The Current State of Renegade X Balance: With Statistics

[Tytonium]

The Current State of Renegade X’s Balance

by: Tytonium

            HEY YOU! Yeah you. You like Renegade X? Well of course you do you are browsing the forums you degenerate. Perhaps you are like me and enjoy Renegade X more so than any other multiplayer game. With its unique style of game play pretty much seen nowhere else (or at least not executed as flawlessly), Renegade X is a shining gem among games. However, beneath Renegade X’s amazing exterior lies a problem. Perhaps you have felt it too during your time playing. And no I’m not just talking about the little baby bitch mines that people can place in the Hand of Nod that go invisible and always end up killing me. No, I am talking to you today about balancing issues in Renegade X. I put a ton of work and 3 months of time and effort into collecting this data and drawing conclusions from it. So I would absolutely appreciate a tiny bit of your time to check out this major issue I am going to highlight with my statistics and MATH! Read over the section of my method, and then you can skip to the picture of the cat IF you are just a normal player and want to see the conclusions without going into too much detail of the actual statistics. However, if you are an avid member of the community or a developer, I implore you to read through all of this in order to understand how I got to my conclusions. So sit on back, grab a cup of coffee, pour some godforsaken rum in that cup, and let’s get into this monstrosity.

 

Method

            So in case you didn’t know, I have been collecting statistics for this game for quite some time now. I started nearly a year ago, but I ended up getting too drunk for my own good, and kind of ruined the statistics with forgetting to add key data points and showing truthful statistics. So… I restarted about 3 months ago.

            The following statistics have been gathered over the course of exactly 3 months, from September 14, all the way to December 14.

            The statistics I gathered dealt with a predicted winning team’s chance of winning past the 5 minute mark in the game.

            So essentially, for my statistics, at the 5-minute mark of the game, I would look at the scoreboard by pressing “k” and seeing which team, if any, was predominantly near the top of the scoreboard. The team that was more on top, became the “predicted winning team” and the other team was the “predicted loser”. I did it in this manner because that would allow for an easy implementation of binomial distribution, and binomial distribution is the sexiest. I promise I did this in the fairest way possible. If say 7 Nod people were in the top 10 at the 5-minute mark, then Nod was the predicted winner. If 4 Nod people were in the top 10, but an overwhelming amount of Nod people were ahead of GDI in the scoreboard, (AKA GDI full of newbs and like 75% of the team makes up the bottom half of the scoreboard) then Nod is still the predicted winner. If however, the scoreboard was quite balanced and neither team seems to be winning at 5 minutes, then the game is “balanced”. I will say this if you are thinking this is a bit too subjective of a stat gathering method, it was really, REALLY, easy to tell which team was predominantly in the scoreboard on nearly all matches. I think this method worked out just fine in the end.

 

Requirements for Data to Count

I put together some conditions that must be met in order for a match to be logged. This is so I can’t just log 10 matches back to back when the player count is at 20 and one team is stacking because everyone wants to play with their friends. So here are the conditions:

1.      Player count must be above 40 throughout the entirety of the match. No sense doing these statistics in anything less than what is considered an actual game with a proper amount of players.

2.      Player count must remain relatively the same. If say 1-8 people end up joining I think that is fair, or if 1-8 people end up leaving, same still. But if you start a match with 40 people and end it with 64, will now that tampers with the data too much.

3.      No Oasis, Snow, or Snow X allowed. Sorry to those map makers, but they just aren’t really suited for 40-64 players at all. It’s always a clusterfuck and some shitty rocket rush usually ends each match.

4.      Match doesn’t count if some unusual factor I couldn’t have planned for happens. Like someone glitching a nuke on a building’s roof when they can’t get there. Server crashing. There being like 10 afk players for the first 10 mins, etc…

 

General Observations

So before we get into the actual statistics I just want to write about what I saw during these matches to explain away some of the results.

·         “Balanced” Games were always the funnest to play, but they were rare. When both teams are going at it with all their ferocity and the best commander is picked on both teams, that is when RenX is at its best.

·         Whenever the “predicted losing team” ends up stealing a win, it can be one of 4 things that caused it in my opinion. A super duper lucky tech/hotty sneak, a super good commander appeared on the losing team that ends up essentially winning the game, a god-tier player shows up and takes the inf path single handedly, Or the winning team is just very uncoordinated or has a commander that doesn’t make plans whatsoever, or just has really bad plans. This is why I added in another stat, that being if the match went on for over 60 mins when the losing team steals the win and no progress was made in the first 60 mins by either team, as this indicates that the winning team probably fucked up since they couldn’t mount a successful offensive.

