Forums / Community / Matchmaking Feedback & Discussion

[Locked] MATCHMAKING FEEDBACK UPDATE – November 5

OP ZaedynFel

  1. 1
  2. ...
  3. 22
  4. 23
  5. 24
  6. 25
  7. 26
  8. ...
  9. 27
ZaedynFel

Been having a lot of quitters/AFK in games recently. Played some Team Arena yesterday, these were my first five games.
  • Game 1: Starts 2v4 - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/f3b62936-fa22-42e0-959a-1f202b2d3f6c/players/nightclerk?gameHistoryMatchIndex=14&gameHistoryGameModeFilter=Arena
  • Game 2: One teammate AFK from the start, another quits. Then we all quit - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/397787f2-eb98-4d82-9c0d-ef761a34eaba/players/nightclerk?gameHistoryMatchIndex=13&gameHistoryGameModeFilter=Arena
  • Game 3: Success! Full teams for entirety of game. No AFKs. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/1b99c0fe-aec4-465e-b9e3-835515fc9aa2/players/nightclerk?gameHistoryMatchIndex=12&gameHistoryGameModeFilter=Arena
  • Game 4: Started with 1 or 2 AFK teammates who eventually quit. We all quit after they leave. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/5e96d950-5af5-40c4-a693-f1dd4dd99822/players/nightclerk?gameHistoryMatchIndex=11&gameHistoryGameModeFilter=Arena
  • Game 5: 2 teammates quit early. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/8c23ea97-1541-44b7-898b-4a49c092c637/players/nightclerk?gameHistoryMatchIndex=10&gameHistoryGameModeFilter=Arena
Fortunately I believe the ranking system worked and my CSR loss was pretty minimal. I think the penalties for quitting and being AFK aren’t strong enough to discourage this behavior. I will say that the matches I get are generally very competitive and TS2 finds balanced games. I’m sure you understand the frustration of going into game after game only to have that time wasted with a quitter.

Has there been any talk of implementing some type of “trust score” that tracks how often people quit? With a score like that, quitters could be matched with other quitters. Rewarding “good” players and punishing quitters in one motion.

This isn’t a criticism of TS2 or the MM system, which seems to be working very well. Just frustrated with these experiences I'm sure you can relate to. Keep up the good work.
NightClerk wrote:
ZaedynFelBeen having a lot of quitters/AFK in games recently. Played some Team Arena yesterday, these were my first five games.
  • Game 1: Starts 2v4 - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/f3b62936-fa22-42e0-959a-1f202b2d3f6c/players/nightclerk?gameHistoryMatchIndex=14&gameHistoryGameModeFilter=Arena
  • Game 2: One teammate AFK from the start, another quits. Then we all quit - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/397787f2-eb98-4d82-9c0d-ef761a34eaba/players/nightclerk?gameHistoryMatchIndex=13&gameHistoryGameModeFilter=Arena
  • Game 3: Success! Full teams for entirety of game. No AFKs. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/1b99c0fe-aec4-465e-b9e3-835515fc9aa2/players/nightclerk?gameHistoryMatchIndex=12&gameHistoryGameModeFilter=Arena
  • Game 4: Started with 1 or 2 AFK teammates who eventually quit. We all quit after they leave. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/5e96d950-5af5-40c4-a693-f1dd4dd99822/players/nightclerk?gameHistoryMatchIndex=11&gameHistoryGameModeFilter=Arena
  • Game 5: 2 teammates quit early. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/8c23ea97-1541-44b7-898b-4a49c092c637/players/nightclerk?gameHistoryMatchIndex=10&gameHistoryGameModeFilter=Arena
Fortunately I believe the ranking system worked and my CSR loss was pretty minimal. I think the penalties for quitting and being AFK aren’t strong enough to discourage this behavior. I will say that the matches I get are generally very competitive and TS2 finds balanced games. I’m sure you understand the frustration of going into game after game only to have that time wasted with a quitter.

Has there been any talk of implementing some type of “trust score” that tracks how often people quit? With a score like that, quitters could be matched with other quitters. Rewarding “good” players and punishing quitters in one motion.

This isn’t a criticism of TS2 or the MM system, which seems to be working very well. Just frustrated with these experiences I'm sure you can relate to. Keep up the good work.
Unfortunately quitting isn't a small group of players that be isolated, it's most players.

Players quit when they are clearly overmatched. Either at the team or individual level, and it's a large portion of the population.
ZaedynFel wrote:
qlimm wrote:
ZaedynFel wrote:
qlimm wrote:
Chimera30 wrote:
qlimm wrote:
ZaedynFel wrote:
qlimm wrote:
GameI was partied up with the other Onyx player, then we got a Plat 6 and a Diamond.

