Angeldust

YouTube Twitch Facebook Instagram Discord Twitter
iTunes App Store Google Play Store Steam (Windows, macOS, Linux) Windows

Hero battle (PvP) ratings

in English in Forums





Post 11–20 of 33

Previous « 1 2 3 4 » Next


# 11

by

Firefly

4 months ago


---[ READ POST #9 ON PAGE 1 FIRST! ]---

So the thing is that absolute 'score' doesn't even matter. If others are playing, the top player's score might drop, even though he/she is still the top ranked player. Only the ratio of your score to the score of the players around you matters.

Maybe I should always have a 10.000 score for the top player and use relative scores for all other players. This way your score better indicates where you rank relative to the top player instead of giving you a cool number that didn't mean anything in the first place.


# 12

by

obi2002

4 months ago


My fidings are that I get about half of my score from the top 4 or 5 players, interestingly enough most from space(68). While Alex is the person who EX me the most, he pulls 47(I'm also his top target).

pool of players - interconnected web of people who have EX eachother

Anyway, we currently have only 1 "pool" of players who really do pvp...but what if we had 2 pools of completely different people or even 3? The scores would change in a completely different way, deppending on what pool/s you're connected to. Just a thought.

How would you determine the top player if you used the 10k system?


# 13

by

Firefly

4 months ago


Multiple pools / webs don't matter. We already have a lot of those, mostly small groups of players PvP'ing together. The algorithm still works as intended, delivering a consistent ranking of everyone no matter in which pool(s) they may be.

As for the 10K remark: I'd just multiply all scores so that the highest one is always 10.000. That's the top player. This way you can see if you relatively gained or lost points. Otherwise you'd just end up with another weird number that doesn't indicate anything.

I mean: going from 8.625 to 8.711 is a clear improvement. Going from 167 to 166 seems like a decrease but you don't know because everyone else might have dropped, too.


# 14

by

obi2002

4 months ago


Maybe having a sort of stocks like system would be interesting, so you would know if youre the only one who dropped, or if it was everyone together.

How does the algorithm connect the seperate pools if there are no mutual targets nor nemeses?

Can't wait to have the leaderboard live and on website, it'll be fun testing it out...will it update real time(with every EX)?


# 15

by

Firefly

4 months ago


Stock like system? Like in a historical graph of your (relative) score? Maybe much later in the future :D

The algorithm doesn't connect pools at all. Which I think is the beauty of it. By calculating everyone's score given the algorithm in post #9 you actually end up with someone who's got the highest score, no matter where the points come from. Of course it's likely to be someone from a larger pool with high value targets, but there is no reason there can't be multiple, separated webs of people scoring a lot and being intermixed in their relative score.

The leaderboard will most likely go live with Angeldust v3.0—there are a few other pieces connected to it that I can't easily untangle from it.


# 16

by

obi2002

4 months ago


A historical graph would be awesome yes.

EX= when someone gets shot down
TE= target exhuastion/s

I was thinking of an "All player graph" or maybe just top 100 players, where you could see 100 lines, the top one being 10k of course. And every other player could use that graph to see whats happening in the elites.
If for example: the #1 player gets a lot of TE, it would decrease everyones relative score...The #1 would still be at 10k, everyone would just drop lower relative to how much TE he got.

The average/smaller scoring would see a drop in their score, but if he looked at the graph, he could see that the "score stock" dropped(everyone is again lower cuz the #1 got "higher"(still at 10k)). Compared to his friends the player wouldnt lose any score, only the score on the way to #1.

The stock part would be a red or green number that says how much everyone got decreased(red)- #1 got more TE, what progress has been made to dethrone him(green).

This is a way everyone could know why their score got lower.


# 17

by

Firefly

4 months ago


If no one has any objections or additional wizard insights into the math (which I think is sound) I'm running with the algorithm and scoring as written down in post #1.

I did try out relative scores with the top player always being at a fixed number and scaling the rest, but that wasn't really as satisfying as the raw, silly numbers that you need to compare. It's all raw and silly in the end anyway :)

Also: the algorithm will currently run at a fixed interval since it takes a bit of time to run. I don't want to run it after every EX for now, but maybe I can do that in the future if the load isn't too bad.

And @obi, to answer another question: I don't think players will need to EX you to surpass you. You can accumulate points from any player and eventually make it to the top, I think. We'll have to see what happens in practice.


# 18

by

obi2002

4 months ago


Sounds good to me.

I think that unless you own like 90% of a persons EXs, they cant surpass you that easily, cuz you profit from their score rising. Idk, practice shall tell how it really is.

Btw, blank accounts, like the roadies, actually give any points? They can't have any score cuz they haven't EX anyone so you cant pull from them.


# 19

by

Firefly

4 months ago


Blank accounts or players with low kills won't have a high score value and won't count for any or very few points. This was one of the major issues I wanted to resolve compared to a plain 'K/D' score/ratio in other games. The current algorithm works around that beautifully in my opinion.


# 20

by

Hummm

4 months ago


so, hypothetically, if I raised the score of all the roadie accounts, Hummm's score would rise without Hummm ever doing any more PvP? ... that doesn't seem like the algorithm you want. It seems that the zero benefit Hummm's score got at the time of the kill should forever be zero benefit. Hummm should only benefit from a roadie score increase if he kills a roadie AFTER roadie raises his score.


Post 11–20 of 33

Previous « 1 2 3 4 » Next