155 replies to this topic

[Gagis]

The most likely player to quit is the player who gets good enough to have "beaten" quick play and has no meaningful content to play other than Comp. the time and scheduling requirements mean that competitions are suitable only for some, so a large number of players at top of tier 1 get bored and quit, leaving a large gap between where they used to be and the best long term competitive players. This means the average skill level of the entire playerbase remains low and keeps getting lower.

What the game would need to prosper is some endgame content that players who have fully learned to play well could graduate into, to find rewarding goals to work towards. Faction Play could have been this, if fully developed, but it is gone now.

Edited by Gagis, 31 March 2022 - 03:01 AM.

[DaZur]

I think some context is being lost in terms of what exactly tier 3 is...

Tier 3 is the composite of players who are good but not good enough. They are either steadily improving and slowing grinding up, maintaining status quo (the vast majority) or on the slide secondary to a volume of reasons. As such, the median skill of involved players is pretty broad, ranging from newly vetted from cadet to grizzled veteran having a bad streak.

Point is, you can play with the flood gates from 5 and 1 and or adjust team sizes all day and there is nothing you can do to alter the core composition of tier 3. They are the mutts of the player base and wild swings in match quality will ebb and sway just within that core demographic. You play during non-peak times when the player pool is low and it will be further influenced by the population release valves.

I don't want to be the negative Nancy but statistical outcome can only be altered so much when you have a sliding scale...

The only real means to influence the quality of median tier match quality is to grossly increase the player base, thus feeding more organic tier 3 level players into the median.

The truly "better" or "worse" players are just passing through...

[dubstep albatross]

DaZur, on 31 March 2022 - 06:23 AM, said:

Tier 3 is the composite of players who are good but not good enough. They are either steadily improving and slowing grinding up, maintaining status quo (the vast majority) or on the slide secondary to a volume of reasons. As such, the median skill of involved players is pretty broad, ranging from newly vetted from cadet to grizzled veteran having a bad streak.

There is a huge amount of variation within any and all tiers. Tier 4 players who just arrived in Tier 4 are quite different than Tier 4 players who are just about to breach Tier 3. We could replace the number with n and the statement holds true. I would also suggest that the skill gap between top of tier and bottom of tier gets worse with an increase in tier.

I think the reason that Tier 3 feels different is the matchmaking system and the tier spread (see below). Tier 3 is the first time a player can be regularly matched against Tier 1 and that can apply a significant amount of downward pressure on one's PSR and game experience, especially if one has just breached Tier 3.

Quote

Point is, you can play with the flood gates from 5 and 1 and or adjust team sizes all day and there is nothing you can do to alter the core composition of tier 3. They are the mutts of the player base and wild swings in match quality will ebb and sway just within that core demographic. You play during non-peak times when the player pool is low and it will be further influenced by the population release valves.

I have observed this ebb and flow. I do not think it is due to the composition of Tier 3 or anything inherent to being a Tier 3 player (whatever that actually means). I think it is due to the matchmaker and the tier spread valving. Tier 3, with +/- 2, can reach into either end of the tier system at any given time. Tier 1 and Tier 3 mixing is much more dramatic than Tier 3 and Tier 5. I would also suggest that the practical effects of increased skill is super-linear (exponential difference, etc.).

Consider if tier spread was limited to +/- 1. Tier 3 players would still get sucked into Tier 2 games, but not Tier 2 games with Tier 1 players. Tier 3 players would also get sucked into Tier 4 games, but not with Tier 5 players present. Tier 3 would suddenly have a very different feel -- less getting smashed on by Tier 1, less running the field on Tier 5.

It is not the player, it is the matchmaker.

Quote

The only real means to influence the quality of median tier match quality is to grossly increase the player base, thus feeding more organic tier 3 level players into the median.

An increase in player base, presuming the same distribution that now exists, would mean that games were much more likely to be (closer to) homogenous with respect to tier. This would improve gameplay quality for everyone, in all tiers. I think everyone intuitively understands that and acknowledges that.

Quote

The truly "better" or "worse" players are just passing through...

That is true for any tier, even including the tiers at the end. The only difference is that there is a mechanical stop at either end. Some Tier 1 players would be Tier 0 and some Tier 5 players would be Tier 6 if the stops were not present.

[LordNothing]

even if the average moves up over time, its still the average.

