About ranking & and stats reset.

[BaKiBaKiBaKi]

I dont like reseting the stats but i can live with that , but the only problem is the system of ranking, i mean next time before you reset the stats please try to find a new way of ranking the players, my advice is somehow to rank players not just by the PSR but also by the kills/deaths/assists, maybe insteed of PSR we should have LEVEL, just saying because many many noobs gain high PSR and ruin high psr games, there is no balance between the skills of the players only PSR balance. Have a nice day admins.

[CoMMoN1337]

ranking people by kills/deaths/assists would not solve the problem, because there are heroes that are better at having good kills/deaths/assists scores than other heroes, and it does not necessarily mean people with more kills less deaths and more assists played that particular game better, an offlane clockwerk for example could easily have a score like 0/6/32 while a carry phantom lancer can have 20/6/20, but there's no way to tell from those stats alone who played better, the clock could've been the only reason the PL won the game, and the PL could've been the only reason the clock won the game, it means nothing, and can be easily abused to gain more if u play heroes that can get kills/assists easily like spectre zeus pudge for example, it does not necessarily mean they played better so it is a bad idea

the psr system as it is, although ridicolously bad, it is the best possible system under these conditions, the only gains should be from winning, everything you do in game works towards that, there will always be people that do more and people that do less, there's nothing you can do about that
some other ideas to improve the ranking system would be to give an exact number of psr for win and exact number of psr for loss, however that is a very bad idea because here you can choose who you play with and against
another idea would be to give +1 psr for every consecutive victory and -1 psr for every consecutive loss, however that is still a very bad idea because here you can choose who you play with and against

there is just nothing more to do to improve this system, even though it's horrible.
the only thing i can think of at the moment that would improve it, is making people win more if they are at a number disadvantage (if they have a leaver in their team) and they manage to win the game, and making people lose less if they are at a number disadvantage (if they have a leaver in their team) and they lose the game

aside from that, there's actually not a whole lot we can do about it you guys

[Lagi]

There would be some methods, but they would be hard to make working. The best would be a ranking system, in a ranking system
What I mean after picks happened, everybody writes a small command according to his role/subrole, or for now, lets say farm priority position. So picks finished, game start, guy with a cm types !pos5, or mid SF types !mid. (examples) At the end of the game, the bot would make a score according to the role of the player. For example cm should get  let's say 10 for each ward planted/dust bought camps stacked etc (about the problems i will write a bit later) while a mid player for example will get his points for lasthits, kda twr dmg hero dmg etc. As a final result they will get a number that describe their performance in their picked role. If this index number <100 minor increase/decrease on the psr, 100-200 normal 200-300 high 300< very high, and they will get +5 +10 +15+20 psr if they win, or loose -5 -10 -15 or -20 if they loose, according to their ingame performance. This way, if you played shit, but your team won, you won't get that much psr as those who actually did well, and vice-versa.

But smth like this can't be implemented most likely because of:
1: the engine limitations. There are so many things that a hostbot can't track ingame, and i dunno if it is possible to hack those functions into the map or not (i ghuess not).
2: it would need a very precise formula to determine the performance of each role as well as possible, since there are things that a formula still won't be able to measure.
3: it would make so much calculation that can consume much more resources than before (maybe nothing hard but, maybe server would became more instable lags etc, dunno)
4: and finally the most crucial part -> The more complex a system is, the more bug/possibilities to ABUSE it has.

And maybe a bit simplier solution would be smth like they made in dota 2 and so many other mobas, an automated matchmaking system.
Here on lagabuse it could work with a matchmaking bot. Someone wants to play go on channel or /w lagabuse.com !search -> bot puts him into a pool, and automatically pairs him up with the same psr guys/with a balanced pool, than creates the game with a generated password, and whispers the gamename/password to every player assigned to that game, they join, autostart if game full.
Problems here are again system abuse, lots of coding, lots of possible resource consume, and later on the long queue times on higher psr, but well, still better than waiting in a lobby, rehosting from time to time and other shits.

Whatever, these are some old ideas of mine maybe someone finds it intersting.

[expert.s]

The problem with this rank reset is that I cannot play now 1700+ PSR games and I have to play whose freaking noob games to get higher PSR for a game that it is worth playing..


I just played one game 1500+ PSR.. and I quited after 15min because of 2 crazy noobs in my team.. This is not worth of my time to play such games. Trust me with PSR reset you are losing players in DOTA (this is why people join garena, LOL, Dota2..)

Greets CECAAAAAA

[CoMMoN1337]

Stop blaming retards for your losses, blame yourself for not beating the enemy retards hard enough to win.

[jeandarc]

Commy is that why u were whining the entire game when i was feeding with furi? Why didnt u win the games solo? EleGiggle KappaPride BlibleThumb MingLee

[Nexxus]

Commy is that why u were whining the entire game when i was feeding with furi? Why didnt u win the games solo? EleGiggle KappaPride BlibleThumb MingLee

gtfo jean, no one likes you here anymore, you arent a attention star whore anymore

[Povratnik]

Commy is that why u were whining the entire game when i was feeding with furi? Why didnt u win the games solo? EleGiggle KappaPride BlibleThumb MingLee

gtfo jean, no one likes you here anymore, you arent a attention star whore anymore

But you are 

[BaKiBaKiBaKi]

okey i get it , i understand you, the idea to get more psr if one of your teammates quits the game or getting less psr if u win a game against 4 players is great. But still this system is a problem and i see you all know that, maybe a little bit of changes would be great, because i dont want to play with players who have 6 kills 10 deaths and have 1750++ psr, when we play high psr game we expect everyone from the team to do their job as a team players, thats all. Bye bye.

