new exp level calculation suggestion

PVP Warfare forum for discussion, suggestions, and bug reporting for current and upcoming versions
User avatar
Deeelite
Donator
Posts: 807

new exp level calculation suggestion

Post#1 » Sat 20 Feb, 2021 12:21 pm

(as already posted in Discord )
Exp level should be more related to the kill/death ratio number rather the playtime according this new algorithm:
playtime itself is not enough at all to define players skill but surely is a good indicating value to understand the level of confidence a player has reached with eutw warfare so I suggest to let this value capable to bring the exp level to a maximum of 4/12 rising up the maximum exp level to 12 as the total sum of the following factors.
Summing up:
4/12 playtime
4/12 kill ratio (players killed as infantry)
2/12 Kill ratio (vehicles killed as infantry)
2/12 Kill ratio from planes ( players + vehicles killed as pilot)

Splitting this way the esperience level it will be much more realistic since I don't understand why my exp is 10 killing 1/10 of the players which Mudokon or many others can but having less exp. than mine.

Introducing this algorithm, players like Mud will rise up to 11-12 while I should go back to 6-7, which is much more correct and useful to balance the game when joining.
"When taking part in a discussion, the most difficult thing is not defending your opinion, but knowing what it is in the first place." André Maurois

Grenadier
Donator
Posts: 97

Re: new exp level calculation suggestion

Post#2 » Sat 20 Feb, 2021 1:26 pm

you can immediately set it to 12 so that others can see it and join to him. as long as it's on the server, there's no point in doing anything.

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#3 » Sat 20 Feb, 2021 6:15 pm

Deeelite wrote:4/12 playtime
4/12 kill ratio (players killed as infantry)
2/12 Kill ratio (vehicles killed as infantry)
2/12 Kill ratio from planes ( players + vehicles killed as pilot)


sounds interesting
i would adjust it to
2/10 playtime (would use a similar formula as the EXP has now with a hard cap to the max hours)
4/10 kill ratio (players killed as infantry)
2/10 Kill ratio (vehicles killed as infantry)
2/10 Kill ratio from planes ( players + vehicles killed as pilot)

(10 EXP for a bit more readability)

i will try to adjust my calculation programm and will try to calculate that

BTW what would you do with people that are new or dont have any kill ratio on a category?
Would it make sense to add a category for fire support (arty and mortar)?
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

User avatar
Deeelite
Donator
Posts: 807

Re: new exp level calculation suggestion

Post#4 » Sat 20 Feb, 2021 7:37 pm

Well Master, the aim of recalculating is meant to better highlight players skills in terms of in-game effectiveness. To give an example, I consider myself good enough in guarding sensitve flags which may be the enemy's next objective. I try to be there before they are prepared to attack and many times I manage to get rid of respawn vehicles or heli insertions. Wether I kill them all or none, my team is immediately warned of the incomin danger, which is the most important aspect. This behaviour of mine can even result in 0 kills but, on the other hand, 0 succesful attempts of 10 enemy tries (in an extreme example). So, yes, unless admins find a method to detect this specified kind of skill I would find my exp. level penalized (time ago, I suggested to score up the player exp. who marks first on map 'emhq' when an actual emhq/mip is nearby, e.g. ). But once players like Mud are in-game my effort get dreastically reduced since, once he reaches the flag, there is no need for his mhq to be there. Most of the times he manages to bring the flag below 75, even alone. Considering this, is neither correct having me and him at the same exp level. I need to reach his kill ratio too.
Summing up, I don't think exp should be a kind of reward for have been killing or playing a lot in yeras: its purpose should be strictly related to the real level a player is dangerous on average.
So, if it happens once on 90 days that I kill units like crazy, this shouldn't count much.
Conversely, talkin about plane pilots or tanks, exp should take in account of the different and splitted episodes since flying a jet or controlling a tank is not something a player can afford on every match. So while the exp in killing enemies as infantry could be adjusted on every match to fit the average level, the exp level in flying planes should be trimmered from flight to flight, even if weeks have passed by in between the flights. Same for the tanks.
Artillery should count as well but I would place it under the tank calculation algorithm.
"When taking part in a discussion, the most difficult thing is not defending your opinion, but knowing what it is in the first place." André Maurois

User avatar
Deeelite
Donator
Posts: 807

Re: new exp level calculation suggestion

Post#5 » Sat 20 Feb, 2021 7:47 pm

Before define the new or final exp calculator, I suggest to wait just awhile for more players suggestions. Maybe something more could be added.
"When taking part in a discussion, the most difficult thing is not defending your opinion, but knowing what it is in the first place." André Maurois

