Change to the PSR system

[KGB_Killer]

This is just my idea of how the formula’s might work. (don’t bash on the exact numbers because they can be altered slightly. Although please do keep in mind i have tested them somewhat with what my friends psr's would be under this system based on there current stats and the assumption the games they played were evenly balanced team wise.

The object of this is to create an accurate PSR based not only on wins but also Kills, Deaths, Assists, Creep Kills, and Creep Denies.

The formula uses these 5 variables along with the percentage each side has to win at the start of the game based on current PSR, as well as the length of the game in question.

Win: (20 x % to lose) + (0.5 x kills) – (0.5 x deaths) + (0.5 x assist) + (CK / game length x 2) + (CD / game length x 8)

Players Game line:

Wins the game (%50 chance to win based on starting game psr)
Kills 8, Deaths 4, Assists 10
CK 120, CD 20
Game length 40 min.

10(win) +4(kills) -2(deaths) +5(assists) +6(creep kills) +4(creep denies) = 27 points for the game.




Loss: (70 x % win)+ (0.5 x kills) – (0.5 x deaths) + (0.5 x assist) + (CK / game length x 2) + (CD / game length x 8)

Loses the game (%50 chance to win based on starting game psr)
Kills 8, Deaths 4, Assists 10
CK 120, CD 20
Game length 40 min.

-35(loss) +4(kills) -2(deaths) +5(assists) +6(creep kills) +4(creep denies) = -18 points for the game.


As you can see the first person gained 27 points for a fairly impressive game line, and winning the match, however the second person lost 18 points for having the same line on the losing team.

Those of you about to argue that the system will allow for people to slowly gain PSR because the person received more points for winning then the second person received for losing are mistaking.

The reason the second person only lost 18 is because despite the loss, that person still played quite well. Someone with a KD of lets say 1 and 10 would lose a lot more points.

Furthermore this PSR system will only recognize the last 50 games each person has played.

[KGB_Killer]

Please post comments on whether you think this idea is an improvement on the current system, and why or why not.

[Slayn]

i already made this suggestion, accidently in the wrong section:
http://forums.heroesofnewerth.com/showthread.php?t=7379

[KGB_Killer]

Your suggestion is a more balanced approach, where as mine has a huge percentage put towards winning the game still. It is almost impossible for someone to gain points in my system if you lose the match. It simply minimizes the damage. Also mine is a tested formula not just random percentages placed that seem balanced. (i tested my formula for me and 2 other friends over a 10 game span and got fairly balanced PSR results.)

[Slayn]

Your suggestion is a more balanced approach, where as mine has a huge percentage put towards winning the game still. It is almost impossible for someone to gain points in my system if you lose the match. It simply minimizes the damage. Also mine is a tested formula not just random percentages placed that seem balanced. (i tested my formula for me and 2 other friends over a 10 game span and got fairly balanced PSR results.)

the percentages are just randomed examples xD
as i said
the idea itself is the same ^^

anyway nice work with the calculatings

[uberdabing]

Complaint:
PSR should use K/D/A or any other in game stats in its calculations or the “I was owning buy my team sucks” argument
Answer:
This encourages play that is detrimental to the team. Including:
1)Encouraging AFKing in the fountain to preserve record
2)Encouraging killstealing
3)Discouraging playing support and doing support actions (buying wards, etc)
4)Discouraging making sacrifices for the team (Vengeful Spirit swap, etc)
5)Encouraging stat farming instead of winning the game

Overall, it encourages playstyles that are generally accepted as bad.

And secondly, how can a PSR only recognize your last 50 games? Your past history has no impact on you, cause PSR is just a +/- and not an average. If you lose 10000 games, your rating can only drop so low before it stops dropping completely. Then you just need to win like 10 games and you'll be back to normal. K/D averages are affected by your past, but not PSR.

[Slayn]

Complaint:
PSR should use K/D/A or any other in game stats in its calculations or the “I was owning buy my team sucks” argument
Answer:
This encourages play that is detrimental to the team. Including:
1)Encouraging AFKing in the fountain to preserve record
2)Encouraging killstealing
3)Discouraging playing support and doing support actions (buying wards, etc)
4)Discouraging making sacrifices for the team (Vengeful Spirit swap, etc)
5)Encouraging stat farming instead of winning the game


the winning/losing thing will still be the biggest part, as i said
1) afk = kick
3) + 4) this would lead to a lose if you think ppl wouldnt buy wards or so --> bad psr, they would just getting selfowned if they change their behaviour in supporting actions
5) losing the game would result still a bad psr, or at least still a malus


And secondly, how can a PSR only recognize your last 50 games? Your past history has no impact on you, cause PSR is just a +/- and not an average. If you lose 10000 games, your rating can only drop so low before it stops dropping completely. Then you just need to win like 10 games and you'll be back to normal. K/D averages are affected by your past, but not PSR.maybe he thought about recording the kills/deaths/etc of the last 50 games, not the whole kills etc

[KGB_Killer]

The purpose of it only including the previous 50 games is to understand that people improve, a persons first 10 games where they played terribly doesnt trully represent a persons skill lvl.

Another reason for only using the past 50 games is to remove the ability of someone who plays 5x as many games as another person to recieve a higher PSR score. Eg. someone who wins roughly 85% will have a lower PSR (assuming the games are against similar competition) if someone who wins 70% of there games plays 3x as many games, perhaps not a fair representation. (just because someone plays more doesnt make them better).

