Community Question: Measuring Player Skill

(tokamak) #201

The thing is, battlesense isn’t what wins you the game either. It contributes to the victory but it’s by far not the only thing that matters.

(Breo) #202

Very true tokamak, but the thread says: Measuring Player Skill, probably I took it too literally :confused:
Maybe adding another multiplier with (1 + (completed objectives/4) * battlesense) on end of game?

(Humate) #203

We used a stats site called rystats for all our competitive matches.
Ive grabbed a few matches just to see if there was a common thread in all of them, and wacked it into a spreadsheet

KD per team was always higher for the winning team.
Kill totals were always higher for the winning team.
Average KD per player was always higher on the winning team.

In BF3 deaths arent counted if the player is revived, but the kill for the shooter is counted even if a player is revived. ETQW obviously doesnt have that so the stats are far more clear cut.

(BrightIs) #204

Skill Points per Min. XP can be used for improving perks and rank etc.

Skill Points to be rewarded as follows.

Individual (Direct)

  • points for a kill

  • points for the skill level of the enemy player killed. The higher the skill level of the enemy killed the more points rewarded for killing them, the lower the skill level, the fewer points.

  • points for how many kills without being killed. For example 1 kill worth 100 points, the 2nd kill in a row is worth 150, the 3rd kill 200 point etc.

Team (Indirect)

  • points for the amount of damage inflicted/health taken away from enemy. This enables those who inflict a lot of damage but have their kill stolen to still be rewarded and those who assist in a kill to be rewarded proportionally. As well as those who kill a weakened enemy to not be rewarded as much as for killing a fully healthy enemy.

Medic; + points for healing a player that immediately goes onto kill or do objective. Instead of points for each health pack used which can be abused.

Field Ops; + points for giving ammo that is used to kill. Instead of points for each ammo pack used which can be abused.

Engineer; + points for repairing a gun that is used to kill

  • points for doing an objective that enables team to progress such as destroying wall or building bridge.

Objective (Overall)

  • points for saving the game such as diffusing dynamite to recapturing object.

  • points for winning

All totaled up and divided by minutes played to create a skill point per minute score.

(Joe999) #205

The best thing I’ve seen so far and which always caught my attention was ET-PRO’s killingspree messages:


K_Sprees = {				
     [5] = "is on a killing spree",
    [10] = "is on a rampage!",
    [15] = "is dominating!",
    [20] = "is unstoppable!!",
    [25] = "is godlike!!!",
    [30] = "is wicked sick!!!!",
    [35] = "is real POTTER!!!!!",
}

as well as deathspree messages:


seems to be having a bad day!
is on the way to getting victim of the day!
is getting his ass kicked!

the other stuff is just numbers

(tokamak) #206

What you’re doing now is taking a selection out of the xp distribution and applying more absolute values to them. I’d love to hear your reasoning behind this as I personally don’t yet see how reducing the fidelity and inclusiveness of something already in place is preferable here.

(Humate) #207

rystats

:oppressor:

(matsy) #208

I’ve always rated accuracy. I prided myself on getting 100% accuracy with the syringe in ET.

(Humate) #209

When everyone is fragged, its a bit hard to miss.

(tokamak) #210

My disguise accuracy is terrible :frowning:

(haternet) #211

Just FYI: ETPro doesn’t have killing spree messages. If you saw them on an ETPro server, it was most likely running ETAdmin Mod.

(Senethro) #212

Xp systems don’t work because bad players vs. bad players will lemming on to an objective, spamming abilities and having wacky random hi jinx. Good players will have a small number of short lethal firefights and when the attackers gain area control they do the objective in one go.

The only system that will work is one that uses difficult to fake measures, and awards you score based on the difference in global ranking between yourself and the people you are playing against. That you are compared against the measured skill level of your opponents is the important bit.

Some measures might be:
Offensive KPM
Defensive KPM
Accuracy and headshot accuracy compared only against similar/identical weapons/classes
Some battlesense like score that measures how engaged in the fight you are and if you are surviving. Can even give a nod to if you are near an objective or whatever.

Also lol at the people who don’t understand elo.

(tokamak) #213

My issue wit this is that it, again, excludes everything else you can do in the game besides killing. It doesn’t matter if you’re the best cunning saboteur or the most heroic medic in the match, it’s that guy that kills the most that will ascend the ladded as the ‘most skilled’ player.

[QUOTE=Senethro;403383]Xp systems don’t work because bad players vs. bad players will lemming on to an objective, spamming abilities and having wacky random hi jinx. Good players will have a small number of short lethal firefights and when the attackers gain area control they do the objective in one go.
[/QUOTE]

You can account for this by making the victory modifier gradually higher in games with higher average ELO ratings per team. If you’re not playing effectively then you will drag down your team’s average. That small modifier will accumulate constantly throughout the matches you play and you’ll end up with lower xp/min rewards from it. Of course it also means lower xp/min penalties, all in all it means the outcome of the match matters less which in turn means that lemminging, a highly ineffective way of generating xp/min isn’t getting you anywhere.

(ailmanki) #214

@tokamak
You are creating a lot of rules for a probably simple problem. And with time, you will find more rules you will need to apply, ad infinitum.
But yes, with enough time and balancing, it can work well. Yet I believe for a fun and intuitive game, there should be not many rules for scoring, at max 7 (3?!) rules (more I cant consider in my head I guess at same time…). While some nerds surely enjoy exploiting a system which has ton of rules; it would not attract the mass I believe for a fast paced fps at least.