Surely the other squad, comprised of three champs and an Onyx, had a higher MMR than us. I am also inclined to believe it was a squad of four because of how they spawn trapped us.

I go an even 7-7 and the rest of my team got destroyed. The Diamond player eventually gave up and went AFK.

I lose 30 CSR. This is the max, right? Why?

This is the kind of result that makes people stop playing ranked.
Your team had a 1518 MMR, the other team 1516.

A 1660 losing to a 1516 will generally drop 30 like you did.

Your post-game MMR was 1410, so losing -30 is definitely the right step to get your CSR down to where it should be after that performance (you played like a 1389 or Diamond 4).
How could their team's average be 1516 when three of them were Champion which is 1800s for CSR? Maybe I am wrong about this and the Champion threshold is lower at the moment? Did the champions have a MUCH lower MMR than their CSR would otherwise indicate? They'd have to have Diamond MMR levels to get an average of 1516.

This sounds retaliatory, childish, and I'd be just plain venting...but I can't not say it: if the system put a Diamond 4 (apparently how I played) in that match instead of me they sure as heck wouldn't break even.
Champs are just the top 200 Onyx players. So the top 200 players with CSR above 1500 will be Champs. They don't need CSR >1800 to be Champs. If that enemy team's average MMR was 1516, then they could have all just had MMRs around the Onyx threshold; none of them would have had to have Diamond MMR for that average to be possible. It is possible, though, that they had inflated CSRs.
I suppose the fact that my MMR dropped 250 points for one game and is now 210 points below my CSR happened to the other team their previous game? That is the only explanation because to have an average MMR of 1516 you must have players on the team below this value.

Currently Champ #140 has a CSR of 1880 so an estimate of the Champ CSR being 1800 is somewhat reasonable. There were three >1800 CSR players on the enemy team and their MMR somehow found its way down 300 points or more? Is MMR supposed to be this volatile?
Depending on the playlist and player, MMR can be quite volatile. SWAT tends to be one of those because of the large swings you can have game to game despite similar teams.

You were 1466 pre-game, so you were probably dropping recently and CSR is catching up.

Your opponents ranged 1455 -1619 in MMR, so are probably in similar situations.

The 1516 is also pre-game.

Since your opponents also had CSRs above their MMRs, most of them only got 1 CSR for the win. One did get 8.
This makes me wonder...I am a "safe" player in SWAT in that I rarely go negative (probably less than 5 times this season, definitely less than 10) and this is reflected somewhat in my 1.748 K/D, but in order to preserve this I may get fewer kills per game so I never "lose" the game but I don't necessarily win it. Is this playstyle stagnating my progression because while I have teammates who go 6-20 or 0-12 I should be killing the enemies faster before my teammates die that many times?
Well, kpm leads to more wins in the data than K/D ironically enough, so if you can increase your kpm against the same difficulty of opponents, you should see more wins and hence a higher CSR.

Though if you increase kpm but end up against worse opponents, it won't change anything.
Does k/d lead to more wins than dpm? Are you using kpm vs raw k/d data or do you filter out the unreliable k/d numbers? If so, what do you use?
Basically higher KPM leads to more wins than a higher K/D. KPM is a more accurate predictor of who will win the match versus K/D. It is counterintuitive but considering H5 is all about pressuring spawns it makes sense that more kills per minute means more pressure and less time on death screen/losing.
Sorry if this has been suggested before, but can you guys bring back headhunter mode from Halo Reach? it was one of my favorite gametypes, but its no longer playable in the halo reach playlists
ZaedynFel wrote:
NightClerk wrote:
...[player][/player]
Unfortunately quitting isn't a small group of players that be isolated, it's most players.

Players quit when they are clearly overmatched. Either at the team or individual level, and it's a large portion of the population.
This sounds like throwing in the towel on stopping quitters because the problem has gotten so out of hand. Unfortunately, I'm not surprised it's most players. But I'd prefer trying to improve and reverse the situation or at least reward players who don't quit/quit the least. Sounds like right now we're resigning and saying, "it is what it is, can't do anything about it."

I understand it's a hard situation. Hopefully with a larger population in Infinite more options will be available.
NightClerk wrote:
ZaedynFel wrote:
NightClerk wrote:
...[player][/player]
Unfortunately quitting isn't a small group of players that be isolated, it's most players.

Players quit when they are clearly overmatched. Either at the team or individual level, and it's a large portion of the population.
This sounds like throwing in the towel on stopping quitters because the problem has gotten so out of hand. Unfortunately, I'm not surprised it's most players. But I'd prefer trying to improve and reverse the situation or at least reward players who don't quit/quit the least. Sounds like right now we're resigning and saying, "it is what it is, can't do anything about it."