[DaZur]

dubstep albatross, on 31 March 2022 - 07:00 AM, said:

There is a huge amount of variation within any and all tiers. Tier 4 players who just arrived in Tier 4 are quite different than Tier 4 players who are just about to breach Tier 3. We could replace the number with n and the statement holds true. I would also suggest that the skill gap between top of tier and bottom of tier gets worse with an increase in tier.

I have observed this ebb and flow. I do not think it is due to the composition of Tier 3 or anything inherent to being a Tier 3 player (whatever that actually means). I think it is due to the matchmaker and the tier spread valving. Tier 3, with +/- 2, can reach into either end of the tier system at any given time. Tier 1 and Tier 3 mixing is much more dramatic than Tier 3 and Tier 5. I would also suggest that the practical effects of increased skill is super-linear (exponential difference, etc.).

Understood and agree to a point... Difference is the chasm in player potential (skill) is narrower in 4,5 and 1,2. Tier 3's median (and statistically should be) are "average" players. Being fed from 4 and 2, in particular as you noted, new residents, are effectively similar enough in skill level to not register. I know tier 3 can on occasion, see a 1 or a 5 but IMHO it's not as common as some would infer so I'm not entirely convinced of their influence.

It's the "transition" players that hold the most sway... Using myself as a perfect example... I've been as high as mid tier 2 and have played this game for a long time thus have the experience and knowledge. That said, right now I'm a high tier 4 because of some serious bad streaks and some questionable choices in play. In tier 4 I can directly impact the quality of play and can manifest outcome based on how well or poorly I play.

dubstep albatross, on 31 March 2022 - 07:00 AM, said:

That is true for any tier, even including the tiers at the end. The only difference is that there is a mechanical stop at either end. Some Tier 1 players would be Tier 0 and some Tier 5 players would be Tier 6 if the stops were not present.

I've actually argued that there should not be a hard ceiling or floor...

There are players who as apex tier players, their matches should only valve in similar skilled players. Conversely, there are players that even in tier 5 struggle and they should be gated until they are truly ready for the full battlefield.

Edited by DaZur, 31 March 2022 - 07:51 AM.

[DaZur]

LordNothing, on 31 March 2022 - 07:43 AM, said:

even if the average moves up over time, its still the average.

True... But that is why a sliding scale washes itself. Your median will and should always represent the "average" player. That average can move up or down based on the player skill potential but at the end of the day it's still "average".

[LordNothing]

DaZur, on 31 March 2022 - 07:49 AM, said:

True... But that is why a sliding scale washes itself. Your median will and should always represent the "average" player. That average can move up or down based on the player skill potential but at the end of the day it's still "average".

i sometimes question if part of the reason the average has moved is people setting up alts to scrub their noob data from the record.

though i suspect the cauldron had something to do with it. by buffing a lot of the trashmechs and weapons, people who use those are doing better as a result. the meta becomes more marginal as an advantage. perfect balance would mean no meta, but as good as it is, perfect balance is impossible.

also you play a game long enough you are bound to get better at it. this is the grognard effect. its also part of the reason new players bounce off the game. the bar to entry is higher than it would be with a novice player base. if new players were sticking it would shift the average further down.

Edited by LordNothing, 31 March 2022 - 08:01 AM.

[DaZur]

LordNothing, on 31 March 2022 - 07:52 AM, said:

i sometimes question if part of the reason the average has moved is people setting up alts to scrub their noob data from the record.

I don't think it's even in question... I personally recognize a LOT of alt accounts when I play. Sadly, there is no way to stop this unless PGI audited IP addresses or hardware markers. Even then there are ways around it for the true buttheads.

[MrMadguy]

Vxheous, on 31 March 2022 - 02:25 AM, said:

That graph literally shows the vast majority of players sit between 201-250 matchscore...which is where you also sit, how is that skewing towards "hardcore" players?

Because it's not even Tier 3, that is dead now. It's also Tier 4. Because if even Tier 4 doesn't have players with adequate skill levels, suitable for 50% percentile player, then, I guess, I will be in Tier 5 soon.

[Nightbird]

Vxheous, on 31 March 2022 - 02:25 AM, said:

That graph literally shows the vast majority of players sit between 201-250 matchscore...which is where you also sit, how is that skewing towards "hardcore" players?

