Win to Loss Ratio or Kill to Death?
#21
Posted 03 July 2012 - 10:39 AM
#22
Posted 03 July 2012 - 10:39 AM
The problem with experience based scoreboards is that they incentivise certain behaviors. So either they have to be thoroughly tested or the developer has to be committed to modifying them frequently.
Personally, I would be okay with just a Trueskill number and match delta being displayed for the scoreboard. It does two things:
1. Encourages winning.
2. Gives the losing side an excuse at the end of the game.
#23
Posted 03 July 2012 - 10:39 AM
kill counts have little to do with wins, for those folks having a free-for-all map/planet that does not even bother to track such a thing as a "win" should do the job.
I hope that the game evolves into an environment where strong teamwork can show itself rather than a constant gear escalation.
#24
Posted 03 July 2012 - 10:52 AM
crabcakes66, on 03 July 2012 - 10:29 AM, said:
Sorry but random chance does not work that way. I could be an awesome player and still lose 60% of the time due to terrible teammates. Go look at WoT.
Human beings =/= dice.
You really don't understand statistics, do you? The other team is just as likely to be terrible as your team (if you are a good player it is in fact slightly more likely to be terrible, because your team already has one good player in it). The only common factor over a large number of battles (again, assuming random team placement) is you. The law of large numbers that I linked earlier says that you won't be put on the "terrible" team 60% of the time; you'll be put on it 50% of the time. Any deviation from that number in wins/battles-played (discounting draws, if these are possible) is an indicator that you influenced your battles toward a win (if over 50%) or a loss (if under 50%).
Edited by Pun Pundit, 03 July 2012 - 10:54 AM.
#25
Posted 03 July 2012 - 10:59 AM
KDR or Kill Death Ratio - measurse the number of time you have died versus the number of time you have killed your opponent. In itself this sounds like a good stat since it is an average of what you have done, but it can also be very misleading. If one's kdr is never reset then an experienced player who has grown in skill will have a lower kdr then their current level of play. If mechwarriors have a weekly kdr this would be more reflective of how they are currently playing, but a run of bad or good luck could also misrepresent the players abilities.
Suggestion: My solution for this (never seen it but love too) is to have your overall kdr and a weekly kdr. From this other mechwarriors could get a better picture of you skills in this area. But if you want and even better judge lets have the kdr also include a range meter that reflects the percentage of damage your doing to other mechs. (i.e. Short 56% Medium 40% Long 4%) This would then let me know that this particular mechwarrior love to mix it up at short ranges. Now I know something about his playing style and how well its working for him.
Winning Percentage - Here's perhaps even a tougher one than the kdr. I've been playing WoT's for over a year and I currently have a 47% win ration, 52% loss ratio and the other 2% are ties. On my latest vehicle though the bottom is dropping out on as I only have 39% win ratio (this is a run of bad luck), but its balance by another vehicle in which I had a 60% win ratio. So my overall win % is more reflective of my over gaming experience, I don't believe its neccessarily a refection of my over tank commander abilities (personal opinion). Winning percentages are most often a refection of simply ending up on the right side.
Now I'm not trying to nay say players skills or teamwork, but in a game where the sides are picked at random a player ought to end up somewhere pretty close to even over the life of a game. Now if you take a game like Combat Arms where players can pick which side their on then the player chance fo winning can increase significantly if he picks the skilled side to be on. So again how valuable is this tool in itself that's the question. So how could we add to this information to make it more useful?
Suggestion: Again include and over wining percentage and a weekly percentage so I see how he's doing. Then include stats for "Objectives Achieve." If this were a percentage number also and geared towards their mechwarrior skills (i.e. A scout successfull revealed all enemy mechs to his lancemates 88% of the time) I could see that while he had won or lost a lot of battles he was very good and doing his job.
For myself I am most interested in how a player peforms as a teamplayer (and I don't mean one who runs over when he see's your opponent is on his last leg and tries to steal the kill). I want the player who sticks to the plan, follow direction (using good judgement when he needs to change the plan) and who communicates whats going on in their sector of fire.
My understanding is, the in game promotional system might reflect some of these last thoughts through the rank that a player achieves. In any event the more input the devs can give us on a player the better we will be at in determining how welll a player performs. Let's face it though some of us will go down as legendary players of the game and with others talking about them for years to come. Others of us will end up as nothing more than cannon fodder or Solahma units for other's to rain down fire and with little chance of making a name for ourselve. In any case I hope all will find the game to be enjoyable, no matter what their abilities are.
Edited by CmdrSpider, 03 July 2012 - 11:05 AM.
#26
Posted 03 July 2012 - 11:02 AM
You need other methods to measure player's skill.
#27
Posted 03 July 2012 - 11:04 AM
its harder to maintain a high win ration in a teamgame because of well idots or bad layers
haveng wsome skills and working in a team are 2 different things
#28
Posted 03 July 2012 - 11:04 AM
Teams will never be random as there is (presumably) a Matchmaker involved.
#29
Posted 03 July 2012 - 11:07 AM
For example:
Mechs lit per match.
Mechs relit per match. Allows you to spot scouts who know how to light, pull back and relight as needed.
Mechs damaged per match by indirect fire while you are lighting them.