I understand it's a hard situation. Hopefully with a larger population in Infinite more options will be available.
Well, there's no way to legit stop anyone from quitting a game, no matter what 343 tries to implement. If they were to ever decide to be heavy-handed in their approach to dealing with quitting (i.e. removing the option to quit entirely once a game starts), people will either just dashboard out of the game (like they do already pregame) and perhaps just give up the game entirely because that system is in place. W/L record and K/D ratio is a huge deal to a lot of people and quitting out of a game isn't a knock on your record and preserves your K/D if a game looks like it's about to get ugly. So, looking at it that way, there's actually more incentive to quit than stick out a game to the bitter end, and that might be the biggest issue when it comes to combatting quitting. People avoiding that nice L on their record.

So long as that doesn't change, I highly doubt quit rates will get any lower with a larger population. Dem stats, mang.
NightClerk wrote:
ZaedynFel wrote:
NightClerk wrote:
...[player][/player]
Unfortunately quitting isn't a small group of players that be isolated, it's most players.

Players quit when they are clearly overmatched. Either at the team or individual level, and it's a large portion of the population.
This sounds like throwing in the towel on stopping quitters because the problem has gotten so out of hand. Unfortunately, I'm not surprised it's most players. But I'd prefer trying to improve and reverse the situation or at least reward players who don't quit/quit the least. Sounds like right now we're resigning and saying, "it is what it is, can't do anything about it."

I understand it's a hard situation. Hopefully with a larger population in Infinite more options will be available.
Well, there's no way to legit stop anyone from quitting a game, no matter what 343 tries to implement. If they were to ever decide to be heavy-handed in their approach to dealing with quitting (i.e. removing the option to quit entirely once a game starts), people will either just dashboard out of the game (like they do already pregame) and perhaps just give up the game entirely because that system is in place. W/L record and K/D ratio is a huge deal to a lot of people and quitting out of a game isn't a knock on your record and preserves your K/D if a game looks like it's about to get ugly. So, looking at it that way, there's actually more incentive to quit than stick out a game to the bitter end, and that might be the biggest issue when it comes to combatting quitting. People avoiding that nice L on their record.

So long as that doesn't change, I highly doubt quit rates will get any lower with a larger population. Dem stats, mang.
There will always be quitters, and sometimes real life happens and you have to quit to take care of something. But those should be isolated incidents.

Right now, ranked is playable, but the quit rate is beyond what I consider reasonable. A good place to start would be setting up a scoring system which ranks players based on the likelihood they finish the game, and match them with other players who finish games, and match the frequent quitters with each other (within acceptable MMR parameters of course).

This would reward players who finish games and pseudo-punish players who don't finish games. Also, MM could put the quitters on the same team so pop isn't fragmented and matches are found faster.
NightClerk wrote:
Right now, ranked is playable, but the quit rate is beyond what I consider reasonable. A good place to start would be setting up a scoring system which ranks players based on the likelihood they finish the game, and match them with other players who finish games, and match the frequent quitters with each other (within acceptable MMR parameters of course).

This would...
[This would] not only cause some player base fracturing, but pairing players on quit rates over MMR will cause less balanced matches which in turn will cause more quits. You've essentially created a system that likely causes more quitting in a self-propelling manner.
eLantern wrote:
NightClerk wrote:
Right now, ranked is playable, but the quit rate is beyond what I consider reasonable. A good place to start would be setting up a scoring system which ranks players based on the likelihood they finish the game, and match them with other players who finish games, and match the frequent quitters with each other (within acceptable MMR parameters of course).

This would...
[This would] not only cause some player base fracturing, but pairing players on quit rates over MMR will cause less balanced matches which in turn will cause more quits. You've essentially created a system that likely causes more quitting in a self-propelling manner.
I should add it would put quitters on teams with other quitters to prevent fragmentation. Also, I said within MMR parameters. MMR would take precedent over the score assigned by the quit tracking system.

Edit: I got my threads confused. In the post quoted I addressed the fragmentation question.
NightClerk wrote:
eLantern wrote:
NightClerk wrote:
Right now, ranked is playable, but the quit rate is beyond what I consider reasonable. A good place to start would be setting up a scoring system which ranks players based on the likelihood they finish the game, and match them with other players who finish games, and match the frequent quitters with each other (within acceptable MMR parameters of course).

This would...
[This would] not only cause some player base fracturing, but pairing players on quit rates over MMR will cause less balanced matches which in turn will cause more quits. You've essentially created a system that likely causes more quitting in a self-propelling manner.
I should add it would put quitters on teams with other quitters to prevent fragmentation. Also, I said within MMR parameters. MMR would take precedent over the score assigned by the quit tracking system.