(Breo) #215

Then you should make win/lose only be counted when you only play the full game on one team.
Join 1 sec on winning team and you win on the other hand you can lose in 1 sec aswell. Seen enough traitors who always switch to winning team :frowning:

(tokamak) #216

Now you’re forgetting that we’re already talking about xp/min here. It doesn’t matter how long you took part in the match, because the time is already accounted for the modifier will always be the same. It’s just that the weight to your overall score depends on how long you played. So there’s simply no way a player ever gets undue credit for winning or undue blame for losing.

(Senethro) #217

Yeah, thats the thing though. Once you’re out of the chump skill level there is no cunning saboteur or heroic medic because they are quickly and efficiently killed. The guy that kills the most, as long as he picks the right area to fight in, is the most skilled player.

So given that cunning and heroism are difficult to measure, largely irrelevent past a certain skill level and usually rely more upon your opponents failing rather than you succeeding, its easier to stick to things like KPM, acc% etc.

You can account for this by making the victory modifier gradually higher in games with higher average ELO ratings per team. If you’re not playing effectively then you will drag down your team’s average. That small modifier will accumulate constantly throughout the matches you play and you’ll end up with lower xp/min rewards from it. Of course it also means lower xp/min penalties, all in all it means the outcome of the match matters less which in turn means that lemminging, a highly ineffective way of generating xp/min isn’t getting you anywhere.

There are all kinds of ineffective behaviours that generate large quantities of XP. Its much more practical to not allow them the chance to get in the way than to spend months of research adjusting them all to a “correct” and then readjusting them with every bug, patch or shift in the meta-game.

(tokamak) #218

[QUOTE=Senethro;403400]Yeah, thats the thing though. Once you’re out of the chump skill level there is no cunning saboteur or heroic medic because they are quickly and efficiently killed. The guy that kills the most, as long as he picks the right area to fight in, is the most skilled player.

So given that cunning and heroism are difficult to measure, largely irrelevent past a certain skill level and usually rely more upon your opponents failing rather than you succeeding, its easier to stick to things like KPM, acc% etc.[/QUOTE]

If a game truly disintegrates at high levels to such a point that everything but combat becomes irrelevant then there’s something seriously wrong with the mechanics.

There are all kinds of ineffective behaviours that generate large quantities of XP. Its much more practical to not allow them the chance to get in the way than to spend months of research adjusting them all to a “correct” and then readjusting them with every bug, patch or shift in the meta-game.

The signature of an SD game is that shooting isn’t everything. There’s so many shooters that do the whole exclusively combat+simple objectives thing so well that I don’t really see why SD needs to get in on that niche as well. It’s the stratagems that exist around the combat that bring so much narrative to match.

I think I’ve named all the ‘xp-leaks’ there are in this thread and they get solved through the result (round or match) modifier, the ‘hot’ zones and players. The ‘hot’ tags are for context sensitivity and the modifier ensures that even without an accurate system a player will still need to ensure for himself that he actually wins or it’s bye bye improved xp/min score.

I haven’t actually seen anyone comment on them in depth or come up with a situation where statpadding can still happen. Yeah I’m pushing an agenda here, I do believe that this is something incredibly strong and perhaps even a new selling point for the game. I love every bit of it but I really need other people to try and put dents into this idea (as opposed to ignoring it or dismissing it outright) to make sure it’s not just me.

Can we just do that? Actually discuss the ideas on their merits? Can someone just please tear my idea to pieces?

(INF3RN0) #219

I think the reality is that at a certain stage of meta-game mastery, the stuff you mention does not hold equal impact weight to killing skill via aim in terms of influencing the outcome of a match. Aim is possibly the hardest skill to master and thus tends to have the biggest impact. These non-shooter variables simply give opportunity to reward or give slight advantage in situational circumstances, but in the end they are still used as a means to the same end- killing. Brink actually tried to force a sway in the balance with all of the buffs, which added absolutely nothing to the meta-game except requiring people to press their f key. There’s a lot of actions being performed during a game, but at a certain level of understanding/skill the only successful actions result in kills/gibs. I think that multipliers based off of the results (and locations ofc) of actions rather than simply the actions themselves, would be a more accurate means of measuring rewards; as well as the overall result of the game w/l. For me- it’s not that your ideas are necessarily disagreed upon, but it’s more so that your agenda is so specific and personalized to your performance/experiences that it becomes much too overcomplicated and often non-essential. Most of us want to keep our FPS priorities in the usual order.

(DarkangelUK) #220

Seems the only plugs have been theory and convolution with a dash of hope. At the risk of repeating myself cough how does this system recognize repetition vs requirement? An engineer repairing the entire teams deployables that have been getting struck from a distance, is he amounting XP or is he using the deployables to keep the enemy at bay? Those AAT/AVT/AIT’s are holding the enemy back, but they’re not all his. On one hand it could be someone that realises that those are key drops that need to be kept active, they’re not at the objective but in a heavy traffic area en-route to the objective so he keeps them going winning the game… awesome guy.

On the other hand he’s an XP farmer, he doesn’t give a crap about the objective, he’s gathering XP and sitting on the by-line waiting till they’re damaged then sneaking in for the repair.

Either the XP farmer is considered skilful and raised up the ranks, or the awesome tactician is considered a farmer and held down the ranks. Which one is it? Lemme guess, more modifiers?

Basically you’re putting value on menial, non-skilful tasks and the system has no way of interpreting intention, luck, abuse or tactics. You plugged nothing dude, you’re not getting replies because you’re not convincing anyone of anything.