MrMadguy, on 31 March 2022 - 10:07 AM, said:

Because it's not even Tier 3, that is dead now. It's also Tier 4. Because if even Tier 4 doesn't have players with adequate skill levels, suitable for 50% percentile player, then, I guess, I will be in Tier 5 soon.

This is why you let people that learned math do math.

Just because the majority of players are between 201-250 avgMS doesn't mean the majority of people in the queue have an avgMS score between 201-250. How this works is simple, one person is one person regardless of how many matches they play in the population level graph, their avgMS is only one data point. However how often a player contributes to the queue depends on how often they play. Because people with high avgMS play more games on average than everyone else, the median avgMS of people in a game is higher than the median across the population, making 201-250 in reality below average and below median in the queue which means downward tier movement.

It's called weighted sampling.

My long term simulation of Jay Z's matchmaking formula shows the center of the population being pushed to T4, and with sufficient time, to T5.

Edited by Nightbird, 31 March 2022 - 10:30 AM.

[Vxheous]

Nightbird, on 31 March 2022 - 10:25 AM, said:

This is why you let people that learned math do math.

Just because the majority of players are between 201-250 avgMS doesn't mean the majority of people in the queue have an avgMS score between 201-250. How this works is simple, one person is one person regardless of how many matches they play in the population level graph, their avgMS is only one data point. However how often a player contributes to the queue depends on how often they play. Because people with high avgMS play more games on average than everyone else, the median avgMS of people in a game is higher than the median across the population, making 201-250 in reality below average and below median in the queue which means downward tier movement.

It's called weighted sampling.

My long term simulation of Jay Z's matchmaking formula shows the center of the population being pushed to T4, and with sufficient time, to T5.

Ok, March just ended for data collecting, so show us the math for the most recent month, and weight it towards games played. Im genuinely curious as to whether or not this game's population actually fits that you're saying. Eyeballing the avgMS of those that play the most, the range looks like the vast majority sit 200-260ish with a few outliers

[Nightbird]

Vxheous, on 31 March 2022 - 03:28 PM, said:

Ok, March just ended for data collecting, so show us the math for the most recent month, and weight it towards games played. Im genuinely curious as to whether or not this game's population actually fits that you're saying. Eyeballing the avgMS of those that play the most, the range looks like the vast majority sit 200-260ish with a few outliers

Jarl's already does it automatically

[Vxheous]

Nightbird, on 31 March 2022 - 04:35 PM, said:

Jarl's already does it automatically

Thats total games played, not monthly games played. Im curious as to current playerbase, as in the last several months

[Nightbird]

Vxheous, on 31 March 2022 - 05:34 PM, said:

Thats total games played, not monthly games played. Im curious as to current playerbase, as in the last several months

Using the regression line, if you dumb down the population to 5000 players with 200avgMS playing 1500 games each, and 5000 players with 300 avgMS and 1800 games played. The average MS of the population is (200+300)/2=250. The average MS in the queue is (200*1500 + 300*1800)/(1500+1800)=254.5

When you move further to the two ends of the regression line, the difference in games played also become more extreme taking the queue average further from the pop average.

No need to scrap leaderboards for this.

[Nightbird]

The difference may seem small, but it means the system is not in equilibrium. People think that players are pushed up down from 2500, but the truth is that 2500 is not the "middle" anymore.

I found an old thread, with predicted PSR across the population
https://mwomercs.com...cy-with-graphs/



See how the middle of the pop is falling from 2500? I expect the middle to be around 2000 today.

[LordNothing]

ive always figured that tiers 2 and 4 were more of a proving ground than any actual tier. always transitionary. from those graphs it looks as if this might be true. you dont see very many t2s and most of the t4s you see are probibly new players who havent bubbled yet.

[dubstep albatross]

DaZur, on 31 March 2022 - 07:46 AM, said:

Understood and agree to a point... Difference is the chasm in player potential (skill) is narrower in 4,5 and 1,2. Tier 3's median (and statistically should be) are "average" players. Being fed from 4 and 2, in particular as you noted, new residents, are effectively similar enough in skill level to not register.

I am not sure I really follow you here. Would it be possible for you to rephrase this or maybe go into a bit more detail?

Quote

I know tier 3 can on occasion, see a 1 or a 5 but IMHO it's not as common as some would infer so I'm not entirely convinced of their influence.

