

Average Score Or Cherry Pick?
#1
Posted 30 April 2013 - 09:24 PM
I would imagine getting an average of your score would be more representative, as long as there is a minimum number of games needed. It would also be the ultimate solution to grind since the more games you play the harder it is to move your average with a single very good or very bad game.
By average I mean adding up the scores for all your games and dividing that by total games played.
#2
Posted 30 April 2013 - 09:36 PM
I like the current setup. The best pilots are rising to the top at a consistent rate, and the lucky guys/cheaters are at a relative minimum. That is the best we can hope for, and it is working pretty well.
#3
Posted 30 April 2013 - 09:55 PM
PEEFsmash, on 30 April 2013 - 09:36 PM, said:
I like the current setup. The best pilots are rising to the top at a consistent rate, and the lucky guys/cheaters are at a relative minimum. That is the best we can hope for, and it is working pretty well.
The only reason he can get away with this is that it doesn't impact him either. The games where he team kills don't count in his stats.
Just punish team killing very severely so that any time this high ranked pilot does this he is destroying his stats a lot harder than the ones of the guy that gets killed.
After today I found that what I don't like about the current system is how you just need to get "those games" where your team is good enough to not throw the game, but not good enough to get kills. Where the enemy team isn't good enough to kill you, where they stick relatively close together so you get to damage and kill all of them.
Plus its kinda boring to get your top games in the first hour or two and then play 5 more hours to get a tiny up in score. Average system would mean every game counts which feels more fun to me.
Edited by Inyc, 30 April 2013 - 09:58 PM.
#4
Posted 30 April 2013 - 10:05 PM
The current leaderboard challenge does the best of both worlds here. 5 matches exemplify the best you can manage on a given day. The Overall board is 20 matches, which really is quite a few, and represents the sort of "average" that you are looking for. Perhaps upping that best 20 matches to 30 or even 40 would help every game feel more meaningful, and avoid the concerns I have raised. I'm open to that in future tournaments.
I also contend (again) that the remarkable consistency that top-ranked players are displaying every day is strong evidence that the system is non-random, skill based, and working.
Edited by PEEFsmash, 30 April 2013 - 10:12 PM.
#5
Posted 30 April 2013 - 10:13 PM
PEEFsmash, on 30 April 2013 - 10:05 PM, said:
I've got a few thousand games under my belt and I team killed accidentally all of once...
Plus I think you're trying to build around griefers too much. Griefing is a fringe event that should be delt with on a case by case basis with reports and bans.
Though having the "best of X" match thing count a lot more games would be nice - but that would only really work for short tournaments because a lot of people would complain about having to play 40 matches per day (that comes out to over 3 hours at 5 mins average games).
#6
Posted 30 April 2013 - 10:46 PM
#7
Posted 01 May 2013 - 01:36 AM
Inyc, on 30 April 2013 - 10:13 PM, said:
I've got a few thousand games under my belt and I team killed accidentally all of once...
Plus I think you're trying to build around griefers too much. Griefing is a fringe event that should be delt with on a case by case basis with reports and bans.
I've accidentally team killed a bunch, sometimes several in the same game. Zoomed in you don't get IFF idents and can be easy to mistake a team mate for an enemy. If they're shut down they lose all idents, and I know I've completely cored a light on my team from full to nothing because he shut down behind their team and I annihilated him before he powered back up. Shooting through trees, mistaking the red square being behind a blue one, everybody being white (or blue) heat vision blobs, team fire is very easy. You only need to hit them with a single laser, once, and that might be enough for the kill. As for the several in the same game, that was just dumbass PuGs trying to get the killing blow and stepping into my alphas.
Regarding griefers, you greatly - greatly - underestimate the lengths people will go to grief others in online games. When I used to play Halo a game without griefers, teamkillers, mechanics abusers and just plain old trolls was such an anomaly that it really stood out.
Actually on topic, we've all discussed the issues with cherry picking results. However what really brings down the "Average" is the face that the drops are solo, the matchmaker is terrible, Elo is shaky at best and so many bugs plague the game that having every game contribute to your score would just be demoralising. How often does a guy D/C right at the start of the game (on either side)? How often does the matchmaker not even give you a full team? How often do you look up, see seven Steiners on the enemy and think "Well, ****"? How often do you have no HUD, no map, get stuck in terrain or CTD yourself? Having all of these games tank your score before you can even start playing would be worthless.
#8
Posted 01 May 2013 - 02:53 AM
yeah the disco/uneven match numbers/bad teams/ninja caps 1 minute into the match/running into premades on the other sides multiple times where person B misses them all/whatever else i forgot issues are going to hurt/help everyone. but if they're only scoring 20 matches no one is really going to be costing themselves a huge amount of time and effort if the luck of the draw is against them. plus no one has to turn themselves into a still somehow functioning centurion at 11% health still running around the battlefield shooting things to reach the leaderboard.
also helps the folk who opt in and realize they need to play hundreds of matches just to break the top 20 of the leaderboard if they're lucky and pretty much give up on the idea immediately but still get counted because they opted in.
#9
Posted 01 May 2013 - 03:00 AM
All problems the current system has would also be present in a set number moff matches setup: Farming friends that just stand there aso.
[edit] i liked the idea off 5 best + 5 worst matches (still client stability would have to be improved) and someone proposed to have a tournament that wasnt announced until its over

