Ladder Stats

[DarnYak]

So, assuming warden doesn't succesfully destroy ghost...

GHost has a new feature allowing maps to effectively store data. I'm not sure that it can inject data into maps yet (if ever), but that's irrelevant anyway. The main purpose for this in EotA would be to track various stats, and to ultimately form some sort of ladder that doesn't suck.

So the basic question is this: What are your guys suggestions for ways to create an EotA ladder? Or, put another way, how do we determine if one player is better than another? The team game nature obviously is what makes this a difficult call, so I'm looking for ideas.

DarnYak

[DarkNemesis]

Well I think a good way to figure this out is to divide the heroes up into sections, and set a standard (loose) of hero/troop kills and see were they fall into that subsequent category. For instance, you could have the following:

Tanks
Support
* Hero
* Unit (spawn)
Hero Killers (DPS)
Magic/AoE dmg (casters)

Obviously, this is a general category list, but imo, that's basically the different "groups" of heros. Now, for say the hero killers, high enemy hero kills in gonna be the target here. So you just set a standard (again loose) of say 15 hero kills in a game? Holding that as the average. If a kid named DarnYak comes along and plays, acquiring say 11, its pretty easy to determine where he stands as far as knowing wth hes doing in that class. The only real problem with this concept is its limited to each class. But again, if you can manage above average in one class, you'd think any other you'd at least be decent. Thoughts?

As far as the other groups go, namely support and tanks, it could be trickier. For support (spawn) a way to determine that persons skill probably would be A) what they buy: items, consums, mercs, and b) which skills they put points into. A good kid playing BP probably isnt going to buy much in the way of items, maybe int and mana, armor to stay alive and keep healing/buffing troops. Too his points would probably be spent into skills other than backfire. But thats just me.

Tanks, I guess you could base this one off of how many bases they help wipe or how often they die?

These are just wild assumptions I'm throwing out here. So feel free to add your own ideas (obviously).

[Ion]

It is complicated, I will throw out some things that are cool to me and I think would be good for a ladder.

- Overall games played, and win/loss record.
- Win/loss record with x hero, x race, on x map or with x comps on one team (might need different determiners for 4v4, or 5v4 with one leaver x minutes in, etc.)
- Unit kills, hero kills (and assists?), tower kills (and assists?)[and (unit/heroes saved? could be determined if a unit is under x% of health and is healed)], deaths per game, mean time between deaths. I'm going to refer to all of this just for this as n. The mean of n could be calculated both overall and for specific heroes. The overall mean of n is kind of useless just a fun fact, but specific hero means would be awesome (determines kills per game basically).

If they were calculatable, this would just be fun/cool facts but wouldn't necessarily have to contribute to one's ladder standing

- Mean time spent in game.
- Win/loss record with x color.
- Win/loss record with x person (see below for clarification)
- Spawns built per game (or top three spawns built most often, or spawn tower built most often or something like that).

You could have hero specific ladders which show the stats of people and the heroes per game, a general ladder for overall/win loss. Because the x person would get too complicated, the only way I can see this working is if it's possible to make a function which could count a threshold number to determine whether the statistic was relevant and using that specifically for certain players (i.e you play with a player 10 times and then ladder will include this statistic).

[noworries]

Seems really hard.

Tracking unit kills/building kills seems stupid. Most of the time last hit within those is luck if it's not done by a skill, and as far as units go not every hero is great AoE dps.

As far as hero kills goes it doesn't seem right to use those as a measure of skill either. A lot of that is last hit, or kill stealing, and another huge part in a hero kill is how good the opponent is. If you're playing vs people who downloaded in the game you're in, you're going to get a lot of hero kills, that doesn't mean you are better than someone them, you just know you way around the items/heroes/map.

I would like to say the amount of building hp repaired and amount of mercs bought should have a big influence on it, but thats limited to gold as well, dominantly comes from hero kills or unit kills, and some heroes are better than that at others.

Wins/losses doesn't seem an effective way either, team stacking happens a lot. You can play your heart out but if there are 3-4 people on your team who have no clue what they're doing then you aren't going to be very lucky against 3 people who play the game regularly.

i managed to provide no ideas in this post.