·         I split up the stats into 2 sections. FIELD, UNDER, AND ISLANDS are in their own separate category. I did this because I feel as though these maps in particular usually result in stagnant matches. Essentially the one thing these matches have in common is the inf path is a complete cluster fuck with vehicles being able to single-handedly stop infantry rushes, and this causes the games to become long and drawn out with neither team being able to truly rush in an effective manner even despite being way better. (AKA arty/MRLS spamming tunnels makes me wanna barf). On all other maps it is a lot harder to do that amount of spam that stops infantry rushes.

 

The Data

FOR FIELD, UNDER, AND ISLANDS ONLY

Total matches: n=30

Winning team won: 25

Winning team lost: 2        did this match go on for 60 minutes with no progress? 2/2

Balanced game: 3

 

FOR ALL OTHER MAPS EXCLUDING FIELD, UNDER, AND ISLANDS

Total matches: n=77

Winning team won: 68

Winning team lost: 5           go on for 60 mins without any progress? 2/5

Balanced game: 4

 

This may not look like much, but we can do quite a bit of things with this data set.

 

Showing My Work

You can skip past this if you want to the cat picture. I’m just showing how I got my conclusions.

Essentially for each data set I did 2 binomial distributions. One for which matches were “interesting” which was balanced and where the losing team won, and another for just “balanced” matches.

So for example for interesting matches in Field, Under, and Islands.

N=30

P(x) aka probability of success. Success = the match being interesting

P(x) = 5/30 = .1666666

Not doing this shit by hand, used a calculator for bidistr.

BinomialCDF(30,.166666,X=0:10)

This will give the percentage chance that this number ____ of matches are interesting in a set of 30 matches.

Then I add up the matches from 0-X where X equals a match number where approximately 99% of all matches will fall in between. So if you do this correctly you can say that you can be 99% confident that 0-X matches will be interesting out of 30 total. You can also turn 0-X matches into a percentage by taking X/n so you get a percentage of how many matches are interesting or balanced in those certain maps.

Doing this we get that

UIF (Under Island Field) – interesting = .33 (AKA 33 percent of all games means X equaled 10. So that means that 0-10 games make up a 99% confidence level that this is how many games would be interesting out of 30. So then take 10/30 = .3333 games are interesting if my data is not a horrible fluke.

UIF balanced = .2333

All others Interesting = .1948

All other balanced = .1299

 

Look at this good boy. So polite.

CONCLUSIONS WE CAN DRAW

After doing all the fun stats we can draw some conclusions on game balance.

FOR UNDER ISLANDS AND FIELD ONLY

We can be 99.3255% sure that essentially 33% or a third of all games played on Islands, Field, and Under, HAVE THE POTENTIAL to be interesting and not just a one-sided stomp. Notice it HAS POTENTIAL. So this certainly does not mean that a third of all games are interesting in some way. It means at a maximum it is one third if the stars aligned perfectly. I’ll show a graph in the next conclusion as it is more important since it deals with balancing.

 

We can be 99.2216% confident that AT MOST 23.33% or essentially a quarter of all games have THE POTENTIAL to be balanced on Under, Islands, and Field. Here is a graph to help you better picture what this means. We essentially have turned binomial distribution into a normal distribution bar graph.

 

 

FOR ALL MAPS EXCLUDING UNDER ISLANDS AND FIELD

We can be 98.4611% sure that only a maximum of 19.48% of all matches have THE POTENTIAL to be interesting on all other maps besides UIF.

 

Here’s a real kicker.

We can be 99.7928% sure that only a maximum of 12.99% of all matches HAVE THE POTENTIAL to be BALANCED on all other maps besides UIF. Here is a graph to visualize this.

 

 

Some other fun things we can do is to figure out if it is even worth trying to fight when you are on the losing team.

Exclude balanced matches because it’s past the 5 min mark and you are losing. So for example, 2 loses / 27 matches = .074074

So IF MY DATA IS NOT A HORRIBLE FLUKE, well… On Under Islands and Field only, if you are losing at the 5 minute mark… you have a 7.41% chance of winning!