[beastiary]

okey i get it , i understand you, the idea to get more psr if one of your teammates quits the game or getting less psr if u win a game against 4 players is great. But still this system is a problem and i see you all know that, maybe a little bit of changes would be great, because i dont want to play with players who have 6 kills 10 deaths and have 1750++ psr, when we play high psr game we expect everyone from the team to do their job as a team players, thats all. Bye bye.

can you expand this idea please i would really like to read more

[CoMMoN1337]

I make money make money make money make money
I make money make money make money make money
I make money make money make money make money
I make money make money make money make money

[bassonja]

I want my stats back, I will suck your penises for 100 psr! I went 1400 after stats reset hahaha.

I freekin had over 2100++ 2 years back, why you reset stats god damn.

It breaks community If you havent noticed.

[div.ide]

Oh wow, I wrote a random fucking essay on hardly relevant topic. Someone give me an internet cookie. Whatever, maybe someone will have any use for it. It's probably half wrong anyway, I need sleep.

Spoiler for Hiden:

Why is it so hard to understand that a player involved in a game of DotA has just one, sole objective?

And guess what? It isn't killing enemy heroes, or creeps, or roshan, it's not important how many deaths you got, neither is accumulating more rapiers or having more buyback gold left. This is not Counter Strike, or Street Fighter, not even Mario Kart.

The objective is destroying enemy main structure (call it World Tree, Frozen Throne, or Ancient, I don't care), which translates directly to either a win or a loss, and as such, w/l ratio is the only really objective measurement of your performance in game. Incidentally, ELO system originally designed for chess, and adapted here under the name of PSR, is based exactly on that. Guess what? Other platforms (or games) pretty much use this too, even if it might be called or scaled differently.

All of the other stats are at best contextual, of situational importance, and of varying weight depending on rather volatile background. And don't even get me started on concept of roles in game or even individual hero picks, which have quite extreme impact on that aspect too. For that to be usable in any comparison, you'd have to do it in very narrow scope of similar cases (i.e. not applicable for an attempt of general skill assessment). If you insist to compare apples to cats, well, don't be surprised if someone calls you an idiot.

The system as it stands, seems to have three main significant weaknesses, to me at least.

Firstly, it's a team game, where each player contributes individually, but you either win or lose as a team. Unfortunately, it's rather difficult to measure each player's individual input. It's hell of a complex thing that is at best an approximation based on bajillion of little details GHost couldn't quite capture (that goes way beyond just raw stats - how do you measure positioning or ward coverage, for example?), and I don't see it captured by a simple algorithm. What that unfortunately means, is your rating is largely rating of your teammates. Sucks? Yes. Can we do much about it? Not really. In longer run, you'll  get as many good as bad teammates, so if you're above your current skill bracket, you'll eventually get out of it. I suppose we could use two-layered rating - one number for general, more slowly changing rating, and another for quickly changing rating reflecting just your trend in last X (some arbitrary number) of games? Possibly what Dota 2 seems to do under the hood, but I'm not even sure how to use it here properly.

Secondly, while assigning each player a rating to make a possibly balanced match, it ultimately doesn't enforce anything: it's in game owner's hands who can join the game, and how are players split into teams. That alone is almost a guarantee your match is not going to be balanced no matter how accurate player ratings may be. Can we do anything about it? I guess automated matchmaker queues could eliminate human factor, but it's not a trivial thing to implement, would require massive amounts of extra work, and without sufficiently high amount players in queue, there's absolutely no benefit from it.

Thirdly, there's just too few players around. Without large enough player pool, it's impossible to make a fair match, simple as that. I don't think you can help that though.

TL;DR rating players based on anything other than w/l is never going to work (unless you try to glorify dagon Furion as epitome of skill); ratings are never going to be perfectly accurate because it's a team game and measuring individual game contribution in game this complex is difficult; it gets further distorted by leaving matchmaking to players, plus playerbase is too small to guarantee a fair match. I have no useful suggestions of improvement that I could see being applicable here.

[cacomonster]

Make it so.. first 5-10 games after reset. . u get x10 psr if u win .. and x1 if u lose.. this way.. decent players / decent teams can make it up fast enough .. and avoid the stupid crap that happens on the 1500~ area that happens with each reset.. This way.. u get an idea of better / good avg players who can win 6-7 of the 10.. and good / nice stackers who go 10/10 .. and if u suck .. u cant stay up there.. unless u keep stacking.. which doesnt happen .. really.  and thoses who are extremely horrible / noobish / retarded .. wont go above 2-3 wins at most.. ( if their team's carry em . )

[beastiary]

Make it so.. first 5-10 games after reset. . u get x10 psr if u win .. and x1 if u lose.. this way.. decent players / decent teams can make it up fast enough .. and avoid the stupid crap that happens on the 1500~ area that happens with each reset.. This way.. u get an idea of better / good avg players who can win 6-7 of the 10.. and good / nice stackers who go 10/10 .. and if u suck .. u cant stay up there.. unless u keep stacking.. which doesnt happen .. really.  and thoses who are extremely horrible / noobish / retarded .. wont go above 2-3 wins at most.. ( if their team's carry em . )

lol