@ PWYFF : It feels aweful when the game's outcome is a crap-shoot because of matchmaking.

Aptest·Sep 3, 2016·20 votes·697 views

This is a rant about premades. Specifically, about how the match maker deals with them.

Sometimes you have premades with large skill difference. for this post a large skill difference means 3 divisions difference or more The matchmaker puts them both in a game. Puts another bunch of premades on the other team.

When the skill difference is significant (2 divisions or more), you'll get "the good lane" where your team's guy is the skilled one. and "the bad lane" where the other team has a substantial skill advantage.

As a solo player you have no control over this. There is going to be "a good lane" and "a bad lane" and you have no influence over which lane is going to be good, which lane is going to be bad, and by how much. the game becomes "their good lane VS yours" and the result is all up to chance and the grace of the matchmaker.

it really feels bad to be on the losing end of this deal. you get to feel like you had no influence over the game like there is no real point in playing.

20 Comments

jaymc1130Sep 3, 20164 votes

Well lets be fair here op. Its DEFINITELY not a crap shoot, its DEFINITELY not pure luck. But no one can deny that luck has been an increased factor on the ladder this season. And yes, you identify the exact reason. The highest skilled player and lowest skilled player having an MMR gap that more than tripled this season across all levels of play is certainly to blame for much of the volatility of the games, that is absolutely true.

But players also can't become so obsessed with the flawed system that it detracts them from thinking critically about the matches they play in. If this season was 100% luck then why do I have no issue regularly one tricking accounts from unranked to diamond as I do every season? It still took me between 150-200 games for most of them, which has been my average pace for several seasons now. So clearly the impact has been felt by the playerbase, but some, imho, CHOOSE to feel it more than others. Yes, I deal with all the same things you deal with, yes, I am JUST as displeased by them (probably more so given just how much I've bitched on this topic over the months), but I don't allow my frustration with an incompetent gaming company and its poorly designed systems to prevent me from achieving what I know I am capable of achieving.

For example, in the scenario you present with the one "bad lane" and one "good lane" there are indeed ways you can use this to your advantage in game. If you can identify which is which soon enough you can make an impact in the "good lane" and then get him to roam with you to the "bad lane" and start a massive snowball that even a bad player will have difficulty giving away. Being able to adapt like this is part of climbing, and its certainly something that is still possible despite the rampant flaws this season's ladder has had.

Now thats no reason to stop demanding improvements! But it is reason to demand them with a grain of salt and to be as detailed and in depth about what you would like as is possible.

AptestSep 3, 20163 votes

My issue is that the proportion of "bad games" has significantly increased.

The number of games where I feel powerless due to allies or opponents "doing their thing" too well is greater.

There are more games lately where I do not get to properly express my skill level. To properly play. To properly be useful or fail.

Yes I can still get to where I deserve, elo wise. The law of large numbers makes it so. But the amount of Crap game i get to experience on the way there is much, much greater.

jaymc1130Sep 3, 20162 votes

{quoted}

My issue is that the proportion of "bad games" has significantly increased.

The number of games where I feel powerless due to allies or opponents "doing their thing" too well is greater.

There are more games lately where I do not get to properly express my skill level. To properly play. To properly be useful or fail.

Yes I can still get to where I deserve, elo wise. The law of large numbers makes it so. But the amount of Crap game i get to experience on the way there is much, much greater.

Well this was always the case. Players have always had to play an extremely large number of games to reach an appropriate skill ranking because the ranking system has always been terribly flawed for this game. You can't take ELO, a systme designed for chess, use it in a team game, and use it to rank the INDIVIDUAL players without EVER at ANY POINT accounting for their individual skill metrics.

That should just be common sense, but hey, Rito is a dumb as fuck company mate. We all suffer the same under the weight of a terribly, horribly, awfully designed system.

