184 replies to this topic

[xRatas]

I do not see much value for this stat really, although I'm a supporter of meaningful statistics, so it is good someone is trying to get something counted. Considering the available (very limited) stats, only thing I'd bother to use would be winratio multiplied with KDR. Only two things that tell something about general efficiency right now are those two.

What I'd like to see, would be:
Matches, Wins, Kills, Deaths, Damage and XP(without premium) per mech, combined totals of every class and all classes together.

With those, one could actually count efficiency in a meaningful way. Although just XP/Match without premium bonus would be halfway there already. That would also give quite good balancing information on different mech chassis, or at least tell what good players usually play... : )

In this game, teams ofcourse screw any efficiency between player who team up and player who don't real hard, but teamplayers should do better anyway, if you ask me. (Otherwise most of mankind would still throw rocks at their neighbours and we wouldn't be playing computer games...)

[Monkey Bone]

20.8

According to the poll, 50% of the voters are around 4.

[BigJim]

3783 x 4.01 / 900 = 16.85

I don't understand what this figure is meant to represent though..

[Test Monkey 13]

755 / 317 Kills / Death

377 / 354 Wins / Losses
[color=#00AA00]
2.38 Kill / Death Ratio[/color]



5.07

I think in general, the current stats are not a true reflection of the player. The OP is proposing something, but can be skewed by outlyer stats.

Try using;

perhaps damage done/games played?

perhaps kills per games played?

I would suggest using:

damage done divided by damage recieved and use that.

[Carrioncrows]

lol

My KDR is horrible.

I know this because I pilot a spider.

Now i suppose if I gave a **** about KDR it would be higher. Or if games meant something. When we start getting community warfare you'll see my KDR skyrocket as I streamline my mechs and load outs of designs that give me the best chance to win.

For now, i'm just having fun.

[Prophet of Entropy]

133.0437755blahblah,


mostly do teams as a heavy or assault but i have everything heavier then a cicada mastered, yes i am an evil pug stomper with inflated stats,

but i did 5 pugs tonight cause i was bored, got 18 kills died twice lost once, was at the top of my side in xp and damage all but one game (the loss) he had 20% more damage then me but i was the 4th on our team to die he was last lived a good 90 secs longer then me. i was in an awsome for 3 of those and 2 ppc 4 srm4 stalker for the others.

[PurpleNinja]

64.2
What does this number mean?

[Zaptruder]

Until you're able to account for mech usage... this stat is not very meaningful.

To whit, I kick serious *** when piloting my DDC. I use it as my go to mech to release some pent up frustration after a few straight losses.

But for the most part, I've been building up a large library of mechs, some of which perform better than others.

[t0ast]

IMHO, we aren't going to get a truly useful metric here until we either have access to better / more stats or recognize the inherent shortcomings in what we do have and limit their scope/polling accordingly.

KDR and W/L are busted due to the wildly varied drop environments available at the moment. Without drop demographic information being tracked, we can't correct it to account for these things. The best we could really do at the moment is break pilots out into segments around how frequently they use pre-made teams, and possibly further, by their usage of various mech weight tiers. This would probably be more of a PITA than it's worth.

Avg. Damage is simply unavailable, and even if it was, it unfairly rewards missile boats and people who can't reliably hit specific armor segments (but can still land hits). This is only useful when comparing pilots in very similar roles, mechs, and/or loadouts, which is nearly impossible to track/accomplish outside of a regular pre-made drop group.

Avg. XP is *almost* ideal. It incorporates both match results and several meritable activities. Unfortunately, we can't scale what it is made up of, and the results portion of the value still shares some issues with KDR and W/L (though with much less impact). Also, there's the issue of cap vs. kill wins providing more/less XP, but unless someone is constantly making an effort towards one or the other, I suspect things will even out when spread across several hundred matches. Still, it is the single broadest measure of a pilot's usefulness/skill we have. The only problem is Founders/Premium time, and it's a pretty huge roadblock at the moment. Without a "raw" stat or exclusively polling those of us who have NEVER activated it (which has its own issues around exclusivity and honesty), an accurate comparison can not be achieved.

All that said, I don't wish to halt this poll/experiment. Even if it is imperfect as an individual metric, there isn't much that can be done about it at the moment. If nothing else, the results are still interesting and can at least help show the distribution of various stat values across the population, and to a lesser extent (whenever demographic info is volunteered), the various modes and styles of play. So by all means, carry on.

Edited by t0ast, 05 February 2013 - 08:18 AM.