My intuition is that it happens enough to be a factor in feel of gameplay, especially during times of lower population. This is what drives my earlier comments about why tier 3 games feel as they do. I have a non-trivial amount of tier 3 gameplay data, but no reliable way to know what each player's tier was at the time of the match. I believe some have suggested using Jarl's and percentile ranking, but I am still not sure if the correlation is strong enough to place enough players in the appropriate tier to make any analysis credible.

In other threads I have mentioned that I feel that tier 3 fills tier 1 games much more often than tier 3 fills tier 5 games. Nightbird does have an interesting point about the distribution of players versus the distribution of players in the queue. I think tier 1 is the smallest tier, but I do wonder what portion of the queue, at any given time, is tier 1. Intuitively one might be tempted to think that more skilled players would have more active game time, but I am not sure.

Quote

It's the "transition" players that hold the most sway... Using myself as a perfect example... I've been as high as mid tier 2 and have played this game for a long time thus have the experience and knowledge. That said, right now I'm a high tier 4 because of some serious bad streaks and some questionable choices in play. In tier 4 I can directly impact the quality of play and can manifest outcome based on how well or poorly I play.

In your example, it makes sense that you can have an outsized effect as a tier 4 assigned player in a tier 4 game (if you truly are a tier 2 player). I think a case of a player reaching tier 2 and then ending up in tier 4 is unusual. The real question is: where does this player really belong? It can be easy to be mis-tiered, especially in the downward direction. Once the cause of that is sorted (back to a practiced build/playstyle, no longer playing drunk with friends in meme builds, etc.) then that player will rise quickly back to where they should be.

Unintentionally or otherwise, some players can rise into higher tiers through certain dynamics of the match score system. As I mentioned in another thread, I think 8v8 exposes these mis-tiered players and pushes them towards a more appropriate tier (at least for 8v8 games).

So it seems to me that you may be suggesting that tier 3 players tend to be mis-tiered more than other players. There may be something to that. Stack a bunch of AMS and sandpaper damage weapons in tier 4, you will get to tier 3, but you will not likely get to tier 2. In tier 2 if you do not keep up with the meta and you do not execute well, you will end up in tier 3, but not likely let it get to tier 4.

In the scenario I just described, then, it is not really about being average. It is about being mis-tiered. The matchmaker cannot balance mis-tiered players.

Quote

I've actually argued that there should not be a hard ceiling or floor...

There are players who as apex tier players, their matches should only valve in similar skilled players. Conversely, there are players that even in tier 5 struggle and they should be gated until they are truly ready for the full battlefield.

Without stops, what do you do when a player is at 10000 PSR? At some point you will have to re-bucket (or transform) the PSR values in order to find matches or you will have players who can never find matches.

What you may really be suggesting is just having more gradients in the tier system. I would imagine we would need quite a few more active players to support this. If the valves are opening to +/- 1 and to +/- 2, we do not have enough players (with the distribution of players we have) to support even the tiers we have.

[MrMadguy]

My logic is simple. If I'm 100% average player, but keep dropping to Tier 5 - then it's obvious, that everything between my current position and my estimated position in mid Tier 3 - IS DEAD, cuz I keep playing against exactly the same >50% rating guys, no matter, how low my own rating is. MM can barely find players for two simultaneous games, so if I die and quit match early, I have to wait for long time for next match to start. Yeah.

[DaZur]

dubstep albatross, on 31 March 2022 - 08:09 PM, said:

I am not sure I really follow you here. Would it be possible for you to rephrase this or maybe go into a bit more detail?

In short, tier 5 and tier 1 are capped, as such, this tier is primarily fed by 4 and 2 respectively (Not saying exclusively to be clear). So in terms of the potential skill level between 4-5 and 1-2 the skill-gap is narrower and there is a much less skill level fall off. Tier 3 is primarily fed by 4 and 2 thus the potential skill gap is substantially wider, ranging from low 4s to high 3s... The palpable result is the potential skill-gap is huge in contrast to 1s and 5s and manifests in the aforementioned wilds swings in quality of play in tier 3'ish matches.

dubstep albatross, on 31 March 2022 - 08:09 PM, said:

My intuition is that it happens enough to be a factor in feel of gameplay, especially during times of lower population. This is what drives my earlier comments about why tier 3 games feel as they do. I have a non-trivial amount of tier 3 gameplay data, but no reliable way to know what each player's tier was at the time of the match. I believe some have suggested using Jarl's and percentile ranking, but I am still not sure if the correlation is strong enough to place enough players in the appropriate tier to make any analysis credible.

