138 replies to this topic

[Nightbird]

https://mwomercs.com...p-match-making/

The more buckets, the fewer people in those buckets, and the worse the quality of matches made. What you want is the MM to know that short range mechs perform worse on long range maps and vice versa. As a result, teams that have non-ideal mechs will get more skilled players to compensate.

[MischiefSC]

Nightbird, on 19 June 2018 - 06:09 AM, said:

https://mwomercs.com...p-match-making/

The more buckets, the fewer people in those buckets, and the worse the quality of matches made. What you want is the MM to know that short range mechs perform worse on long range maps and vice versa. As a result, teams that have non-ideal mechs will get more skilled players to compensate.

It would have to pick the map before it makes the team to do that.

Possible, but you're talking about something better suited to MWO2 than MWO at this point.

[Nightbird]

MischiefSC, on 19 June 2018 - 11:01 AM, said:

It would have to pick the map before it makes the team to do that.

Possible, but you're talking about something better suited to MWO2 than MWO at this point.

The order as I see it being implemented for solo queue:

1. Population high, make 12vs12 teams
2. Wait until 72 players in queue
3. Launch 3 games, top 24 players based on pilot skill in one game, mid 24 in another etc..
4. Players select map and mode
5. During loading time, server adjusts pilot value based on map and mech(including loadout, skill tree, range, heat efficiency etc), arranges two even value teams
6. Do a tonnage check, swap players if necessary

If population is low, MM can target 10vs10 or 8vs8 reducing the wait to 60 or 48 players.

For QP GQ, assymetric team modifers, group size modifiers comes into play, and tonnage checks removed.

Edited by Nightbird, 19 June 2018 - 11:12 AM.

[Nightbird]

https://mwomercs.com...622-matchmaker/

The problem is the MM doesn't take into the account of each player's skill and mech choice, this leads to huge variances in skill between the teams in each match. The proposed MM will calculate skill and mech choice in a meticulous manner to minimize stomps. Note, this doesn't mean stomps won't happen, just that the odds are minimized.

[MischiefSC]

Nightbird, on 19 June 2018 - 11:11 AM, said:

The order as I see it being implemented for solo queue:

1. Population high, make 12vs12 teams
2. Wait until 72 players in queue
3. Launch 3 games, top 24 players based on pilot skill in one game, mid 24 in another etc..
4. Players select map and mode
5. During loading time, server adjusts pilot value based on map and mech(including loadout, skill tree, range, heat efficiency etc), arranges two even value teams
6. Do a tonnage check, swap players if necessary

If population is low, MM can target 10vs10 or 8vs8 reducing the wait to 60 or 48 players.

For QP GQ, assymetric team modifers, group size modifiers comes into play, and tonnage checks removed.

Drop step 5 (which would require a lot of work) and you have something that we could get now. This is especially nice because it gives you matches ending at approximately the same time.

Dynamic but matched team sizes is another thing with a lot of potential; 10v10, 8v8 when population is low. Reducing total team sizes to increase opportunity to make balanced teams.

I know that dropping step 5 removes a potentially strong balancing leg but A ) it's going to be hugely resource intensive by comparison and B ) involve some very significant changes and telemetry. It's also only going to be relevant when the population is thick enough that there's enough other choices available to fix it.

You can come back to 5 later - it's a solid idea and has merit but the others are (potentially) achievable with a small scale investment.

That's a great set of suggestions, let's build a business case, fishbone and various pretty charts and pitch it.

[Nightbird]

Can list step 5 as a part B or something. I disagree that it's just a potentially important point, since my effectiveness with a meta mech versus an unskilled event mech is pretty much the difference between the moon and the earth. Ok hyperbole, but Solaris one division merge isn't possible/fixable without it.

As far as pictures, without step 5 it becomes much easier. I would just need the final results and player lists of 1000 matches (including kill count on each side), and I could combine this with leaderboard data for each player to create a predictive model with existing leaderboard metrics.

That been said, unless this data is just laying around, I couldn't be bothered to collect it just to get PGI's attention. Too much effort

[MischiefSC]

Nightbird, on 20 June 2018 - 06:32 AM, said:

Can list step 5 as a part B or something. I disagree that it's just a potentially important point, since my effectiveness with a meta mech versus an unskilled event mech is pretty much the difference between the moon and the earth. Ok hyperbole, but Solaris one division merge isn't possible/fixable without it.

As far as pictures, without step 5 it becomes much easier. I would just need the final results and player lists of 1000 matches (including kill count on each side), and I could combine this with leaderboard data for each player to create a predictive model with existing leaderboard metrics.

