john.oaklander
New member
Hey,
Could we please be provided with some transparency on how the ranked system works? Its very hard to give feedback on it when its a complete mysterybox.
It seems from some anecdotes that you guys might be implementing some sort of a "true skill" system, where instead of ranking purely by wins/losses you are taking some statistics (like perhaps kills, deaths, assists, damage dealt, healing dealt, structure damage) into account.
This has been tried in many games before, and it has never worked, because it is entirely misguided. The goal of the game is to destroy the enemy patron, its the only objective that truly matters. The goal of the game is not to optimize statistics. However, if you are ranking not only based on win/loss but also statistics you are shifting the objective of the game from winning to a mixture of winning and stat optimization.
Good statistics only matter if they lead you to winning the game, which they generally will on average, and if they do lead to you winning your winrate will be high and your rank will adjust, meaning there is no need to further adjust based on good statistics, and if they do not lead to you winning they do not matter and should not be adjusted for.
It is entirely possible to have consistently good statistics and be a worse or equal player to those you are matched with as statistics do not tell you the whole picture.
One could be very good at playing the laning phase, and perhaps play characters that are good in the early game, and hence get many kills early when its fast to do as respawn times are short and deaths relatively inconsequential. But then they could be bad at playing late game, and get picked late a couple of times when it really matters and singlehandedly lose their team the game, all while having 15-2 stats because they went 15-0 in the early game before losing the game by mispositioning late.
Similarly a person could ignore all objectives and greedily farm while his team keeps him in the game, and lose the game for his team singlehandedly, but simutaneously have few deaths and highest farm in the game.
If a player consistently has good statistics, but is not winning games then he is not a better player than the people at his rank, his playstyle just optimizes stats. Since the only constant in a players teams is himself if he is truly better than the people at his rating his team will statistically have a higher probability of winning, and law of large numbers will ensure that they win more than they lose over a large sample of games. Any given game can go any which way due to randomness, but over a large set of games all randomness will even out, leaving only the impact of the difference between the player and the average player at his/her rank. This is statistics 101 and the basis of all ranking systems.
Please tell me Im wrong on this, and you are aware of the misguidedness of true skill systems and are not implementing such a system. But if this is the case there are some unexplainable anomalies in ranks (such as people that played their games against people that ranked into phantom, lost most of said games but somehow got ranked into eternus) which might indicate that there are unintented bugs in the system.
A chess player that captures alot of pieces because they offer many misadvantegous exchanges, and play into every advantageous sacrifice of the enemy but lose all their games due to the poor board state those captures leave him with is not better than a chess player that captures less pieces but wins sometimes. You can capture more pieces and still lose, because capturing is a means to the goal, not the goal of chess, and isnt always advantageous.
Furthermore, please add drafting to ranked, drafts really matter ALOT and in the current system you can just about automatically lose games based on random drafts.
heres just a couple of examples that point towards the game not functioning on a mmr system, but rather some "true skill" system under the hood, or being completely broken: (you can check the matches from the ID using the watch tab)
1.
https://tracklock.gg/players/158685386
Player plays ranked matches against low ranked people that mostly end up in phantom with a couple of low ascendants. out of the recorded games he loses 4 and wins 3. For his sub 50% winrate against mostly phantom players he was rewarded with eternus 3.
2.
player plays ranked matches against low ranked people that exclusively end up in emissary/archon, wins against them, and ends up in eternus 2. This is pretty nonsensical, the man was never tested against real opposition, he beat people in low rank and got placed in the highest rank for it.
3.
tracklock.gg
Wins against people who are mostly phantom or low ascendant -> ends up in eternus 6.
the 2nd and 3rd examples point towards some true skill system with heavy weight on individual performance, which enables people that are essentially smurfing to get jumped to the highest ranks without ever being tested against real opposition.
the 1st example is hard to explain, the person loses his games against phantoms but gets eternus, and then goes on to play casual matches against oracles and lose. makes 0 sense.