Isn’t that GREAT!

Similarly on any other map, if you are losing at the 5 minute mark, you have a 6.85% chance of winning that godforsaken match! Good luck buddy.

 

            I have to admit I was expecting a bigger difference between the two data sets. Because they are so similar I decided to go ahead and do the distributions for all the matches combined for INTERESTING matches, not balanced, as balance surely changes from those 3 maps compared to all the rest.

N=107 P(x) = .1308  Binomialcdf (107,.1308,0,23)

 

We can be 99.4645% sure that between 7 and 23 matches out of 107 total matches will be something other than a stomp. Or in other words:

We can be 99.4645% sure that at a maximum, 21.50% of all matches have the potential to be something other than a stomp fest. Graph to help you visualize.

 

 

Lastly

If you are on the losing side of a match by the 5 minute mark, you have a 7% chance of winning that match on any given map.

 

What to think?

            If it is not evident to you, there is a major problem with team balancing. Having this system of anarchy is not going to cut it as the overwhelming majority of matches can be seen as being over in the first 5 minutes as indicated by this data. So what is the point of playing if you are losing? You have a pretty good idea of how the story will end.

Why do I think this is such a problem?

A better question is just how important is balancing to the game? Well, Renegade X is a special case when it comes to multiplayer games. My issue with balancing and why I am making such a big deal about it is because BALANCING IS DIRECTLY PROPORTIONAL TO PLAYER CHOICE AND THEREFORE HAS AN IMPACT ON MATCH ENJOYMENT. What other game is out there where doing bad as a team results in your options being limited? Sure, on most games doing bad as a team results in being limited to map control or being unable to get momentum based options (like kill streaks in call of duty). In Renegade X doing bad as a team results in severe map control loss (being locked in), and potentially losing buildings which completely cut off player choice in a severe way for the entirety of the rest of the match (like no infantry or vehicles. That is major!).

My point is, the balancing of teams MUST be good, or nearly every match will be a complete and utter stomp where only one team has fun, and the other people are just left to suffer.

I’m sure we have all thought about how to balance this game at some point or another, and I have noticed a team balancing mutator on FPI popped up recently. Didn’t really seem to have much of an impact though but I am curious as to what method it balances by.

You need to answer this question in order to get balancing right: What is the biggest indicator of player skill in this game?

Figure that out and you can balance the game.

In my opinion the biggest indicator is the amount of time a player has been in game. I would much rather have experienced players on my team who knows what they are doing than really good fps players. I think balancing for time played would be something to try out if you haven’t already done it. You could go about requiring Steam for the game to be played but I don’t know what kind of problems that would bring with it. There just needs to be a way to track ALL player stats in my opinion so you can go about balancing through this method or another similar one.

I am not a game developer

I can merely do these relatively simple statistics to show you a problem and hopefully shed some more light on this issue which we all know has been in the game for quite some time.

I really hope these stats can help in the development of this game, it’s the only multiplayer game I play anymore and all I want is for more matches to be fun.

Whatever route you all end up taking to tackle this crippling issue, I wish you luck in your endeavors. If you have trouble figuring out a solution… just remember,

 

 

Also fuck invisible HON mines

Edited December 21, 2019 by Tytonium

[Akbaro]

the absolute STATE

[TomUjain]

Team balance is a tricky thing to get right, and even a manual shuffle fails to perfect it at times. The fault, I believe is how the game is designed. In a nut shell: a bad team is more likely to lose a key building, further and permanently handicaping the already bad team. The shuffler has a high chance of re-applying the same awful team thus mimicing the potential problem for upwards of 3 games in a row. 64 players is also not helping adding to the 'bottle neck' effect we feel on maps such as field / under where one team holds the field / tunnels / everything -- because it can. Add to that the buff AOE and VP ranking system and you can have (potentially) 32 heroic rocket men melting your entire base before you can blink. Having units wail on a building for 2 hours on AFK mode isn't something we should be rewarding. To counter this I believe VP should be awarded to buildings, increaseing armour / health the more damage they take over time to encourage more tactical takedowns.

So how do we perfect the team balance system? We need it to shuffle people based on their last game and score up both teams based on that data. This, naturally poses a threat to the whole 'play with friends' commity so we need a system in place that rewards people for moving to the losing team by either making them spawn with a special unit / credits, or allows them to move over with friends.

