Stacking

[Violator]

This game seems to do its best to stack teams. The algorithm seems to be to take the top half highest ranked players and place them on the same team, bar a couple. I suspect its completely random though but I would say 80% of the time it ends up in a stack. This has been the way since day one but it seems far worse with the recent couple of patches. A really simple idea, even just based on player level (this could be applied to k/d, xp / round etc. as well). I’m still surprised nothing like this has been implemented:

Sort all players by descending level / ranking / k/d etc. Alternate placing players on each team until none are left.

The team that sticks together invariably wins, its very rare that both teams do this though as one team usually ends up with all the low-ranked players who haven’t yet learned this style of play.

This should never happen:

[Anti]

There is actually a ‘Public Skill Rating’ used to shuffle teams, that is derived from player performance over a number of matches. The problem is more likely to do with when shuffles occur and that we don’t yet Quick Match by that skill.

We’re keeping an eye on both and planning to improve/address them over time.

[Szakalot]

This is going to be very hard to nail down imo.

DB favors teamwork heavily; and two very-good players working together can destroy half of the enemy team. The other half being heavily outgunned at this point, so much that the weakest players on the strongest team will have much better stats than good players on the losing team. I’m not sure how that factors in the shuffling of the previous matches, but if you get lucky with good teammates, your rank might be boosted considerably.

It is the curse of DB’s format that individual player’s performance is heavily dependent on their teammates, which is hard to factor in.

Perhaps a system looking at that player’s performance over a longer period of time would be more accurate?

Other potential problems:

• multitude of people ragequit before the map ends
• different maps/stages are attacker/defender biased

My hope is that some type of ranking system can be extended to casual; to isolate newbies from the alpha-slayers. Or a casual matchmaking like in CSGO.

[tangoliber]

[QUOTE=Szakalot;524641]This is going to be very hard to nail down imo.

DB favors teamwork heavily; and two very-good players working together can destroy half of the enemy team. The other half being heavily outgunned at this point, so much that the weakest players on the strongest team will have much better stats than good players on the losing team. I’m not sure how that factors in the shuffling of the previous matches, but if you get lucky with good teammates, your rank might be boosted considerably.

It is the curse of DB’s format that individual player’s performance is heavily dependent on their teammates, which is hard to factor in.

Perhaps a system looking at that player’s performance over a longer period of time would be more accurate?

Other potential problems:

• multitude of people ragequit before the map ends
• different maps/stages are attacker/defender biased

My hope is that some type of ranking system can be extended to casual; to isolate newbies from the alpha-slayers. Or a casual matchmaking like in CSGO.[/QUOTE]

I think the “ranking score” should be different based on the class being used. If they are playing Proxy, then I think number of deaths should be almost thrown out the window. Reparing/planting/defusing, etc, + kills should determine the ranking.

If playing medic, I think it should be number of revives + healing, with a penalty for # of deaths. (Number of kills don’t matter as much, but you don’t want your medics dying too much.)

Snipers should really be rated on # of kills, and Fragger/Skyhammer/Arty should really be rated on actual KDR and KPM …with bonus points for kills from using grenades/artillery.

[tokamak]

[QUOTE=Szakalot;524641]This is going to be very hard to nail down imo.

DB favors teamwork heavily; and two very-good players working together can destroy half of the enemy team. The other half being heavily outgunned at this point, so much that the weakest players on the strongest team will have much better stats than good players on the losing team. I’m not sure how that factors in the shuffling of the previous matches, but if you get lucky with good teammates, your rank might be boosted considerably.

It is the curse of DB’s format that individual player’s performance is heavily dependent on their teammates, which is hard to factor in.

Perhaps a system looking at that player’s performance over a longer period of time would be more accurate?

Other potential problems:

• multitude of people ragequit before the map ends
• different maps/stages are attacker/defender biased

My hope is that some type of ranking system can be extended to casual; to isolate newbies from the alpha-slayers. Or a casual matchmaking like in CSGO.[/QUOTE]

The current way the games are played it too frivolous and transient, at least compared to other competitive F2P games.

In games like WoT and Mobas you play one match at the time (rather than hopping in continuous servers). The ranked mobas even have a team selection round before the game starts.

All of this eliminates the externalities like dropping out of games and biases. At the same time it allows to control for players that are playing with friends. With such a system it becomes much easier to track a single player’s performance. There’s the ELO (usually hidden) and the visible ranking. Combine the two and the ladders actually start to mean something.

[BioSnark]

[QUOTE=Anti;524637]There is actually a ‘Public Skill Rating’ used to shuffle teams, that is derived from player performance over a number of matches. The problem is more likely to do with when shuffles occur and that we don’t yet Quick Match by that skill.

We’re keeping an eye on both and planning to improve/address them over time.[/QUOTE]

Since teams are already reshuffled at map start, which makes player balancing over multiple maps hard, please reshuffle based on skill rating at the beginning of each match.

[spookify]

Just because you have a high level doesn’t mean you are the best… or even good for that matter…

I have seen a few people with level 15, 16 , 17 and even a 19 already.

One or two of these players might be very very good but not all these high ranks are the best…

Play Time does not equal skill. Other things go into it.

Plus joining a pub is so random… I use the open server browser rather then the MM Pub…

[sunshinefats]

Something else that factors into this whole thing is players who intentionally try to stack teams. I’ve encountered this a few times recently, where the player will quit the server and rejoin on the other team intentionally so as to, presumably, change the overall balance of the teams. This sort of thing can make a big difference, especially when there are already several good players on the other team.