ThreeDots
Community Member
Posts: 74

Re: new exp level calculation suggestion

Post#6 » Sat 20 Feb, 2021 8:06 pm

There is multiple ways how to change the algorithm, first I would choose between these two options:
- xp based on other players
- xp based on hard cap

first is how the xp is computed now, there is no hard cap, the cap is the player with most hours and others are computed based on him
second is introducing a specific number, let's say 200h for playtime, where you say "ok, after 200h of playtime I consider this player to have 100% of xp for playtime category"

Then you choose some categories that are measurable back in time: time on the server, avg k/d, avg points/min, ...

You add weight to those categories like 30% playtime, 30% k/d, 40% points/min

Introduce some condition for low amount of data (to make sure person who just joined and killed 5 people without dying won't brake the stats)

And finally put everything together...

I don't really care what numbers you put in and what bounds you set, just pointing out the way to do it.

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#7 » Sat 20 Feb, 2021 9:42 pm

TheMasterofBlubb wrote:
Deeelite wrote:4/12 playtime
4/12 kill ratio (players killed as infantry)
2/12 Kill ratio (vehicles killed as infantry)
2/12 Kill ratio from planes ( players + vehicles killed as pilot)

...

BTW what would you do with people that are new or dont have any kill ratio on a category?
Would it make sense to add a category for fire support (arty and mortar)?


My question was about what value you want to assign to these, strictly mathematically as you need some for an avarage?
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#8 » Sat 20 Feb, 2021 10:15 pm

Deeelite wrote:So, if it happens once on 90 days that I kill units like crazy, this shouldn't count much.
Conversely, talkin about plane pilots or tanks, exp should take in account of the different and splitted episodes since flying a jet or controlling a tank is not something a player can afford on every match. So while the exp in killing enemies as infantry could be adjusted on every match to fit the average level, the exp level in flying planes should be trimmered from flight to flight, even if weeks have passed by in between the flights. Same for the tanks.
Artillery should count as well but I would place it under the tank calculation algorithm.


my current method would imply that the calculation is run periodically to get new data (the initial set will take quite some time to run, the data collection alone takes about 16h of 1 Core CPU Time, though it can be made parallel very easy, and i can go back to almost the start of available records) im not sure on how good it will run live in terms of taking live kills into account, but it would
still be more meaning full than the current EXP
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

User avatar
Deeelite
Donator
Posts: 807

Re: new exp level calculation suggestion

Post#9 » Sat 20 Feb, 2021 11:42 pm

BTW what would you do with people that are new or dont have any kill ratio on a category?


New people = 0 exp as usual
People with no kill ratio will gain exp from playing time but only up to 4/12

Of course we will have to assume that 4 exp means the current 10 exp of playing time, so a new player having 4 exp after few weeks of eutw will mean now that he must be a good one since part of that scoring has to be fruit of his killing skill and, trust me, you CANNOT kill a lot in our servers if you don't understand how to play. You have to be reactive, fast in teleporting, defending and all this IMPLIES you have understood enough, quite accurately, up to the point that a new player having a costant good kill ratio should have the time exp algorithm rised up quiker.
We have never seen new players collecting tons of kill in one match but unable to understand how to play or move. If you kill a lot, this means you have understood enough about eutw engine. Of course we can meet players who teleport to MHQs having not experience on driving/placing them at all but in 95% of cases they are supposed to learn that part easily. What we have now is players with high experience that still don't move mhqs properly or at all. And yes, also moving mhqs should be nice to implement in the new exp formula, but we don't wanna make things too complicated.

Having this new calculator working we wll know that 6 is already a good level but 4 doesn't mean much. 4 means only you are a well trained member, which not bad

Would it make sense to add a category for fire support (arty and mortar)?


Not that much, since arty effectiveness is so aleatory for everyone: sometimes you strike out everyting, sometimes you score 0 kills. I would count it as infantry kills, to be honest

What about composing the exp number by two digits? The right one tells about your playtime, while the left one your kill ratio, each from 0 to 9.

So 19 could be my exp (1 in kill ratio and 9 in playtime), while Mudokon playing eutw for the first time would have 90 :)
Think about
"When taking part in a discussion, the most difficult thing is not defending your opinion, but knowing what it is in the first place." André Maurois

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#10 » Sun 21 Feb, 2021 12:19 am

Deeelite wrote:
What about composing the exp number by two digits? The right one tells about your playtime, while the left one your kill ratio, each from 0 to 9.