And as for your argument that it encourages poor teamplay as I mentioned above it is almost impossible to recieve a positive PSR in a game you lose. Also Ksing would have no benefit as assist give just as much of a boost to PSR as the kill itself.

The system is balanced towards support and carries as i mentioned. and this is why.

Support players gain extra points in game based on the high number of kills they are involved in. (Generally carries will have about 15 kills and 8 assist (just roughly) thus being in on about 23 kills.

where as support players may only get 5 kills but they generally assist on about 25. Thus being in on 30 of the teams kills. This is do to the fact that support players have aoe spells which hit all heroes in team fights and based on the fact that if they do ward then they will always be involved in ganks while the carry is farming.

The carry makes up for the reduced points by gaining more points then support players on the creep kill scoring system.

As i mentioned the system is very balanced towards all play styles and simply makes a measurement based on level of play the player maintained through the game.

[zzSleeper]

Including anything other than wins and losses as the primary inputs for a ranking system only decreases its accuracy in a game as complicated as HoN. Moreover, it encourages people to play the system, rather than play the game.

Edit:
After actually reading your proposal (rather than skimming), it appears that my initial concerns were far too generous. I cannot help but get the feeling you don't fully understand how Elo works and what makes it a reasonable rating system. While I agree with your sentiments (insofar as it would be nice if players' skill could be measured more accurately), your suggestion is completely unfeasible. Explaining why exactly requires significant math, so I will instead refer you again to the Wikipedia article:
http://en.wikipedia.org/wiki/Elo_rating_system

[KGB_Killer]

Once again someone not understanding the numbers, wins and losses are by far the primary inputs for the ranking system they account for around 50% of the PSR for the game. The other stuff is just to tweek the amount each person on the lossing team loses and how much each person on the winning team gains.

[Sov4l]

Once again someone not understanding the numbers, wins and losses are by far the primary inputs for the ranking system they account for around 50% of the PSR for the game. The other stuff is just to tweek the amount each person on the lossing team loses and how much each person on the winning team gains.

You still are not addressing the issue of "it encourages people to play the system, rather than play the game."

Example: Your team is losing, but not extremely badly. One of your teammates is a pessimistic stat-whore. Your suggested PSR system can further encourage this teammates to AFK at the fountain to prevent damaging his precious record. In DotA (with no stat-tracking at all), this guy would likely be still playing the game, or have left the game.

Tweaking the PSR according to creep kills/creep denies is okay. However, I disagree with any adjusting of PSR according K-D ratio. It's just detrimental in a team game.

[KGB_Killer]

If the person stays in the base for 10 minutes it will largely effect his creep kills and and thus his overall score. Eg. he had 120 creeps over the first 30 minutes if he kept that pace for 10 more minutes he would have 8 points under his creep score, if however he decided to afk his creep score would drop to just 6 points.

Therefore unless he would die 4 + times in that 10 minute span he would still be better off to go out and try to help the team.

This doesnt even take into account the fact that he also loses points from his denies.

And also if the person does stay afk and gets kick he will lose a lot more points for a disconnect then if he just plays the game out.

I did not put in my origonal post but any disconnect would result in an automatic loss of 40 points for the game.

[KGB_Killer]

Bottom line is that playing the system is playing the game to win, right up until the end.

[Venerable]

sounds like an interesting idea, however wouldn't support heros get screwed? and we all know its many times that the support allows the carry to carry via chick/wards harrass ect. other than seems ok

[Slayn]

to sum it up:
this system just decreases your PSR malus while losing when you still fight, defending your base, killing enemies, denying and so on.
when you're afk for like the last 10 mins of losing you get for example a malus of 35, but when you were fighting til the end it's like 30 (for example)
just a few points but still a difference

supporter heroes get assist points, so they wont get screwed at all

[MeVe]

I agree that PSR systeam must look at K/D/A, ever based on win/lose and the must value the current PSR ( lower PSR = + gain, higer = lower gain )
assistpoint cant be like Kill point, i think 0.2 will be fine because not only 1 take the assist
I didn't see many ks, actualy all want to kill everyone, so the killer is random, the important is that all try to kill

[KGB_Killer]

The purpose of the assist being worth the same as kills is to avoid ksing which u mentioned but its main purpose is to balance the number of points support heroes will recieve in comparison to carry heroes.

By making assist worth just as much it will encourge people to play differnt roles instead of playing a carry every game.

[CloudKen]

i don't think they have this feature added, which would be a great one as my PSR is fairly low now....
i have a even score (204/209/270) with avg game length of 45 mins and 130 creep kills and 6 denies.

i think that when the person leaves a game, the PSR calculations should change as well.
i've had too many games where one or 2 people left and my PSR still dropped the same, its a bit unfair.
and conceding makes ur profile look bad, so i don't think doing that every time there is a leaver.

[Slayn]

i think that when the person leaves a game, the PSR calculations should change as well.
i've had too many games where one or 2 people left and my PSR still dropped the same, its a bit unfair.
and conceding makes ur profile look bad, so i don't think doing that every time there is a leaver.

yepp

[ma5]

I like my system better, +1 point if you go positive in K/D and +1 point if you go positive in A/D. :)

[KGB_Killer]

I would like this forum to be closed and everyone interested in a new system look at my revised idea which is based on a study of 25 players. I will post a link once i post the exact formula.

[Gutless]

I believe K/D/A need to be associated in some way, but I also see were people say that it will encourage solo play in a team game.