[Szakalot]

Something else that factors into this whole thing is players who intentionally try to stack teams. I’ve encountered this a few times recently, where the player will quit the server and rejoin on the other team intentionally so as to, presumably, change the overall balance of the teams. This sort of thing can make a big difference, especially when there are already several good players on the other team.

in all likelihood its friends wanting to play together. Not much can be done about that, and i dont think the game should try to address it in any way

[Szakalot]

Small revive but something that might have been missed regarding balancing.

Currently auto-balancing based on skill or whatnot is done before loading the new lobby. However, substantial amount of players leaves the lobby at this stage, new people come in - map is chosen and even more people leave&join, screwing up whatever plan you had to balance the teams properly.

Voting on the map before the lobby might solve this problem considerably. So would loading the map before the lobby, as the people that leave would be more likely to leave before the map is loaded (so as to avoid loading the map, duh)

[Anti]

[QUOTE=Szakalot;525879]

Voting on the map before the lobby might solve this problem considerably. So would loading the map before the lobby, as the people that leave would be more likely to leave before the map is loaded (so as to avoid loading the map, duh)[/QUOTE]

Funny, we discussed both those options again yesterday. I think we’d like to do this in the long run, although it’s a considerable code change so it could be quite some time away, but it’s definitely an interesting option.

[spookify]

Funny, we discussed both those options again yesterday. I think we’d like to do this in the long run, although it’s a considerable code change so it could be quite some time away, but it’s definitely an interesting option.

OR

Right before the map loads the teams are shuffled to “Avoid Stacking”…

As soon as a round ends people are coming and going. This makes for uneven teams during the map voting, which leads to team stacking. Clans have a chance to get on the same team during this time.

I would suggest that when the map is locked and teams are locked a Shuffle then occurs right as the map loads. Maybe even before people can see teams! Boom in map with your team and the count down starts and pub match begins.

[Humbugsen]

[QUOTE=spookify;525898]OR

Right before the map loads the teams are shuffled to “Avoid Stacking”…

As soon as a round ends people are coming and going. This makes for uneven teams during the map voting, which leads to team stacking. Clans have a chance to get on the same team during this time.

I would suggest that when the map is locked and teams are locked a Shuffle then occurs right as the map loads. Maybe even before people can see teams! Boom in map with your team and the count down starts and pub match begins.[/QUOTE]

you can’t plan your team composition then, but i don’t think that anyone does that in a pub anyway
so yes i think thats better than a “considerable code change”

but we really need something to do in the lobby/endgame like opening cases or watch more detailed (weapon) stats, seems like lots of people leave because that takes to long

[Zenity]

I think this really is the biggest issue for player enjoyment while the game is still in closed beta. I don’t think I’ve ever played a game where teams are this stacked on a regular basis. I’ve seen so many skilled players switch teams at the earliest chance to stack teams and get an easy win, it’s just sad. Unfortunately this behaviour is strongly encouraged by the reward system.

I understand that you want to be lenient to allow players to work out the teams and friends to play together, but this only works well if the community is being nice about it. Unfortunately the community is being an asshole about it, so I strongly believe that you need to do the opposite and enforce some kind of team balance instead.

Shuffle by skill right before the game starts sounds extreme, but I do think it would be the best solution at least temporarily until MMR gets going and we have more players.

[DarkangelUK]

[QUOTE=spookify;524740]Just because you have a high level doesn’t mean you are the best… or even good for that matter…

I have seen a few people with level 15, 16 , 17 and even a 19 already.

One or two of these players might be very very good but not all these high ranks are the best…

Play Time does not equal skill. Other things go into it.

Plus joining a pub is so random… I use the open server browser rather then the MM Pub…[/QUOTE]

Play time still equals a greater familiarity with the game, knowledge of the short cuts and quick tricks to get things done. It may not = shooting skill but longer play time still gives a greater advantage over new players, and a full team of long term players against a team of new guys will still have the upper hand.

[spookify]

Play time still equals a greater familiarity with the game, knowledge of the short cuts and quick tricks to get things done. It may not = shooting skill but longer play time still gives a greater advantage over new players, and a full team of long term players against a team of new guys will still have the upper hand.

I really dont think that is true. Especially if you are talking level 20 vs level 10…

Level 20 vs level 3 then maybe but once that person has reached level 5 or 6 then vs a level 20 the better player should win.

I have something like 70 hours in CS:GO I just put it on my PC and I am owning people with 5 year service badges…

The amount of play time or level does not equal skill. I am not 100% sure but I think SD might use a skill based calculation to balance rather then level which is great.

Just Shuffle the teams by this skill rank right before the map loads and problem solved.

[Szakalot]

you’d still need lobby after the map like i said. otherwise people can’t pick mercs based on their teammates, which SD strongly favors

[spookify]

you’d still need lobby after the map like i said. otherwise people can’t pick mercs based on their teammates, which SD strongly favors

Who said anything about the After and Before Map Lobby going away?

Of Course you need that…

[Violator]

Some more analysis:

42 vs 29

44 vs 36 (though they had potty ;))

29(4) vs 24(5)

70 vs 54

43 (6) vs 63 (7) - this was how the match started.

I don’t understand how this consistently happens - the top X ranked players are all on the same team. Of course there is the defender bias on Bridge but there are plenty of reverse examples. It is quite rare for there to be an even balance of player levels in a given match.

[Violator]

Some more. Note that these are typical, not unusual

58 vs 76

49 vs 59

42(5) vs 28(4) - again match started like this

33 vs 64