Edit: I got my threads confused. In the post quoted I addressed the fragmentation question.
The fact that it attempts to divide or segment players in any measurement beyond MMR is a form of fragmenting that's not in line with creating the most balanced matches. Data overwhelmingly shows that balanced matches (50:50 odds) result in less quit behavior.

I read where you said "within acceptable MMR parameters" but that essentially prioritizes quit rates over MMR up to a certain quality threshold that the system then is told to enforce. This in turn will cause more matches to be created not on the merits of ideal skill balance, but on quit behavior. Then as the population lessens more matches are likely created closer to the MMR threshold which leads to more quitters. As I said, it's essentially a self-propelling system that leads to more quitters as matches become less likely to have predictive odds around 50:50. The system needs to match solely on skill (MMR) in order to reduce the likelihood of undesired behavior.
eLantern wrote:
NightClerk wrote:
eLantern wrote:
NightClerk wrote:
Right now, ranked is playable, but the quit rate is beyond what I consider reasonable. A good place to start would be setting up a scoring system which ranks players based on the likelihood they finish the game, and match them with other players who finish games, and match the frequent quitters with each other (within acceptable MMR parameters of course).

This would...
[This would] not only cause some player base fracturing, but pairing players on quit rates over MMR will cause less balanced matches which in turn will cause more quits. You've essentially created a system that likely causes more quitting in a self-propelling manner.
I should add it would put quitters on teams with other quitters to prevent fragmentation. Also, I said within MMR parameters. MMR would take precedent over the score assigned by the quit tracking system.

Edit: I got my threads confused. In the post quoted I addressed the fragmentation question.
The fact that it attempts to divide or segment players in any measurement beyond MMR is a form of fragmenting that's not in line with creating the most balanced matches. Data overwhelming shows that balanced matches (50:50 odds) result in less quit behavior.

I read where you said "within acceptable MMR parameters" but that essentially prioritizes quit rates over MMR up to a certain quality threshold that the system then is told to enforce. This in turn will cause more matches to be created not on the merits of ideal skill balance, but on quit behavior. Then as the population lessens more matches are likely created closer to the MMR threshold which leads to more quitters. As I said, it's essentially a self-propelling system that leads to more quitters as matches become less likely to have predictive odds around 50:50. The system needs to match solely on skill (MMR) in order to reduce the likelihood of undesired behavior.
"The fact that it attempts to divide or segment players in any measurement beyond MMR is a form of fragmenting that's not in line with creating the most balanced matches. Data overwhelming shows that balanced matches (50:50 odds) result in less quit behavior."
  • Fragmentation would be minimal to nonexistent because low score players would still be in matches with high score players, just on different teams. Matches would still be based on MMR. See below.
"I read where you said "within acceptable MMR parameters" but that essentially prioritizes quit rates over MMR up to a certain quality threshold that the system then is told to enforce. This in turn will cause more matches to be created not on the merits of ideal skill balance, but on quit behavior."
  • I don't think I stated it explicitly previously: MMR would take priority over this new score and unbalanced matches would not result from the system. I never said the new score would take priority over MMR and would not support the new system taking priority over MMR.
"As I said, it's essentially a self-propelling system that leads to more quitters as matches become less likely to have predictive odds around 50:50. The system needs to match solely on skill (MMR) in order to reduce the likelihood of undesired behavior."
  • Match odds would remain within current acceptable values.
NightClerk wrote:
"The fact that it attempts to divide or segment players in any measurement beyond MMR is a form of fragmenting that's not in line with creating the most balanced matches. Data overwhelming shows that balanced matches (50:50 odds) result in less quit behavior."
  • Fragmentation would be minimal to nonexistent because low score players would still be in matches with high score players, just on different teams. Matches would still be based on MMR. See below.
How are the matches *primarily* based on MMR if you're allowing player quit rate "scores" to have influence over the team picker/balancer verses what the ideal MMR breakdown would dictate?
Quote:
"I read where you said 'within acceptable MMR parameters' but that essentially prioritizes quit rates over MMR up to a certain quality threshold that the system then is told to enforce. This in turn will cause more matches to be created not on the merits of ideal skill balance, but on quit behavior."
  • I don't think I stated it explicitly previously: MMR would take priority over this new score and unbalanced matches would not result from the system. I never said the new score would take priority over MMR and would not support the new system taking priority over MMR.
Again, how does MMR take *priority* over this new quit rate "score"? Please, explain more clearly how you perceive the system methodically working on a systematic level.
Quote:
"As I said, it's essentially a self-propelling system that leads to more quitters as matches become less likely to have predictive odds around 50:50. The system needs to match solely on skill (MMR) in order to reduce the likelihood of undesired behavior."
  • Match odds would remain within current acceptable values.
I understand that the system could potentially enact a quality threshold based on MMR just as it does now by preventing odds beyond 61:39. But, how else will MMR take priority over this new quit rate metric's influence with regards to the team picker/balancer system?
How are the matches *primarily* based on MMR if you're allowing player quit rate "scores" to have influence over the team picker/balancer verses what the ideal MMR breakdown would dictate?
  • Matches would continue to be based on MMR. Other scores would NOT have influence OVER the MMR balance. When possible and within acceptable MMR parameters, teams would be assigned to match quitters on the same team, or put quitters into matches against each other. This would satisfy MMR balance requirements. Matches would remain balanced.
Again, how does MMR take *priority* over this new quit rate "score"? Please, explain more clearly how you perceive the system methodically working on a systematic level.
  • MMR remains the primary driver of matches and team balancing. When possible, quitters would be matched together in a lobby or on the same team. The system would scan the pool of searching players and establish matches with quitters playing quitters or quitters playing against "good standing" players. MMR will continue to take precedent and some matches may have to forgo the "quit score" factor to keep game within acceptable prediction percentages.
I understand that the system could potentially enact a quality threshold based on MMR just as it does now by preventing odds beyond 61:39. But, how else will MMR take priority over this new quit rate metric's influence with regards to the team picker/balancer system?
  • MMR is and will remain the primary driver of MM and balancing matches. Players who quit frequently would be put in a lobby or at least on the same team. However, matches would not be made unbalanced because of "quit score."
I understand how some things I suggested could be misconstrued as a "quit score" taking precedent over balanced matches which is not the goal or how the system would function in practice. Hopefully, those concerns and others have been addressed. The goal is a better MM experience for players who play games through to the end.