Edited by Budor, 01 May 2013 - 03:02 AM.
#10
Posted 01 May 2013 - 04:36 AM
farmers would have a bit bigger risk of dropping with good players on their side (not counting 8 mans of course) which could disrupt their efforts too. happens once of twice when you're playing 50+ matches no big deal- plenty of chances to make it up/drop those matches from the score, happens twice in 20 and that's 10% of your matches where peak numbers won;t be achieved.
main issues/problems are still around but in mall enough doses so more people could feel they'd have at least some chance which many don't feel they have atm. i'm just looking at it in the time available to play perspective which eliminates people before they start.
i do like the idea of the stealth tourney though

#11
Posted 01 May 2013 - 06:17 AM
#12
Posted 01 May 2013 - 06:44 AM
Edited by Joseph Mallan, 01 May 2013 - 09:44 AM.
#13
Posted 01 May 2013 - 12:29 PM
I really dont know who of PGI think this current system is a good system. Its terrible.
Just a pure average system of all the matches played with a min of 10 games is way better than this.
#14
Posted 01 May 2013 - 01:30 PM
Joseph Mallan, on 01 May 2013 - 06:44 AM, said:
Why bother with a tournament at all, just give the prize to the player with the highest ELO score.
The only problem I have with the Cherry Picking method is the ability to "game the system", but it is a better solution than my first 10 matches count, and on the first match we had 3 disconnects, and the second I crashed to desktop. End of tournament.
Edited by Agent 0 Fortune, 01 May 2013 - 01:31 PM.
#15
Posted 01 May 2013 - 02:47 PM
This would pretty much remove lucky factor and unlucky disconnects, crashes etc.
Random factor still would exist in maps, if someone is running very hot build and he gets lucky and get to play on 7 cold maps he can perform better.
Also its used in some sports so why not ? ;]
#16
Posted 01 May 2013 - 03:23 PM
Even better, find the standard deviation of a player's score and toss all the values that are a certain number of deviations outside the mean.
Winner is highest average score. Tie-breaker is lowest standard deviation.
#17
Posted 01 May 2013 - 03:46 PM
It's a difficult balance but I'm sure a certain number could be reached that makes sense. Like any time you destroy a component should be a full assists = to a kill. Any time you deal more damage to a component than it has armor should be the same as well. Something like this.
I mean you look at the current system... you get a game with 3 kills, 3 assists, 700 damage where you don't die and you win. What is that game worth? Sweet **** all. Its only 177 points which isn't even close to being a top 10 game.
And that's just stupid. That type of game is an amazing game where you carried your team to victory and it's worth nothing? That's why I believe average would be a better metric.
Hell what about this: Your score is the average score from all your wins during the tournament, divided by the number of wins you got. Sure people might game the system by voluntarily losing a game they do badly but... That'd be really hard to game. I could only see someone do it if they were the last person alive and don't cap when they could have capped to win.
#18
Posted 01 May 2013 - 07:28 PM
#19
Posted 02 May 2013 - 06:00 AM
Agent 0 Fortune, on 01 May 2013 - 01:30 PM, said:
Why bother with a tournament at all, just give the prize to the player with the highest ELO score.
The only problem I have with the Cherry Picking method is the ability to "game the system", but it is a better solution than my first 10 matches count, and on the first match we had 3 disconnects, and the second I crashed to desktop. End of tournament.
I will have to disagree with you Agent. Neither situation is acceptable. The matches with drops and crashes are the NEGATIVE fliers (Worst case) while Awesome Matches that are far above your norm are Positive fliers (Best Case).
My Average damage for the 3,700+ games I've played(since Open Beta) is around 350 damage per game. I have some 1,000+ matches & I have some 19 point matches... Heck I even have a match where I scored 1 point of damage! Why would I be rated by the 50 or so matches where I scored 1,200 damage? Is that really a snapshot of how good I am? Or is the majority of games a 350 a better indicator of how I do?
Edited by Joseph Mallan, 02 May 2013 - 06:01 AM.
#20
Posted 02 May 2013 - 06:08 AM
3 user(s) are reading this topic
0 members, 3 guests, 0 anonymous users