27 October 2015 - 10:43 AM
There needs more ranks or buckets.
queue time can be fixed otherways. Like my suggestion thread - being able to queue as many mechs as you want at once, or all weight groups at the same time... therefore matchmaker has more choices and can make games faster.
my suggestion thread for faster queues
http://mwomercs.com/...ost__p__4752584
Edited by Karamarka, 27 October 2015 - 10:46 AM.
#6
Posted 27 October 2015 - 10:58 AM
#7
Posted 27 October 2015 - 11:09 AM
MM still hasn't changed much in the balancing of W/L ratio - it still tries to get that to 1/1. Sadly, that means in T2 you are going to get a full team of T4 guys and face a group of T2 and T1 guys to make sure you lose to get your win/loss back to 1/1. At least it seems that way sometimes. Every time I get a group and I see a jenner with a ppc and a LRM 5 on my team, I wonder why the MM hates me so much. Gah.
#8
Posted 27 October 2015 - 11:19 AM
However it is the population that's causing the wait times by being to small.
According to Russ in a twitter, its all I.G.P's fault.
It wasn't I.G.P that did a power point talking about how great Cw was going to be, and we got a map with dots, that was Apparently delayed by making the very bad U.I. M.k.2 I can't remember if that was I.G.P's fault, but I'm sure Russ would say it was.
However it wasn't I.G.P that designed and introduced P.S.R
It wasn't I.G.P that put even more weight restrictions on Group play, which did nothing to stop the stomps, as the team of hunchbacks we fielded proved conclusively.
All evidence points to, in fact that it's P.G.I's lead developers fault.
#9
Posted 27 October 2015 - 11:21 AM
Without a BattleValue (BV) type system that balances based on mech loadout AND PSR, we will continue to be split into these buckets.
If we allow a BV system where pilots can downgrade their builds into less min/max killers and take more mixed loadouts in exchange for a lower teir game, it would allow players to play for fun again and not feel like they are gimping their team in the process.
#10
Posted 27 October 2015 - 12:53 PM
FlipOver, on 27 October 2015 - 08:43 AM, said:
To my understanding, this is due to me being teamed with any player from T4 up to T1, this includes the tryhards and also the total new players.
Now, if I'm seeing this correctly, we should have at least double the amount of Tiers.
If we consider >50% T1 to be the new T1 and <50% T5 to be the new T10, I'd be placed at Tier 4 (<50% T2).
Now, the MM when looking for a team, would start to search for Tiers the same as mine (T4) and when not finding, increasing the range, but since the Tiers would be much better separated, an old T4 player would now be either T8 or T7, so the MM would leave the T8 for last when searching for a team.
This would at the very least, leave the least balanced match for latest to be picked.
In this example, the MM would look for T4, then T5 (>50% old T3) and T3 (>50% old T2), then T6 (<50% old T3) and T2 (<50% old T1), then T7 (>50% old T4) and T1 (>50% old T1)... and so on.
So with this, the chances of finding better matches to be made would increase.
What do you guys think?
So Halo 5 just launched...their CSR system, the equivalent of PSR, has ranks associated as 'Iron -> Bronze -> Silver -> Gold -> Platinum -> Onyx -> Champions' then inside of the named tiers for ranking it goes 1 through 3 for the sub ranks. That makes out 18 actual tiers for matchmaking.
But the counterpoint is it has millions of players, while we have thousands...So the scaling there makes a big difference.
#11
Posted 27 October 2015 - 01:04 PM
FlipOver, on 27 October 2015 - 08:43 AM, said:
...
Since we don't have all the data, there is no way to know for sure how well the tiers reflect skill ... however ...
1) Assaults and heavies tend to do more damage and receive higher match scores as a result in most of the games I have played. Lights typically have lower match scores but there are exceptions like many ACH I have seen recently.
2) LRM boats can have high damage and high match score also.
3) Match score is a significant contributing factor to PSR.
4) Your PSR always goes up on a win and can either stay even or go up on a loss with a high enough match score.
5) There are more opportunities for PSR to go up than down and PSR is not conservative meaning that the net change in PSR in most matches is positive across all players.
6) This leads to the PSR score being more strongly influenced by the number of games played rather than the actual player skill. In fact, almost everyone will be able to grind to tier 1 or tier 2 given a long enough time played. (Paul even said this in one of his posts about the PSR system).
What does all of this mean? " it's not uncommon for me to be teamed with either much better or much worst players." ... this is the likely result if PSR doesn't actually work which is what one might conclude from the points above. Most of your opponents may be tier 1 to tier 3 ... but you will find bad players who have played a lot of games in tier 1 and good players who have only played a few and are still in tier 3 or 4. The bottom line is that the number of tiers is absolutely irrelevant if the underlying PSR is broken which I suspect is likely the case.