Could we please be provided with some transparency on how the ranked system works? Its very hard to give feedback on it when its a complete mysterybox.
It seems from some anecdotes that you guys might be implementing some sort of a "true skill" system, where instead of ranking purely by wins/losses you are taking some statistics (like perhaps kills, deaths, assists, damage dealt, healing dealt, structure damage) into account.
This has been tried in many games before, and it has never worked, because it is entirely misguided. The goal of the game is to destroy the enemy patron, its the only objective that truly matters. The goal of the game is not to optimize statistics. However, if you are ranking not only based on win/loss but also statistics you are shifting the objective of the game from winning to a mixture of winning and stat optimization.
Good statistics only matter if they lead you to winning the game, which they generally will on average, and if they do lead to you winning your winrate will be high and your rank will adjust, meaning there is no need to further adjust based on good statistics, and if they do not lead to you winning they do not matter and should not be adjusted for.
It is entirely possible to have consistently good statistics and be a worse or equal player to those you are matched with as statistics do not tell you the whole picture.
One could be very good at playing the laning phase, and perhaps play characters that are good in the early game, and hence get many kills early when its fast to do as respawn times are short and deaths relatively inconsequential. But then they could be bad at playing late game, and get picked late a couple of times when it really matters and singlehandedly lose their team the game, all while having 15-2 stats because they went 15-0 in the early game before losing the game by mispositioning late.
Similarly a person could ignore all objectives and greedily farm while his team keeps him in the game, and lose the game for his team singlehandedly, but simutaneously have few deaths and highest farm in the game.
If a player consistently has good statistics, but is not winning games then he is not a better player than the people at his rank, his playstyle just optimizes stats. Since the only constant in a players teams is himself if he is truly better than the people at his rating his team will statistically have a higher probability of winning, and law of large numbers will ensure that they win more than they lose over a large sample of games. Any given game can go any which way due to randomness, but over a large set of games all randomness will even out, leaving only the impact of the difference between the player and the average player at his/her rank. This is statistics 101 and the basis of all ranking systems.
Please tell me Im wrong on this, and you are aware of the misguidedness of true skill systems and are not implementing such a system. But if this is the case there are some unexplainable anomalies in ranks (such as people that played their games against people that ranked into phantom, lost most of said games but somehow got ranked into eternus) which might indicate that there are unintented bugs in the system.
A chess player that captures alot of pieces because they offer many misadvantegous exchanges, and play into every advantageous sacrifice of the enemy but lose all their games due to the poor board state those captures leave him with is not better than a chess player that captures less pieces but wins sometimes. You can capture more pieces and still lose, because capturing is a means to the goal, not the goal of chess, and isnt always advantageous.
Furthermore, please add drafting to ranked, drafts really matter ALOT and in the current system you can just about automatically lose games based on random drafts.
heres just a couple of examples that point towards the game not functioning on a mmr system, but rather some "true skill" system under the hood, or being completely broken: (you can check the matches from the ID using the watch tab)
1.
https://tracklock.gg/players/158685386
Player plays ranked matches against low ranked people that mostly end up in phantom with a couple of low ascendants. out of the recorded games he loses 4 and wins 3. For his sub 50% winrate against mostly phantom players he was rewarded with eternus 3.
2.
3.
craigboone2 - tracklock.gg - Deadlock Stats
craigboone2 player profile for valves new game Deadlock.
the 2nd and 3rd examples point towards some true skill system with heavy weight on individual performance, which enables people that are essentially smurfing to get jumped to the highest ranks without ever being tested against real opposition.
the 1st example is hard to explain, the person loses his games against phantoms but gets eternus, and then goes on to play casual matches against oracles and lose. makes 0 sense.
Believe it or not, there were about 3-4 games where I was supportive in voice chat and somehow we managed to turn the game 180 degrees after my "speeches" xD