[Kalrithus]

Just goes to show you how immensely difficult it is to create a fair ladder in a game with so many variables.

[Ion]

Tracking unit kills/building kills seems stupid. Most of the time last hit within those is luck if it's not done by a skill, and as far as units go not every hero is great AoE dps.

As far as hero kills goes it doesn't seem right to use those as a measure of skill either. A lot of that is last hit, or kill stealing, and another huge part in a hero kill is how good the opponent is. If you're playing vs people who downloaded in the game you're in, you're going to get a lot of hero kills, that doesn't mean you are better than someone them, you just know you way around the items/heroes/map.

It depends on how Yak can structure the ladder. If it could track assist triggers it wouldn't be such a big deal, but I think the big thing you missed (fair enough it was a phat paragraph) my saying that overall these stats would be largely useless in general. However, I think for specific heroes it's very important. In judging how good an Inf. is or a Harpy is you really do want to look at how many PKs they get. For Gravel, assist building kills, unit kills and building kills are relevant. If assists can't be counted it's more for fun, but it still reflects some degree of skill which needs to be taken into account.

That's why I said for specific heroes, since you could judge each hero then by other heroes. It makes the statistics more fair and relevant. You do however bring up a good point in that it would be hard to tell whether you're playing vs. pubs or not. It just depends on what's available in constructing the ladder =p

[Dekar]

You could also play a Smokecloud/Explosives/Sabotage Inf with the main purpose of AoE damage against waves and sieging.
So obviously it depends on the skill choice, how a hero should be rated in different categories.

What would certainly be interesting is to track the skills/damage/heal/kills/deaths of every hero in 5 minute steps in addition to the map played. Primarily for immediate balance purpose, but if enough data has been gathered after 6 months you could use it to balance ladder ratings on it.

The tracking of different stats reminds me of an old thread where you asked what we wanted to see stored in a new code system.

[Ion]

Yeah Dek's idea is good. It would be a lot of data to get though =p

[BustroQuick]

Might be an idea to also take into account what players you're up against, so people playing against those with better stats could get more experience, if that's how the ladder works.

[Rockness]

The most simple way to go about this in my opinion (theoretically, I don't have a clue about coding) would be to give each player a rating each game, the ladder would average out the rating based on the number of games played.

Say for instance the rating was on a scale of 1 to 100, 1 being the bare minimum and 100 being optimal.

There would be categories which gauge a players performance, each worth a certain number of points. Seeing as how heroes serve significantly different roles some categories would be worth more than others (depending on the hero being played).

So just for now lets say the categories are as follows:
Heroes Killed (in relation to number of deaths)
Units Killed
Hero Level
Total Healing (from skills)
Total Damage
Buildings Produced/Repaired
Buildings Destroyed
Units Hired

So that's 8 categories, although not every category applies to every hero & chances are there is something I left out. In any event I'll continue just to lay out the basic concept I have in mind.

The most important categories in my opinion are Hero Level, Hero kills/deaths & Units Killed, thus they should be worth more points. As should total healing for heroes with heals.

So basically each category is worth a chunk of the 100 available points; the following can be used as an example:
Hero Level: 25 points
Hero Kill/death: 15 points
Unit Kills: 15 points (10 for healers)
Total Healing: 10 points (excluded for non healers)
Total Damage: 15 points (10 for healers)
Buildings Produced/Repaired: 10 points
Buildings Destroyed: 10 points
Units Hired: 10 points
= 100 total points

Now I'm not sure about the values as it relates to what level would equal 25 points etc. or if it would even be static, that is to say perhaps it would be better to make it based on the game values and not predetermined ones as some games last 2 hours and some last 40 minutes. Thus the values should be based on time in my opinion for instance X damage done/30 minutes = full points and so forth for each category (I used X because I have no idea what would be a good number at the moment).

Anyhow, I'll wait to see how people like the concept before I go any further with it.

The best part of this system from where I'm sitting is that it encourages a balanced play style, that is to say you won't get 100 if you only go for hero kills, or unit kills. You must contribute in all avenues that apply to your hero in order to get 100.

*Note* The most important part of any ladder to me is true reperesentation, that is to say, every game counts, that means no opting to not save if/when you have a bad game, that defeats the purpose.