That been said, unless this data is just laying around, I couldn't be bothered to collect it just to get PGI's attention. Too much effort

The issue with 5 is you need a lot of calls per player to make it happen. Sure, you could probably do it all with SQL databases but you're adding a whole other layer. The QA alone would be a huge hassle. It's a complete and complex independent system that would need developed, tested and integrated.

I'm saying only potentially useful because it doesn't change who's in the queue. Yeah, it might result in shuffling a couple people from one team to the other and I don't disagree that it would be a positive impact but all the rest of it is a huge impact without creating a new system, just a new process.

It's a sound set of ideas. Maybe PGI will come back with "sure, step 5 is easy with what we have" and boom. However the rest of it would still be a big improvement.

[Hal Greaves]

late night solo queue is frustrating, to say the least

For me, by the time I get into position 3-4 people are already dead and I barely have time to even do much before the whole team is rolled. Was a bit frustrating to see how bad the rolling gets at night.

Edited by Hal Greaves, 20 June 2018 - 08:55 AM.

[Nightbird]

I'm not concerned at all about the database actually. For a match, PGI already has to load each pilot's mech, from armor values, to equipment, to skill tree, to paintjob and decal positioning. The harder part is implementing the algorithm that turns each mech into a number that corresponds to its effectiveness. The easy thing would be me crunching all the numbers and getting the coeffecients for an algebraic equation that is easy to program, the much harder part is the modelling to get those coefficients. The latter I don't expect PGI to have the know how to do.

For example, people use the Jarl ranking average today to compare teams, I think it's better than PGI's tiers. But, between a team of players with low W/L ratio on the first page, or high W/L ratio on the second, it's clear the latter would win. This is because the ranks are not based on a player's expected contribution to winning from even an intuitive point of view.

Data mining is the future, yes, it sucks for ads, but for medicine or a good match maker why not.

[Nightbird]

https://mwomercs.com...ocus-on-fp-now/

Yes and no, new features are pointless without a new MM because balanced fights are more critical to this mode than QP as respawns magnify team imbalances.

[Nightbird]

https://mwomercs.com...s/page__st__460

After 24 pages, how about instead of changing QP SQ into seal clubbing grounds like GQ and FP, we just create balanced teams hmm?

[Haipyng]

Nightbird, on 22 June 2018 - 08:13 AM, said:

https://mwomercs.com...s/page__st__460

After 24 pages, how about instead of changing QP SQ into seal clubbing grounds like GQ and FP, we just create balanced teams hmm?

I see where they are coming from. While fixing GP MM is undoubtedly the right action, after 5+ years, multiple game changes, patches, and assorted updates, the best that has been done for balancing QP MM is 3/3/3/3, then tonnage restrictions and a tweak to tonnage restrictions. All that, despite years of complaints about it and there is still no sign of any interest from PGI in fixing GP MM. Folks are looking for solutions that PGI is able or willing to do.

A fixed MM system that can create balanced matches would be a big boost to FW, QP/GP alike. You would think it would be a no brainer.

[Agent of Change]

I created a Thread on this a few days back in feature suggestions:

https://mwomercs.com...07#entry6118607

It lays out a quick and dirty mechanic by which average game score might be used to better establish PSR rating and thereby create better matchmaking in a way that doesn't require a whole lot of work on PGI's part which would hopefully make it more likely.

[Nightbird]

Haipyng, on 22 June 2018 - 09:38 AM, said:

I see where they are coming from. While fixing GP MM is undoubtedly the right action, after 5+ years, multiple game changes, patches, and assorted updates, the best that has been done for balancing QP MM is 3/3/3/3, then tonnage restrictions and a tweak to tonnage restrictions. All that, despite years of complaints about it and there is still no sign of any interest from PGI in fixing GP MM. Folks are looking for solutions that PGI is able or willing to do.

A fixed MM system that can create balanced matches would be a big boost to FW, QP/GP alike. You would think it would be a no brainer.

I understand where they are coming from as well, but their suggestion will make the situation worse. As someone who is not even on the first page of Jarl's list, my teams already win 75% of the time with many solo carries. If I can bring a buddy with the same or greater skill, this will jump to 90%. I don't want SQ to be a stomp fest by power duos the way that GQ is a stomp fest. I get what the idea posters want, but wanting it is not the same as having the right method of achieving it. The path to hell is paved with good intentions and all that.