Make no mistake here, 'good players' are one thing, but good commanders are another. Almost all matches are defined by the strength and tactical prowess of the commander, not the skillful players - this naturally becomes a problem when one side focuses on K/D while the other focuses on winning.

 

 

 

 

Edited December 19, 2019 by TomUjain

[Killertomate]

IMHO this whole stacking/balancing issue discussion is just a waste of time and energy.

However, as mentioned by op, relatively early in the game it is pretty clear which team is going to loose. I don't see a problem in that. The problem is that the match might still take 60 minutes and people get bored and leave the server.

So basically the game needs to be over faster! Make the crates more powerful, make the harv faster and stuff like that ...

Accept that having lots of cool tactical fair matches on pub servers is an impossible goal or a "Luftschloss" as we Germans would call it 

[Rups]

Interesting read, although the results do not surprise me.

1 hour ago, Tytonium said:

In Renegade X doing bad as a team results in severe map control loss (being locked in), and potentially losing buildings which completely cut off player choice in a severe way for the entirety of the rest of the match (like no infantry or vehicles. That is major!).

Just to add to that a little:

When you're losing, you lose map control, which includes tib fields (if present in center of map), tech buildings, crates, etc. Attackers gain bonus points for killing people in their base. Winning means more VP which results in more veterancy. If the teams get stacked intentionally the impact goes up even further.

Losing means you have less money, your soldier deals less damage and has less armor on average (due to veterancy AND money advantage), you're locked in a base where the defenders are at a disadvantage when they try to break out...... Once you lose a building everything just spirals downwards even further. Losing bar/hon usually effectively ends the game.

All in all, the game has some massive snowballing mechanics. Combined with an utter lack of teambalancing and a rich history of rampant and pretty hardcore teamstacking, the results Tytonium showed really aren't shocking. In fact, I expected the amount of *balanced* games to be lower. The amount of snowballing needs to be toned down significantly. Veterancy just has a habit of screwing over the losing team and new players. The impact of losing a building certainly could be toned down a notch. Bases in general should be a little easier to defend (and by that I mean, make the attacker have less of a stranglehold on the vehicle exits). I understand a lot of these mechanics were implemented to avoid stalemates, but man, I'd love some stalemates over one team stomping the other repeatedly. 

And to be honest, I think repairs are an underlying issue in this game in general. Repairs are so effective that a team can siege the enemy base for 60 min straight. (And repairing isn't exactly fun gameplay in general from the perspective of the engineer/hotwire, just think of how many games have AFK repairs in them).

tl;dr: teambalance is badly needed, but tweaking/changing game mechanics would certainly go a long way in decreasing the detrimental effects losing has on your odds of winning. A combination of the two has the best chance of making the average game less one-sided. 

 

3 minutes ago, Killertomate said:

IMHO this whole stacking/balancing issue discussion is just a waste of time and energy.

A lot of people care very strongly about it. The only thing I'll agree with on you is that games that are a clear loss that still last 60 min are a waste of time. However, I do feel that that can be changed with better balance and/or tweaking some game mechanics. A small percentage of shitty games that are sort of a time waste are unavoidable and acceptable, however it feels like currently only a small percentage of games is interesting, balanced and truly worth playing.

[euan-missile]

I needed more godforsaken rum to get through that post, I agree with your points mostly but I think map design is something overlooked. Getting pinned down on field is completely different than having it done to you on field X, but aside from that I'd agree. 

Edited December 19, 2019 by euan-missile

[roweboat]

8 hours ago, Tytonium said:

I’m not just talking about the little baby bitch mines that people can place in the Hand of Nod that go invisible and always end up killing me.

 

oh...

[roweboat]

But yes, having played Renegade / RenX for many years, and really seeing the same problems, I think it does come down to the nature of the gameplay. One of its strengths is also its faults.

The reward of a team coming together and taking down an enemy building is ALWAYS so satisfying that honestly even after nearly 15 years of playing "command and conquer" mode, it never gets old when it all comes together.

But sadly, yes, the same building as a loss to the other team is extremely detrimental and since all buildings have an immense amount of sway to one team or the other (some effects more immediate than others [if not a loss actual gameplay options, the player's ability to experience the whole gamut of options is greatly reduced]).

 

It's like how the stats show, there's always a chance the other team could come back... and its those few times that I've seen it happen that make it worth coming back for. in fact, one thing the numbers don't show is how gratifying those "23%" or "12%" matches are when they come together. 