My suggested system is basically a version of the "Trust Factor" system Valve uses for cheaters in CS:GO. Low trust factor players match with other low trust factor players. The system has generally been considered a success.
NightClerk wrote:
How are the matches *primarily* based on MMR if you're allowing player quit rate "scores" to have influence over the team picker/balancer verses what the ideal MMR breakdown would dictate?
  • Matches would continue to be based on MMR. Other scores would NOT have influence OVER the MMR balance. When possible and within acceptable MMR parameters, teams would be assigned to match quitters on the same team, or put quitters into matches against each other. This would satisfy MMR balance requirements. Matches would remain balanced.
When you say "When possible" what specifically are you saying?

So from my understanding, and Josh can correct me if I'm missing critical info regarding the systematic steps, the Matchmaker first checks your selected search preference when you begin a search.

Based on the Focused, Balanced, or Expanded options the Matchmaker will alter what data centers it'll use to locate players for constructing a match.

After that Ping and MMR take priority.

On the MMR side, if you're searching as a solo player it'll look to find players around your individual MMR and if you're searching with a group your fireteam's average MMR (+ adjusted boost) is used as the target.

If MMR is turned-off for the Matchmaker it means that the search process begins with the allowable odds restriction and player skill-gap at their max limits. Player skill-gap limits may differ per the environment (ranked, social, warzone) and even perhaps the individual playlists too. The max limits still prevent the matchmaker from grabbing players that the team picker/balancer can't optimally place onto teams. In other words, it'll ignore players who would cause the match odds to become 61:39 or greater even in the most fair circumstances skill-wise.

If MMR is active for the Matchmaker it means that the search process begins with the allowable odds restriction at around 50:50 before gradually expanding. Also, the skill-gap will begin much closer to your individual skill (as a solo searcher) or the skill-gap of your fireteam before gradually expanding. The initial restrictions may be different per the environment with the ranked environment often being more restrictive. Same for the skill-gap. Plus, the rate of expansion for the limits are often different too between the environments with social tending to expand quicker.

So how and when should a "trust metric" factor into this process?
Quote:
Again, how does MMR take *priority* over this new quit rate "score"? Please, explain more clearly how you perceive the system methodically working on a systematic level.
  • MMR remains the primary driver of matches and team balancing. When possible, quitters would be matched together in a lobby or on the same team. The system would scan the pool of searching players and establish matches with quitters playing quitters or quitters playing against "good standing" players. MMR will continue to take precedent and some matches may have to forgo the "quit score" factor to keep game within acceptable prediction percentages.
You say MMR should remain the *primary* driver. So, essentially what you're advocating for is exactly the process that takes place now, but in the instance (which I imagine is extremely rare) where a player, or players, may be interchange-able between two optimally balanced teams (per MMR) to let the "trust metric" pair them to the side that's more fitting to their personal trust metric.