Agent of Change, on 22 June 2018 - 09:58 AM, said:

I created a Thread on this a few days back in feature suggestions:

https://mwomercs.com...07#entry6118607

It lays out a quick and dirty mechanic by which average game score might be used to better establish PSR rating and thereby create better matchmaking in a way that doesn't require a whole lot of work on PGI's part which would hopefully make it more likely.

Average match score is not much more useful than Tier. You can see on Jarl's list that people with high match score can have HALF the win rate of people with low match score. It may be a little better, or it may be as bad. PGI excels at making half hearted attempts and giving up, so that's why I only want to advocate a surefire way of resolving this mess.

[Mystere]

Nightbird, on 20 June 2018 - 09:20 AM, said:

I'm not concerned at all about the database actually. For a match, PGI already has to load each pilot's mech, from armor values, to equipment, to skill tree, to paintjob and decal positioning. The harder part is implementing the algorithm that turns each mech into a number that corresponds to its effectiveness. The easy thing would be me crunching all the numbers and getting the coeffecients for an algebraic equation that is easy to program, the much harder part is the modelling to get those coefficients. The latter I don't expect PGI to have the know how to do.

For example, people use the Jarl ranking average today to compare teams, I think it's better than PGI's tiers. But, between a team of players with low W/L ratio on the first page, or high W/L ratio on the second, it's clear the latter would win. This is because the ranks are not based on a player's expected contribution to winning from even an intuitive point of view.

Data mining is the future, yes, it sucks for ads, but for medicine or a good match maker why not.

Ah, data mining!

I think I should go back to being a filthy (rich) consultant again.

[Mystere]

Nightbird, on 21 June 2018 - 06:18 AM, said:

https://mwomercs.com...ocus-on-fp-now/

Yes and no, new features are pointless without a new MM because balanced fights are more critical to this mode than QP as respawns magnify team imbalances.

How about instead of "balanced" teams in CW, we get proper and engaging "battle scenarios" and not those eSports-be-damned "matches"?

Edited by Mystere, 22 June 2018 - 12:01 PM.

[Nightbird]

Mystere, on 22 June 2018 - 11:41 AM, said:

How about instead of "balanced" teams in CW, we get proper and engaging "battle scenarios" and not those eSports-be-damned "matches"?

Scenarios against bots? Against people the teams will continue to be a problem.

[dimachaerus]

Another thing we could do, is get rid of "matchmaking" entirely. Been playing a lot of MW:LL lately and have re-discovered the joy of a game that lets You pick what team you want to be on, what map you want to play, and the people you want to play with. Give us respawn, and objective based win conditions, not "kill everything in all modes to win" conditions, and suddenly you have more varied gameplay.

The playerbase for MWO is so small currently that a matchmaker just makes no sense honestly.

[SFC174]

dimachaerus, on 22 June 2018 - 05:10 PM, said:

Another thing we could do, is get rid of "matchmaking" entirely. Been playing a lot of MW:LL lately and have re-discovered the joy of a game that lets You pick what team you want to be on, what map you want to play, and the people you want to play with. Give us respawn, and objective based win conditions, not "kill everything in all modes to win" conditions, and suddenly you have more varied gameplay.

The playerbase for MWO is so small currently that a matchmaker just makes no sense honestly.

Sure it does. But you have to have balance. It's ok to have 3 potatoes, 6 regular joes, 2 95%ers and a God Tier player on each team as long as _each team_ has a similar range of players. Yeah, potatoes will potato, but they'll be exposed to good players and most will learn at least a little bit from being in a game with those guys.

Doesn't mean I think Tier5 players should be thrown to the wolves, but we have plenty of potatoes in Tier1 as it stands. Re-ranking them as Tier 3 of 5 or Tier 6 of 10 (take your pick) doesn't mean they wouldn't see many of the same (better) players they see now. But at least they all wouldn't end up on the same team. And it also might give them a chance to play in matches with players largely of their caliber, which is generally more fun than getting ROFLstomped by 3 top flight players on the opposing team while their best player is a 90%er.

[Nightbird]

dimachaerus, on 22 June 2018 - 05:10 PM, said:

Another thing we could do, is get rid of "matchmaking" entirely. Been playing a lot of MW:LL lately and have re-discovered the joy of a game that lets You pick what team you want to be on, what map you want to play, and the people you want to play with. Give us respawn, and objective based win conditions, not "kill everything in all modes to win" conditions, and suddenly you have more varied gameplay.

That's FP today. Feel the team balance or PvE when ignoring mechs is better?