[Tytonium]

1 hour ago, roweboat said:

I think it does come down to the nature of the gameplay. One of its strengths is also its faults.

 

8 hours ago, Rups said:

Once you lose a building everything just spirals downwards even further. Losing bar/hon usually effectively ends the game.

 

8 hours ago, TomUjain said:

Team balance is a tricky thing to get right, and even a manual shuffle fails to perfect it at times. The fault, I believe is how the game is designed. In a nut shell: a bad team is more likely to lose a key building, further and permanently handicaping the already bad team.

I suppose I spent a bit too much time on that topic.

You are all absolutely correct. There is a snowballing effect in this game, but it typically happens during the mid-game. The early game though is really just a sort of all out anarchy type of situation for both teams. Maybe kill the harv, or take a silo or course, but in the early game players will get the majority of their points from just killing other people. That essentially stops at about 5 minutes and the mid-game starts, and that is where the snowballing begins.

My stats deal with the early game pretty much entirely and allow for you to predict the chances of success. So in essence, I can figure out if the teams are notably unbalanced due to one team being higher on the scoreboard (indicating one team is overwhelmingly better than the other by sheer skill in infantry combat, and usually this will mean all forms of combat and better game sense).

Now this issue CAN be resolved if a proper balancing method is found and implemented.

I absolutely understand that this game helps the winning team along during the mid-game and can lead to an easier time achieving victory, and I am fine with that. The mid-game is often times my favorite part of a match. What I do have a problem with is the fact that a game is essentially over in about 3-5 minutes just by looking at the scoreboard.

"Oh look at that, NOD is overwhelmingly at the top of the scoreboard and it has been 4 minutes! Looks like I have roughly a 7% of winning this game, very nic- GDI INFANTRY BARRACK DESTROYED"

I dunno, I just feel like something can be done to make matches more balanced for a bit longer if the teams are balanced properly. Eventually one team will always get the upper hand, but having one team be inherently advantaged (having better players) at beginning of the match is a bit saddening.

 

[euan-missile]

@Tytonium Like we see with Thanos, balance requires stability and stable this game is not.  It's important to remember,

The inherent design of Ren X is deliberately unbalanced, this goes from everything down from weapons, vehicles and to map design. Like i've described before how GDI is a hammer and Nod is a knife.
So no one team is designed to have an equal hand, so from the start no one team is designed to have the upper hand, always just different one.

Which is where this game is different from others,  power can shift in a blink of an eye all because one person failed to uphold their task. That's the nature of this game and that's why it's different to others. - Many other multiplayer games care incredibly little for the action of one person, unless you're the last alive in S&D in CoD. The term "one man army" has never been more applicable to an online game than Ren-X, and in some multiplayer games is considered an offence to your team, but in here it's a style.

 

But people treat Ren-X like a standard FPS game, either they'll get infantry characters and play it like CoD, dying in seconds after charging aimlessly at the enemy, or they'll play it like Battlefield and get a vehicle and do the exact same thing. 
No one stops to consider, how their actions will affect their buddy next to them and then subsequently the whole damn game. 

How many times have you been on a rocket rush, been late, missed your shot(s) and the building is on that mystical 0 hp? then your team loses.

 

Everything single damn thing has a domino effect, whether you lose your $1000 character/vehicle to an enemy's $500, be late for a rush, repair the wrong vehicle, miss that one damn shot, or vote roweboat to be commander. You too can cause the entire demise of your 32 player team. 

 

Here, I updated the recommended specifications to play Ren X.

 

• Windows 7
• 2.7+ GHz multi-core processor 
• 8 GB system RAM
• NVIDIA 200 series or higher graphics card
• 15 GB free space 
• Brain 

 

Which is funny, because when people use their brains and start coordinating together, people call it Stacking because they're frustrated how OP it is, even though that's exactly how this game is supposed to be played (which is why it's powerful) even when just two-three people coordinate together versus a whole team of headless chickens, they're gonna get the cuck, and i dont mean the rooster kind either. 

 

Anyway, long story short, my take away is this,

Chinese. It's my favourite food.  

 

 

 

Edited December 20, 2019 by euan-missile

[roweboat]

52 minutes ago, euan-missile said:

vote roweboat to be commander.

...wait a minute.