So 19 could be my exp (1 in kill ratio and 9 in playtime), while Mudokon playing eutw for the first time would have 90 :)
Think about


Im not sure if that would make sense though XD, it would lead to a lot of confusion


Deeelite wrote:
BTW what would you do with people that are new or dont have any kill ratio on a category?


New people = 0 exp as usual



Let me get some numbers and i can show you what come off of that idea
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#11 » Sun 21 Feb, 2021 6:47 am

As it takes some time to run that stuff and i need to make some proper output (to be able to show it)

Here are some changes i did to the idea:
- There are multiple Groups for Vehicles and other stuff

- Those get summarized into the Groups from Deelites idea (basically a Pawnee is a Attack Heli and goes into Deelites Air Group)

- Kills of Vehicles are defined a bit differently (i cant see if a vehicle specifically gets destroyed, but i can see if someone gets killed IN a vehicle) so a vehicle kill gets defined as kill of a player inhabitating a vehicle in any position

-Kills from vehicle are different too, those are defined as kills from someone occupying any relevant slot (driver, gunner, commander), kills by passengers are defined as Infantry kills

-Kills from Static Weapons (AA, AT, HMG, GMG, Mortar) and mines are grouped into Infantry kills

-Kills from Arty Tanks are grouped into vehicles (but have their own category if needed)

-I changed the Vehicle Metric to Kills / Deaths when in Vehicles (as its easier to get)

-I removed Teamkills from the statistic (they are rare, but i can add them if needed)

-Will make the data into a 10EXP scale (for readability) with the sheme:
4/12 Playtime as a logarhitmic scale (with max 200h playtime)
4/12 INF contribution
2/12 Vehicle contribution
2/12 Air contribution

contribution = KD * % of total kills in the category (probably will need some factor to up the number a bit as its pretty tiny, but its not only a good KD like in total 10kills and 3 death since the beginning of time but you also need to do your share of work), will see how that works out in total
Last edited by TheMasterofBlubb on Sun 21 Feb, 2021 7:07 am, edited 1 time in total.
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#12 » Sun 21 Feb, 2021 6:57 am

CRUNCHY TIME
Image
~1800 Sessions (only normal ones currently, as i need to add a third filter to get all sessions)
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#13 » Sun 21 Feb, 2021 7:04 am

Some pre calculation data as it is pretty interesting
Highest Kills as Inf:Darth
Highest Deaths as Inf:Darth
Highest Kills as Vehicle:longrifle
Highest Deaths as Vehicle:fynn
Highest Kills as Air:Darth
Highest Deaths as Air:Darth

And bare with me, this is no joke:
3rd most Kills as infantry is ..... the AI defending points.

The total Kills for this run where 443231 (will change as its a full run, so any session played will up that number), also not all kills are counted towards players as some kills are not properly registered with weapon etc some are undefined so i drop those (somewhere around 150 or so in total)
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#14 » Sun 21 Feb, 2021 7:16 am

I hope that i can make a cool table with the new formula until tomorrow, if you like you can write down some names (pls not so many its quite some manual work involved currently to get the per player data) that you would like to compare
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

User avatar
Deeelite
Donator
Posts: 807

Re: new exp level calculation suggestion

Post#15 » Sun 21 Feb, 2021 9:18 am

Good Master,
I'm rly glad you jumped into this idea with interest and motivation, leading me to think we will probably get through.
I'm also considering that many players want to see their feedback, somehow, after been giving lot of effort. This new exp. calculation will give'em some.
The principle, the way data should be applied, should lead to express the exp. value as a ratio of dangerness or contribution a player can rapresent when in-game. Done this, players will should have an higer consideration of what it means to join a team when exp. sum is unbalanced. They will surely realize soon that an apparently little difference is telling the truth. Too many times we have been discussing about the questionable unbalancing in exp. summation, while the in-game final resul was supporting our doubts. This brought most of us not to consider that value so much.

So, what are the data that more should influence the exp. value when fighting?