Mechs damaged per match that are destroyed.
Objectives completed per match
Mech kills per match. Not totally important, but still a live mech is still shooting. like to see them finished and someone has to do it.
Damage sustained per match while surviving. A bone for those brawlers who know how/when to engage AND when to disengage.
Victories while surviving to Losses while surviving ratio. a key metric for showing who is usefully camping vs not so much.
Completion of CO orders per match.
# of CO support abilities that hit per match. ie successful arty calls, air strikes etc.
Proximity kill participations per match. Basically how many bad guys die while you are nearby. A high number on a brawler good. High number on long range support mech... bad.
#31
Posted 03 July 2012 - 11:11 AM
#32
Posted 03 July 2012 - 11:12 AM
Nik Van Rhijn, on 03 July 2012 - 11:04 AM, said:
Teams will never be random as there is (presumably) a Matchmaker involved.
Good points. I would suggest though that stats when used to determine weaknesses and strengths for self improvement are a good thing. Used to judge who you will associate with.... not a good thing.
I would really like it if any planned stat database can be flagged public, private, or restricted to your company/clan/guild only. That way you can choose whether you want that public.
Oh and since someone will eventually bring it up I suspect.... I have and do use XVM for WoT. I do not use it to trash people, but I do use it to identify who I can expect to perform above average in a pug match and who I should not expect great things from. I also use it to identify who on the enemy team is a significant threat above and beyond what is typical for the tank they are using. ie a high rated player in a scout tank probably needs to be dealt with quickly if possible, b/c unless a rookie scout tank driver he will likely use it effectively.
#33
Posted 03 July 2012 - 11:19 AM
Pun Pundit, on 03 July 2012 - 11:07 AM, said:
The problem with win loss ratio is its a stat that can be padded by stacking games potentially. Also people try to use it to say person A with 1% better win rate than person B is a better player. The truth might be that person A gets more games in with stacked teams, while person B could be soloing all the time and is a really good player.
Case in point. Another game I happen to play I have a 60% win rate. When I originally was leveling myself through the trees in that game my win rate was somewhere around 52-54%. Then I got in clan of really good players and saw it creep up to around 57-58%. Then I left the game for RL for awhile and came back. Now the server pop is a bit different and I am able to sway more games in my favor so my win rate has crept up further even though I solo a lot now, and do some group stuff maybe 30% of the time. Am I a good player? I would say yes. Am I a better player now than when I was when I had that 57-58% win rate? I dont think so. Just my competition has changed in such a way as to favor my playstyle. As a side note I used to get a lot more awards for killings lots of enemies in matches and survived far fewer matches. Now I get less kill awards and survive more. Did I change how I play? No. Just circumstances have changed.
#34
Posted 03 July 2012 - 11:20 AM
- Make stats private by default, where no other users or APIs can draw on them to display info without a player giving express permission in an account setting.
- Have the internal account info pages, which a player can access, able to draw on anonymous statistical data to give you an idea of where you stand as a player. This allows self-development without having to reveal your stats publicly (or depending on others to do that, in order to have a frame of reference).
- Keep a copious amount of statistics for those who are interested, and have an interface for comparing one to the other - even with things that wouldn't seem on the surface to make sense. Kills to Wins, Damage Dealt to Accuracy... let players match around any stats they want privately, and see what interesting metrics end up coming out of that.
#35
Posted 03 July 2012 - 11:32 AM
#36
Posted 03 July 2012 - 11:37 AM
ArmrdChaos, on 03 July 2012 - 09:56 AM, said:
SuperDuck1337, on 03 July 2012 - 10:33 AM, said:
UnexpectedDmg, on 03 July 2012 - 10:39 AM, said:
kill counts have little to do with wins, for those folks having a free-for-all map/planet that does not even bother to track such a thing as a "win" should do the job.
I hope that the game evolves into an environment where strong teamwork can show itself rather than a constant gear escalation.
I want to belong to a merc corp or clan, with this mindset when it comes to playiong MWO. This what is needed in real life combat, and this is all I know! Semper Fi Devildogs......OOOOHHHRAAAHHHH!!!
Edited by MuffinTop, 03 July 2012 - 11:40 AM.
#37
Posted 03 July 2012 - 11:37 AM
Fire and Salt, on 03 July 2012 - 09:49 AM, said:
I strongly disagree with you. Skill is the only base for win/lose ratio. In WoT, I am usually a lone wolf and I have superb stats. I do however I try to go where I'm needed most.
#38
Posted 03 July 2012 - 11:46 AM
However the other stats should still be maintained somewhere for you to be able to see them.
I also like the idea suggested that if a player contributes to downing a larger mech while in a smaller mech that should count for something.
What ever the case maybe Im sure the dev's have already got all this figured out. They have already got a system in place to calculate xp, dish out cbills, etc.
I would say all this speculation is somewhat pointless at this stage.
#39
Posted 03 July 2012 - 11:49 AM
Does that mean that this game should not have these stats availlible to me? I don't know. However if they are present players will be judged by them, for good or ill. But I think few players want less choice in who they play with.
#40
Posted 03 July 2012 - 11:51 AM
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users