[roweboat]

Very good points @euan-missile ... well almost...

 

Erm but yes that's the glorious nature of the game being a REAL TIME STRATEGY FIRST PERSON SHOOTER.

 

Strategy is key.

 

Even though half the time I'll randomly charge the enemy because what's scarier than some dude randomly charging at your for no hope or reason. 

[RypeL]

JFYI playerstrenght for Teambalancing is currently calculated like this: ln(lastroundscore) + ln(lastroundkills - lastrounddeaths)/2   (with ln meaning natural logarithm and lastroundkills  are only taken into account if > lastrounddeaths)

That its just lastround score leads to players that just connected dont having a score for the balancer to take into account etc. So a first step to improve this could be to persist like the average score from the last three matches for every player so the balancer always has something to go by.

[Gex_str]

2 hours ago, euan-missile said:

2.7+ GHz multi-core processor 

UE3/UDK only uses one core(blame Epic games for that), so it's better to have a single strong core for RenX.

[Tytonium]

1 hour ago, RypeL said:

 

That its just lastround score leads to players that just connected dont having a score for the balancer to take into account etc. So a first step to improve this could be to persist like the average score from the last three matches for every player so the balancer always has something to go by.

That is a very intriguing idea, and I like it. It would certainly be interesting to see how matches play out with a system in place like this.

[Mystic~]

I don't really understand where you're going with this in relation to team balancing? You say your data and conclusion shows that the winning team based on score at the 5 minute mark, becomes the final winning team at game end; that's useful stuff to see in statistics and it suggests that early game is really important (most experienced players know this already) and that almost 9 times out of 10 if your team occupies more top score leader board positions (has lots of kills) your team is highly likely to win when past this point.

Maybe you could popup a Surrender vote at the 5 minute mark. The probability of winning is 7%; would you like to surrender?

Usually which ever team manages to destroy the opposing team's first harvester and save their own are going to quickly power up and win the match, so long as they capitalise on the opportunity and don't just waste their economy by being killed/vehicles destroyed too easily. It would be good to have some sort of in-game accessible statistic for commander wins/losses as this makes a huge difference and can often turn a game around entirely. 

I don't think I agree that just because score is balanced it's an enjoyable game - could be an awful stalemate.

The better team doesn't always necessarily mean they have the better players, although it helps, especially if there's a competent commander.

[Daxter]

biggest flaw where you could actually balance the game out is the way exp are contributed to the players.

 

whenever a team loses in a fight and is under siege, the other team gets lots of exp by e.g. destroying the harvy multiple times, getting lots of exp attacking buildings and getting HUGE amounts of exp when destroying a building. this in turn makes it even more difficult for the losing team to strike back. not only did they lose a building, the enemy is often one rank higher due to the amount of exp they got when destroying a building. this system really favors the stronger ones and not the weaker ones. since the introduction of this system we face a lot more stomping matches than before.

 

next up are the ratings of a player. the system doesn't recognize who plays e.g. a hottie or a sniper a lot. more often than not you might have one team with lots of hottie / tech players and on the other team a lot of lone wolf snipers or tank drivers that don't have any hottie / tech players at all. a K/d, tank and repair score would help out a lot to determine what the players strong abilities are.

Edited December 23, 2019 by Daxter

[Reivax]

hi  

hmm.. the problem about balance is known, but i think it's impossible to balance it in PUB
i.e., my case:

my skill depends of my mood and my concentration, i sometimes play like a newb and sometimes.. i feel better cuz it's the Kane's will  
i'm more focused on KD now (yea, it's one of reasons who explain why i've changed my name   ) because teamwork is rly interesting, but i've often been alone to try teamwork stuffs, so.. now i try to play alone, headshot all these little gunners who try to strike my holy Hand of Nod, hunt these little candies - i mean, SBHs -, crush these littles rabbits under my wheels - i mean, engineers - or just doing my shitty stuffs

noooow about balance

i think i'm not alone to focus only one part of the entire gameplay possibilities on RenX. after all, focus a part is fun, because you'll become good, and when you're good you can find new challenge by trying something else like AT inf or tanks  

but sometimes, players will just chilling around instead of playing seriously. and sometimes they will play other stuffs than usually.
it's why i think you can't really balance a game by picking players or just using their stats like KD, mine disarming, nukes/ions uses, tank kills counts, bldng shoots, etc..