In other words, you believe 343i should invest resources into developing and incorporating a "trust" metric formula [based on quit behavior] that would only rarely have any real influence; therefore, I think it's fair to question what actual amount of positive impact that investment would actually provide. I'd guess some, but overall it'd probably be fairly negligible.
Quote:
I understand that the system could potentially enact a quality threshold based on MMR just as it does now by preventing odds beyond 61:39. But, how else will MMR take priority over this new quit rate metric's influence with regards to the team picker/balancer system?
  • MMR is and will remain the primary driver of MM and balancing matches. Players who quit frequently would be put in a lobby or at least on the same team. However, matches would not be made unbalanced because of "quit score."
You see here you make it seem as if you'd rather see the trust factor metric take some precedence over MMR by the team picker/balancer as long as the max limits relating to match quality aren't infringed upon. And this is how I was perceiving your previous posts. It's just not very clear as to what you're asking for here without presenting thorough details.
Quote:
I understand how some things I suggested could be misconstrued as a "quit score" taking precedent over balanced matches which is not the goal or how the system would function in practice. Hopefully, those concerns and others have been addressed. The goal is a better MM experience for players who play games through to the end.

My suggested system is basically a version of the "Trust Factor" system Valve uses for cheaters in CS:GO. Low trust factor players match with other low trust factor players. The system has generally been considered a success.
I think we both have the same goals in mind and I appreciate what your suggestion was attempting to do, but I question whether it's worth the effort; however, I am interested in hearing all the particular details of CS:GO's systematic method of incorporating the "Trust Factor" into its matchmaking system. Perhaps there is something there worth incorporating into Halo. Maybe not in Halo 5 but perhaps Infinite as long as it doesn't undermine the role of MMR as the true *primary* factor.
eLantern wrote:
NightClerk wrote:
How are the matches *primarily* based on MMR if you're allowing player quit rate "scores" to have influence over the team picker/balancer verses what the ideal MMR breakdown would dictate?
  • Matches would continue to be based on MMR. Other scores would NOT have influence OVER the MMR balance. When possible and within acceptable MMR parameters, teams would be assigned to match quitters on the same team, or put quitters into matches against each other. This would satisfy MMR balance requirements. Matches would remain balanced.
When you say "When possible" what specifically are you saying?

So from my understanding, and Josh can correct me if I'm missing critical info regarding the systematic steps, the Matchmaker first checks your selected search preference when you begin a search.

Based on the Focused, Balanced, or Expanded options the Matchmaker will alter what data centers it'll use to locate players for constructing a match.

After that Ping and MMR take priority.

On the MMR side, if you're searching as a solo player it'll look to find players around your individual MMR and if you're searching with a group your fireteam's average MMR (+ adjusted boost) is used as the target.

If MMR is turned-off for the Matchmaker it means that the search process begins with the allowable odds restriction and player skill-gap at their max limits. Player skill-gap limits may differ per the environment (ranked, social, warzone) and even perhaps the individual playlists too. The max limits still prevent the matchmaker from grabbing players that the team picker/balancer can't optimally place onto teams. In other words, players who would cause the match odds to become 61:39 or greater.

If MMR is active for the Matchmaker it means that the search process begins with the allowable odds restriction at around 50:50 before gradually expanding. Also, the skill-gap will begin much closer to your individual skill (as a solo searcher) or the skill-gap of your fireteam before gradually expanding. The initial restrictions may be different per the environment with the ranked environment often being more restrictive. Same for the skill-gap. Plus, the rate of expansion for the limits are often different too between the environments with social tending to expand quicker.

So how and when should a "trust metric" factor into the process?
Quote:
Again, how does MMR take *priority* over this new quit rate "score"? Please, explain more clearly how you perceive the system methodically working on a systematic level.
  • MMR remains the primary driver of matches and team balancing. When possible, quitters would be matched together in a lobby or on the same team. The system would scan the pool of searching players and establish matches with quitters playing quitters or quitters playing against "good standing" players. MMR will continue to take precedent and some matches may have to forgo the "quit score" factor to keep game within acceptable prediction percentages.
You say MMR should remain the *primary* driver. So essentially what you're advocating for is exactly the process that takes place now, but in the instance (which I imagine is extremely rare) where a player, or players, may be interchange-able between two optimally balanced teams let the "trust metric" pair them to the side that's more fitting to their personal trust metric? In other words, invest resources into incorporating a "trust" metric formula that would only very rarely have any influence?
Quote:
I understand that the system could potentially enact a quality threshold based on MMR just as it does now by preventing odds beyond 61:39. But, how else will MMR take priority over this new quit rate metric's influence with regards to the team picker/balancer system?
  • MMR is and will remain the primary driver of MM and balancing matches. Players who quit frequently would be put in a lobby or at least on the same team. However, matches would not be made unbalanced because of "quit score."
You see here you make it seem as if you'd rather see the trust factor metric take some precedence over MMR by the team picker/balancer as long as the max limits relating to match quality aren't infringed upon. And this is how I was perceiving your previous posts.
Quote:
I understand how some things I suggested could be misconstrued as a "quit score" taking precedent over balanced matches which is not the goal or how the system would function in practice. Hopefully, those concerns and others have been addressed. The goal is a better MM experience for players who play games through to the end.