In other threads I have mentioned that I feel that tier 3 fills tier 1 games much more often than tier 3 fills tier 5 games. Nightbird does have an interesting point about the distribution of players versus the distribution of players in the queue. I think tier 1 is the smallest tier, but I do wonder what portion of the queue, at any given time, is tier 1. Intuitively one might be tempted to think that more skilled players would have more active game time, but I am not sure.

I can't speak intelligently in regards to what you experience but I personally don't see it a great deal. To be fair, I usually play during Euro prime-time or North America prime-time so I rarely deal with shallow player pools. Conversely, I could just not be as sensitive to their presence.

dubstep albatross, on 31 March 2022 - 08:09 PM, said:

In your example, it makes sense that you can have an outsized effect as a tier 4 assigned player in a tier 4 game (if you truly are a tier 2 player). I think a case of a player reaching tier 2 and then ending up in tier 4 is unusual. The real question is: where does this player really belong? It can be easy to be mis-tiered, especially in the downward direction. Once the cause of that is sorted (back to a practiced build/playstyle, no longer playing drunk with friends in meme builds, etc.) then that player will rise quickly back to where they should be.

Unintentionally or otherwise, some players can rise into higher tiers through certain dynamics of the match score system. As I mentioned in another thread, I think 8v8 exposes these mis-tiered players and pushes them towards a more appropriate tier (at least for 8v8 games).

Again, agree to a point. MWOs tier mechanics to progress or regress are not symmetrical. It takes notable individual effort to progress forward and or be grouped with a upwardly mobile team to rise in tier rank. Conversely, a string of losses or piloting less-than-optimal mechs (which was my downfall from 2 to 4 btw) can have you backsliding in short order.

dubstep albatross, on 31 March 2022 - 08:09 PM, said:

So it seems to me that you may be suggesting that tier 3 players tend to be mis-tiered more than other players. There may be something to that. Stack a bunch of AMS and sandpaper damage weapons in tier 4, you will get to tier 3, but you will not likely get to tier 2. In tier 2 if you do not keep up with the meta and you do not execute well, you will end up in tier 3, but not likely let it get to tier 4.

In the scenario I just described, then, it is not really about being average. It is about being mis-tiered. The matchmaker cannot balance mis-tiered players.

To the contrary... I believe the vast majority of tier 3 players are where they belong. It's the "transitional players" i.e. players who's functional skill level is above or below the median tier 3 players that are "passing through" tier 3 either on their way to tier 2 or tier 4 respectively.

dubstep albatross, on 31 March 2022 - 08:09 PM, said:

Without stops, what do you do when a player is at 10000 PSR? At some point you will have to re-bucket (or transform) the PSR values in order to find matches or you will have players who can never find matches.

What you may really be suggesting is just having more gradients in the tier system. I would imagine we would need quite a few more active players to support this. If the valves are opening to +/- 1 and to +/- 2, we do not have enough players (with the distribution of players we have) to support even the tiers we have.

You are correct.. I was kind'a vague in my explanation. Yes, there should be a more tiers ( 11? maybe ) with the top 1/3rd and bottom 1/3rd gated. This would allow for a more granular transition in skill-gaps (Mitigating the injection of non-compatible skill levels at non-peak times) keeps fresh cadets and apex players from comingling.

Obvious downside to this is it would directly impact wait times and doesn't affect say a tier 1 player riding the coat tails of a tier 5 player in group.

This is why the only true solution is a huge increase in the potential player base which just isn't going to happen for a myriad of reasons.

[pbiggz]

MrMadguy, on 01 April 2022 - 05:29 AM, said:

My logic is simple. If I'm 100% average player, but keep dropping to Tier 5 - then it's obvious, that everything between my current position and my estimated position in mid Tier 3 - IS DEAD, cuz I keep playing against exactly the same >50% rating guys, no matter, how low my own rating is. MM can barely find players for two simultaneous games, so if I die and quit match early, I have to wait for long time for next match to start. Yeah.

You are absolutely not a 100% average player, and you play at way off hours.