1- kill ratio:
Killing a lot without being killed must be more influent that killing a lot but getting killed every time, that's fore sure: you cannot be that effective if you are constantly respawning or patched up.
Lots of kills stop enemy actions but not always are stopping the other side from winning anyway, because the killing team has to stop flags from being captured too. So we have to better merge theese data in order to distinguish 'who kills a lot' from 'who kills a lot and contrasts the captures', if we can.
It's a bit complicated since I can be the one killing a lot from outside and my mates the ones who are contrasting the flags. Also, how to computate a team when it is killing a lot without defending its points loosing anyway?
Somehow, killing a lot is important but not as much as being able to defend flags.

2 - contrasting skill
Due to what is said at point one, we'd better find a method to detect and imply this very important skill (maybe the most one) into the exp. data value. One way is by introducing new scripts to take in account the surviving time after being firing a weapon summed to the quantity of kills acheived after starting the engagement and the distance from an attackable flag where all this occurs, since killing enemies in the middle of nowhere is important but not as much as killing them when close to flags. Example:

Tom shoots: no kills -> a 20 sec countdown timer starts (0 exp pts) -> Tom shoots again and kills: the timer goes back to 20 sec ( + 2 exp pts. 1 for the kill + 1 for surviving and thus for have been resetting the timer) -> Tom shot again within 20 sec and kills again ( 3 exp pts: 2 for resetting the timer twice, 1 for the kill). If more than 20 sec has passed but Tom is still there, the process will start again from the beginning as soon as he fires. All the scored points have to be multiplied to a certain value as high as the victim was close to the flag (1 in the middle of nowhere, 1.2 within 300 mts, 1.3 within 200 mts up to 2 when inside the circle, e.g.) and another multiplier has to valutate if Tom was inside the flag circle or not. The timer is needed to distinguish and better reward a dangerous defender from a normal defender.
Consider that having now an exp. level = 8 must mean a really good value, so I would give more importance to the 'contrasting skill' rather the 'kill ratio'.
Maybe you can find a way to analyze data this way, without scripts, I don't know but the principle above should be respected, somehow.
Last edited by Deeelite on Sun 21 Feb, 2021 2:11 pm, edited 2 times in total.
"When taking part in a discussion, the most difficult thing is not defending your opinion, but knowing what it is in the first place." André Maurois

User avatar
Deeelite
Donator
Posts: 807

Re: new exp level calculation suggestion

Post#16 » Sun 21 Feb, 2021 9:28 am

Oh, I forgot: thank you.
"When taking part in a discussion, the most difficult thing is not defending your opinion, but knowing what it is in the first place." André Maurois

User avatar
Elvis_FIN
Donator
Posts: 1663

Re: new exp level calculation suggestion

Post#17 » Sun 21 Feb, 2021 1:27 pm

Few ideas.

- For playtime use also 'Flag def' and 'Flag off' values. That is imporatan info to tell person has probably been in a right place (defending or attacking a flag)
- For kills can you add more exp if kill was within attack or defence bonus. Then more likely it was a good kill in a right place
- Make TKs as minus exp. Maybe ratio like 1 TK is minus 3 enemy kills (do not count mine TKs)
- Since many times AT/AA guys are shooting vehicles without vehicle kill can you actually summ the money AT role has made with hits/kills for vehicles and that is value for exp calculation

Of course you cannot put everything in formula. Someone can be very useful to team but not gathering so much kills and not being considered that much of a threat. E.g. if pilot like Grebo just flies HB by moving troops, might be quite low exp with these formulas but still very much of a threat to enemy. So I think maybe not the best idea to make exp mostly based just around kills. Also might make some players to start playing more for kills than like driving MHQ.

If exp system is more reliable in future, then introduce new EUTW rule: "Intentional team unbalancing is fordidden" ?

Edit:
- Add MHQ driving time to have weight in formula
- Add EMHQ hit bonus to have extra weight in formula
- Do not add any exp when playing alone on server

User avatar
Deeelite
Donator
Posts: 807

Re: new exp level calculation suggestion

Post#18 » Sun 21 Feb, 2021 2:07 pm

Elvis_FIN wrote:1- Add MHQ driving time to have weight in formula
2- Add EMHQ hit bonus to have extra weight in formula
3- Do not add any exp when playing alone on server


1 - mhq driving is not telling that much about the level of danger that player rappresent, expecially when I see even experienced players driving them for a long time and park them where is not needed or in a bad position, vanishing all the effort and the time spent. I don't want this new calculator to encourage players to drive mhqs just for the reward, unless admins find a way to detect and distinguish a good and lasting placement from a crap action. Remember that sometimes is better not to drive an MHQ at all and there's no need to reward someone who moved it when this was coincideing with the worst action to be done. For this reason I would reserve a very little percentage to this aspect, even tho I know it should deserve much more.