My suggested system is basically a version of the "Trust Factor" system Valve uses for cheaters in CS:GO. Low trust factor players match with other low trust factor players. The system has generally been considered a success.
I think we both have the same goals in mind and I appreciate what your suggestion was attempting to do, but I question whether it's worth the effort; however, I would be more interested in hearing all the particular details of CS:GO's systematic method of incorporating the "Trust Factor" into its matchmaking system. Perhaps there is something there worth incorporating into Halo. Maybe not in Halo 5 but perhaps Infinite?
I’ve explained all of this pretty thoroughly. The “quit score” would be another metric used during MM but would never lead to imbalanced matches. MMR would be the primary driver of matches and no one would experience games outside of acceptable odds. “Quit score” would yield to MMR balancing. MMR takes precedence, and odds could very well be different because of a “quit score,” but never outside of MMR parameters (e.g. 61/39). “Quit score” may be disregarded or weighted less in some instances to balance search times with match quality. With larger populations, it would be more effective, as MM systems are.

I recommend looking up videos on CS:GO’s Trust Factor system and picturing that instead of cheating it factored in quitting. If a hyper-competitive game like CS:GO can implement a system like this, Halo certainly can.
The MMR system already accounts for quitting.

What is the main reason for quitting, aside from "real life?"

It's the feeling of being over-matched. If you are over-matched, that (typically) means the other team's skill/MMR is higher than yours.

Assuming a perfect world/perfect 50:50 win estimate, that means your skill/MMR is actually lower than the other team's.

Therefore, your MMR decreases.

Obviously, if you are way overmatched (40:60 win/loss estimate) the decrease is lessened.
ZaedynFel wrote:
NightClerk wrote:
ZaedynFelBeen having a lot of quitters/AFK in games recently. Played some Team Arena yesterday, these were my first five games.
  • Game 1: Starts 2v4 - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/f3b62936-fa22-42e0-959a-1f202b2d3f6c/players/nightclerk?gameHistoryMatchIndex=14&gameHistoryGameModeFilter=Arena
  • Game 2: One teammate AFK from the start, another quits. Then we all quit - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/397787f2-eb98-4d82-9c0d-ef761a34eaba/players/nightclerk?gameHistoryMatchIndex=13&gameHistoryGameModeFilter=Arena
  • Game 3: Success! Full teams for entirety of game. No AFKs. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/1b99c0fe-aec4-465e-b9e3-835515fc9aa2/players/nightclerk?gameHistoryMatchIndex=12&gameHistoryGameModeFilter=Arena
  • Game 4: Started with 1 or 2 AFK teammates who eventually quit. We all quit after they leave. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/5e96d950-5af5-40c4-a693-f1dd4dd99822/players/nightclerk?gameHistoryMatchIndex=11&gameHistoryGameModeFilter=Arena
  • Game 5: 2 teammates quit early. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/8c23ea97-1541-44b7-898b-4a49c092c637/players/nightclerk?gameHistoryMatchIndex=10&gameHistoryGameModeFilter=Arena
Fortunately I believe the ranking system worked and my CSR loss was pretty minimal. I think the penalties for quitting and being AFK aren’t strong enough to discourage this behavior. I will say that the matches I get are generally very competitive and TS2 finds balanced games. I’m sure you understand the frustration of going into game after game only to have that time wasted with a quitter.

Has there been any talk of implementing some type of “trust score” that tracks how often people quit? With a score like that, quitters could be matched with other quitters. Rewarding “good” players and punishing quitters in one motion.

This isn’t a criticism of TS2 or the MM system, which seems to be working very well. Just frustrated with these experiences I'm sure you can relate to. Keep up the good work.
Unfortunately quitting isn't a small group of players that be isolated, it's most players.