[MischiefSC]

Actually, entertainingly enough, I think this isn't a bad metric to use after Elo is released to help sort people into tiers. It's a good metric for seperating premade teams from pugs.

[Khobai]

Im kindve confused why youre multiplying (kills x kills/death) / (losses)... youre just counting kills twice. why? And why arn't wins counted at all?

Wouldnt a better metric be (wins / losses) x (kills / deaths)? And maybe factor in average XP somehow?

Quote

KDR and W/L are busted due to the wildly varied drop environments available at the moment.

KDR and W/L are probably a pretty good indication of how skilled a player is provided theyve played over 1000 matches. Since those stats will tend to average out once youve played that many matches.

Edited by Khobai, 05 February 2013 - 07:44 AM.

[Bryan Bekker]

Interesting.
4040 Kill
11.35 KDR
260 Loses.
176.36

What does this do exactly? Cannot be bothered to read through 5 pages.

[t0ast]

Khobai, on 05 February 2013 - 07:39 AM, said:

KDR and W/L are probably a pretty good indication of how skilled a player is provided theyve played over 1000 matches. Since those stats will tend to average out once youve played that many matches.

By drop environments, I mean dropping with a group (of any size) vs. going lone wolf. My KDR and W/L are high because I drop with a coordinated group most of the time. If/When I go solo, group coordination/competence/ability becomes a roll of the dice, so W/L will be worse. When your team is well on its way to losing, there isn't much you can do to avoid dying. Even if I can kill as many people as I normally would during pre-made drops, it still comes at the cost of additional deaths. As such, my KDR stat would be negatively impacted by external factors. Without being able to accurately account for this, we simply can't compare pilots with different pre-made drop habits in this manner.

Edited by t0ast, 05 February 2013 - 08:04 AM.

[Grimlox]

Bryan Bekker, on 05 February 2013 - 07:49 AM, said:

Interesting.
4040 Kill
11.35 KDR
260 Loses.
176.36

What does this do exactly? Cannot be bothered to read through 5 pages.

It tells us what we already know. Anyone with a KDR above 5 and a HC above 10 likes to pug stomp. The higher the number the better they and their team are at stomping pugs.

What's funny is the absolute best players in this game running competitive 8v8's versus challenging opponents will have worst stats than your average 4 man pug stomper.

Edited by Grimlox, 05 February 2013 - 08:25 AM.

[Khobai]

Quote

What's funny is the absolute best players in this game running competitive 8v8's versus challenging opponents will have worst stats than your average 4 man pug stomper.

Theres a lot of 8-mans that are worse than PUGs.

[Carrioncrows]

Khobai, on 05 February 2013 - 09:20 AM, said:

Theres a lot of 8-mans that are worse than PUGs.

That is because it's a more balanced game play of team vs team as opposed to 4 man team and 8 guys on the side in trial mechs looking up at the stars trying to figure the game out.

[Agent of Change]

4.99

kills 1115
kdr 2.04
losses 455

[Grimlox]

Khobai, on 05 February 2013 - 09:20 AM, said:

Theres a lot of 8-mans that are worse than PUGs.

Yep... never said all 8 mans are awesome players. I said the very best players who are pushing the competitive 8 man tactics and skills to the test are going to have mediocre stats if they end up facing off against similarly challenging opponents. Of course there will be pushover 8 mans, but the vast majority of them will (and do) just go run 4 mans.

[HC Harlequin]

I guess the first thing to think about is that the numbers submitted resulted in a traditional populated bell curve. It is shallow and weighted but so far it is a bell curve. That indicates some form of validity in the comparison between ranks.

[Grimlox]

HC Harlequin, on 05 February 2013 - 06:50 PM, said:

I guess the first thing to think about is that the numbers submitted resulted in a traditional populated bell curve. It is shallow and weighted but so far it is a bell curve. That indicates some form of validity in the comparison between ranks.

Using kdr is a bad idea. It's all over the map because it depends on your team performance too much. You rarely ever die with a good pug stomping team.

Here's my humble suggestion. Use kills done over total games played by taking kills and dividing by sum of wins and losses.

Is the guy that rides his teams coattails for 2 easy kills each match and rarely dies for like a 20:1 kdr really that much better than the solo pug that gets the only 2 kills on his team before getting focused down by the other 6 mechs? He has a 2:1 kdr and your stat puts them in a whole different range by leaps and bounds. Using kills per game played is way more accurate. The team player still has a big advantage because he lives for the whole match 90% of the time so he has more opportunity for kills, but he has to compete against the other good players on his team for those kills per match.

Let me know if that makes more sense to you as well. Seems a no brainer to me.