2 - Yes, good point Elvis, not to be forgotten: mip/mhqs destroyers are a pain in the arse.

3 - Yes and no but, ok, I can agree.
"When taking part in a discussion, the most difficult thing is not defending your opinion, but knowing what it is in the first place." André Maurois

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#19 » Sun 21 Feb, 2021 5:02 pm

After a quick read through some feedback on the adfitional ideas:

- "Flag def" and "Flag off" should be possible, but are quite difficult to achieve, right now, so im gonna leave that out (i technically have the data, but currently the collector is based around kill events and i would need to add a quite huge portion to make a position analysis)

- same for mhq driving time

- kills close to points is possible though

-MHQ kills are actually not logged as it seems, so cant track that specifically, i can see when it gets killed though and who is close if that helps any(again would require a position analysis)

- Deelites Timer idea, is most probably not viable, due to vastly different implementation of ingame capture and my calculator.

- sum the money for AT and AA sadly doesnt work, its almost impossible to recreate vehicle hits with no kill, it would need additional loggin in-game

-count TK, sure can do

- do not count when low player count, is implemented, there is a threshold for when a kill event gets analysed in min playercount on both teams
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#20 » Mon 22 Feb, 2021 12:01 am

Some Comparison values:
Formula for EXP: (4* (Playtime EXP) + 4 * (INF KD * Contribution in %) + 2 * (VEH KD * Contribution in %) + 2 * (AIR KD * Contribution in %) ) /12

Playtime EXP = same as currently, but with hard cap at 200h
Contribution = Personal Kills / Total Kills per category

Values Coming Soon (TM) (the previous ones were broken due to excel beeing nasty)
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#21 » Mon 22 Feb, 2021 3:01 am

Some Values:
3,9 Leroy
7,2 Mudokon
3,7 Deeelite
3,5 Elvis
9,9 Darth (Darth is a very active player and has a super high Air KD AND contribution)
3,5 TheMasterofBlubb

Thats the first draft, any feedback is appreciated
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#22 » Mon 22 Feb, 2021 4:04 am

Image
The values for the calculation
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

User avatar
Elvis_FIN
Donator
Posts: 1663

Re: new exp level calculation suggestion

Post#23 » Mon 22 Feb, 2021 4:52 am

I'm actually quite humble person (also quite realistic), but thinking how I play and looking values here seeing myself about one third of "value" compared to Darth is way off. I see myself and Darth quite close to each other as value for a team.

Like K/D ration. You can be a sniper who has good K/D ration, but still you are just sitting outside of the circles and doing not good for capturing or defending. Why that should be better exp when you are actually doing worse?

I think there are too many factors that cannot be added to formula that it will never be very precise. But I like the idea to get exp more precise. But if most of things are based on K/D ration it will never be very precise. At EUTW there are many more factors that can be even more important than just pure K/D ration.

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#24 » Mon 22 Feb, 2021 12:32 pm

True.

Darth is that high due to the contribution factor. I used that to basically get an adjustement of an "active player". And he has by a longshot more INF kills and more AIR kills than ANYONE. Pair that with a 14 KD for AIR it drives his EXP up.

Btw this EXP scale isnt linear, its basically:
<3.3 learning player, needs to learn basics, might get some boost from his general skill
3.3 - ~4 is an average player that contributes his share to the team

>4 player is very good in 1 or more categories and can be a allround asset for the team

The pure numbers can be adjusted for better readability (there currently is no cap on how high the EXP can go)
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

TheMasterofBlubb
PVP Warfare Admin
Posts: 738

Re: new exp level calculation suggestion

Post#25 » Mon 22 Feb, 2021 12:36 pm

I can mangle in some other values too if you like instead of Kills-contribution i could use Kills per hour or so.

When i get some time and some proper idea on how to do it, i will add a weight to kills close to flags for a bonus.(like normal kills is factor 1, in 250m proximity is 1.5 and in 50m (inside circle) is 2).
The vehicle kills have a similar weighting for transport kills as those are rewarding transports more than combat vehicles due to the passenger counts
My Steam Profile


  • Worth: WAAYYY TOO MUCH
  • Games owned: 323
  • Games played: 205 (63%)
  • Hours on record: 7,946.8h

Who is online

Users browsing this forum: No registered users and 12 guests