Players quit when they are clearly overmatched. Either at the team or individual level, and it's a large portion of the population.
Does that not mean the matchmaking did/does not work properly ? It happens alot in WZ Arena and btb playlists often. Players either solo or small fireteams matched with very low ranked players against larger teams in wz or just just very good communicating players, it leaves you frustrated when looking at stats to see your 20 plus positive and lower down your team playing out a -20 , gut im sure you know this. Keep trying and iwill also try not to quit out of unbalanced matches.
Z1000B wrote:
ZaedynFel wrote:
NightClerk wrote:
ZaedynFelBeen having a lot of quitters/AFK in games recently. Played some Team Arena yesterday, these were my first five games.
  • Game 1: Starts 2v4 - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/f3b62936-fa22-42e0-959a-1f202b2d3f6c/players/nightclerk?gameHistoryMatchIndex=14&gameHistoryGameModeFilter=Arena
  • Game 2: One teammate AFK from the start, another quits. Then we all quit - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/397787f2-eb98-4d82-9c0d-ef761a34eaba/players/nightclerk?gameHistoryMatchIndex=13&gameHistoryGameModeFilter=Arena
  • Game 3: Success! Full teams for entirety of game. No AFKs. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/1b99c0fe-aec4-465e-b9e3-835515fc9aa2/players/nightclerk?gameHistoryMatchIndex=12&gameHistoryGameModeFilter=Arena
  • Game 4: Started with 1 or 2 AFK teammates who eventually quit. We all quit after they leave. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/5e96d950-5af5-40c4-a693-f1dd4dd99822/players/nightclerk?gameHistoryMatchIndex=11&gameHistoryGameModeFilter=Arena
  • Game 5: 2 teammates quit early. - https://www.halowaypoint.com/en-us/games/halo-5-guardians/xbox-one/mode/arena/matches/8c23ea97-1541-44b7-898b-4a49c092c637/players/nightclerk?gameHistoryMatchIndex=10&gameHistoryGameModeFilter=Arena
Fortunately I believe the ranking system worked and my CSR loss was pretty minimal. I think the penalties for quitting and being AFK aren’t strong enough to discourage this behavior. I will say that the matches I get are generally very competitive and TS2 finds balanced games. I’m sure you understand the frustration of going into game after game only to have that time wasted with a quitter.

Has there been any talk of implementing some type of “trust score” that tracks how often people quit? With a score like that, quitters could be matched with other quitters. Rewarding “good” players and punishing quitters in one motion.

This isn’t a criticism of TS2 or the MM system, which seems to be working very well. Just frustrated with these experiences I'm sure you can relate to. Keep up the good work.
Unfortunately quitting isn't a small group of players that be isolated, it's most players.

Players quit when they are clearly overmatched. Either at the team or individual level, and it's a large portion of the population.
Does that not mean the matchmaking did/does not work properly ? It happens alot in WZ Arena and btb playlists often. Players either solo or small fireteams matched with very low ranked players against larger teams in wz or just just very good communicating players, it leaves you frustrated when looking at stats to see your 20 plus positive and lower down your team playing out a -20 , gut im sure you know this. Keep trying and iwill also try not to quit out of unbalanced matches.
Yes, this happens due to pop depending on the playlist. MM is tuned so if the pop is there at your skill level, you will have similar teammates against similar opponents, without large gaps in performance.

But in playlists like Warzone that require 24 players, the best players will almost never find all equal teammates because there are rarely 24 top players all searching at once solo.
NightClerk wrote:
The “quit score” would be another metric used during MM but would never lead to imbalanced matches.
Just to clarify, when you say "...would never lead to imbalanced matches" you refer to the max limit of 60:40, correct?

In my opinion, 60:40 odds does not equate to a "balanced" match. But, it is an acceptable max limit regarding allowable imbalanced matches.
Quote:
MMR would be the primary driver of matches...
How so?

Please feel free to use the breakdown that I provided of the matchmaking system to explain how.
Spoiler:
Show
Quote:
...and no one would experience games outside of acceptable odds.
As I noted here:
Spoiler:
Show
It seems as though I was correct as you make some conflicting comments here:
Quote:
“Quit score” would yield to MMR balancing. MMR takes precedence...
You say SBMM takes precedence ^here^, but then follow that up with a statement (below) that suggests it doesn't truly take precedence.
Quote:
...and odds could very well be different because of a “quit score”
Stating that MMR will be used to prevent match odds from going beyond the max allowable limit does not equate to it taking precedence within team balancing.
The quit discussion is interesting, but like I said, there aren't a lot of players who consistently quit.

Creating a quit score and matching on it would result in no material difference in experience to the players because the top quitters only account for a tiny fraction of the overall quitting.

The rest of the quitting is spread out among everyone, so everyone would have a similar score and just match same as today.

Except for maybe those who, for whatever reason, NEVER quit, which is a teeny portion. If we matched on a quit score, that tiny portion of stalwart players would have much longer wait times, which doesn't seem like a good reward for their behavior.
  1. 1
  2. ...
  3. 22
  4. 23
  5. 24
  6. 25
  7. 26
  8. ...
  9. 27