P.S. The previous Elo based matchmaker typically had teams matched within a few points of Elo in the solo queue. I suspect that the PSR based matchmaking is likely no worse in the PSR spread but unfortunately, the PSR is likely much less representative of actual player skill than Elo was ...
Edited by Mawai, 27 October 2015 - 01:07 PM.
#12
Posted 27 October 2015 - 01:18 PM
Cathy, on 27 October 2015 - 11:19 AM, said:
However it is the population that's causing the wait times by being to small.
According to Russ in a twitter, its all I.G.P's fault.
It wasn't I.G.P that did a power point talking about how great Cw was going to be, and we got a map with dots, that was Apparently delayed by making the very bad U.I. M.k.2 I can't remember if that was I.G.P's fault, but I'm sure Russ would say it was.
However it wasn't I.G.P that designed and introduced P.S.R
It wasn't I.G.P that put even more weight restrictions on Group play, which did nothing to stop the stomps, as the team of hunchbacks we fielded proved conclusively.
All evidence points to, in fact that it's P.G.I's lead developers fault.
"All evidence points to, in fact that it's P.G.I's lead developers fault."
I beg to differ. The lead developer's don't get to make those decisions. They may have an opinion but the decisions on what to implement come from the top in this company (it is a SMALL company). In the last townhall, Russ stated that the tonnage limits for groups in group queue was HIS idea .. he pushed it through despite a lot of feedback about problem ares like the one you cited.
A number of the other decisions regarding game design may also be his, Finally, as the head of the company and acting in such a hand's on role ... he pretty much has to take responsibility for most of the decisions made to address concerns by whatever vocal subset of the community he actually listens to ... since it isn't the folks posting on the forums and apparently hasn't been for years.
#13
Posted 27 October 2015 - 01:26 PM
AlphaToaster, on 27 October 2015 - 11:21 AM, said:
Without a BattleValue (BV) type system that balances based on mech loadout AND PSR, we will continue to be split into these buckets.
If we allow a BV system where pilots can downgrade their builds into less min/max killers and take more mixed loadouts in exchange for a lower teir game, it would allow players to play for fun again and not feel like they are gimping their team in the process.
It would not hurt to try such a system for about a month to see how it works. And your suggestion would have an interesting effect on the Meta (dual gauss + dual PPCs).
#14
Posted 27 October 2015 - 01:49 PM
Cathy, on 27 October 2015 - 11:19 AM, said:
However it is the population that's causing the wait times by being to small.
According to Russ in a twitter, its all I.G.P's fault.
It wasn't I.G.P that did a power point talking about how great Cw was going to be, and we got a map with dots, that was Apparently delayed by making the very bad U.I. M.k.2 I can't remember if that was I.G.P's fault, but I'm sure Russ would say it was.
However it wasn't I.G.P that designed and introduced P.S.R
It wasn't I.G.P that put even more weight restrictions on Group play, which did nothing to stop the stomps, as the team of hunchbacks we fielded proved conclusively.
All evidence points to, in fact that it's P.G.I's lead developers fault.
I am on the verge of making multiple accounts to like this post even more.
#15
Posted 28 October 2015 - 01:35 AM
Mawai, on 27 October 2015 - 01:04 PM, said:
The bottom line is that the number of tiers is absolutely irrelevant if the underlying PSR is broken which I suspect is likely the case.
P.S. The previous Elo based matchmaker typically had teams matched within a few points of Elo in the solo queue. I suspect that the PSR based matchmaking is likely no worse in the PSR spread but unfortunately, the PSR is likely much less representative of actual player skill than Elo was ...
My suggestion here would be to not trash the PSR and try to work with it. Giving it time and a chance to balance matches a bit better.
I still think the PSR needs a few more months and a few more players to work properly, and I do believe it might work.
That's why I imagined a Tier system a bit more specific.
Worst case scenario, wouldn't make any significant difference.
Best case scenario, matches created by the MM would be much more balanced according to the example on the OP.
Of course, all this is depending on the amount of players waiting to get a match at the same time as you.
#16
Posted 28 October 2015 - 02:24 AM
1) Increase PSR range (maybe to 7-10 ranks)
2) Put new players at the lowest rank instead of somewhere higher, it is really a downer for new players to get into the game totally fresh and then getting punished (losing PSR) for beeing newbies, instead of letting them progress upwards from lowest PSR.
3) Change the Matchmaker, so that players of equal rank are matched against each other and mix all PSR tiers in the same matches, instead of excluding highest tiers from lowest. So if one team has a Tier 1 player, the other team also gets one. (I know, this might get a bit complicated with matching PSR tiers and mech classes at the same time, but this should be counterbalanced by the larger pool of players).
#17
Posted 28 October 2015 - 02:28 AM
Lugh, on 27 October 2015 - 08:45 AM, said:
I think the changes made this month and the projected ones for november will cause far more people to leave and increase wait time far more than making more tier buckets
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users