If they had a PROPER ranking system for this game I would never need more than 30 games to take an account from unranked to diamond. It gets pretty old going 15-5-15 with an 85% win rate til I hit the Troll City MMR range that is plat5-d5 and start to get trolled every single game and then having to slog through 100 games in plat where I perform at that same level but have to play so many 3v5s or have intentional feeders that I can't maintain a win rate that fairly represents that skill set. I mean, I totally understand where you are coming from on that note and I get that frustration. I've been demanding that be changed for years. But I've gotten no where for 4 and a half season bitching about that topic (even doing the work FOR Riot and INVENTING a system DESIGNED to assist in this called Champion Mastery which was supposed to use the end game performance score to affect MMR gains and losses based on individual performance back in preseason 3 that only wound up half ass implemented several seasons later) and eventually you just learn to abuse the flawed system and use its flaws to your advantage.

AptestSep 4, 20161 votes

Show me a single individual skill metric that is reliable. KDA is not reliable. It is not indicative of skill. It is susceptible to so much "noise" and quasi-random events that it can only be used under the most extreme circumstances as a measure of "skill". You play a tank? KDA not work. You play support? KDA not work. You do zoning instead of killing? KDA not work. You show up late and pick up revenge kills after the other team got a tower and a dragon? KDA not work.

The only metric which is shown to reliably say you have contributed to a win, is you having a win. The problem with "individual skill metrics" is that they are completely unreliable. Yes there is a correlation between KDA and winning. But you do not know that KDA causes winning. _Yo do know that winning causes KDA, _ though. So winning is a reliable metrics. KDA is not.

And the fact that I can get "S" on supports super easy just by gaming the system a little bit is proof that RIOT, at least, does not have a reasonable understanding of individual metrics that allow them to make proper use of those.

ELO based ranking, though a relatively simplistic and primitive method, has been shown to work. To argue that it does not is to outright dismiss years and years of hard evidence in multiple sports. Yes it's designed for 1v1 games. Yes it's been adapted in the most rudimentary and simplistic way. But at the end of the day the method is so beautifully robust that it just works.

and the "designed for chess, not for LOL" argument has, by itself, no meaning. It's just a "no true Scotsman" statement.

jaymc1130Sep 4, 20162 votes

{quoted}

Show me a single individual skill metric that is reliable.

Actually, individual skill metrics are incredibly reliable in terms of indicating a players skill level.

Lets go through some examples, I'll compare some of my accounts to some others and yours.

Lets start with a Bronze 5 caliber player, I'll use known quantity and recent boards celebrity, the legend himself, xxWukong360elite:

Wukong, as you can see from any third party site, is a classic bronze level player. He lacks many basic fundamental skills that are required to attain high level play and this reflects in his ranking. Furthermore you can even tell WHAT skills it is he lacks by looking at some of these individual skill metrics. For example, he manages to average less than 3 cs per minute, regardless of role. This indicates that his basic mechanical skills, in lane, or as a jungler, are very weak in the early portions of the game. He has poor lane phase trading habits and this helps to keep him from being able to farm, and having a low income amount makes it difficult for him to impact games. We can also see that he dies about 11 times per match! That's a lot of deaths. This indicates that his positioning skills and decision making skills are pretty lackluster. He cannot average a 2.5:1 KDA ratio either which further supports his poor decision making, however, he acquires a ton of kills per game, often averaging over 20, and participating in close 30. This suggests that his teamfight abilities are actually fairly decent. His kill participation rate averaging around 55% also indicates that he is a player who works well with his teams as he is often at the right places at the right times to contribute to the team goals when fights break out, although it also supports the idea that he is prone to chasing kills rather than going for objectives. This can be confirmed by his tower take down ratio, in about 330 ranked games this season he has managed only 130 tower kills. Thats a ratio of about 33%. And lastly, but not least, we see that he has a negative win rate of 46%, which you would expect given these other individual skill metrics. In short, this is a classic bronze player with bronze player level skills.

Lets take a look at a player a step up from that, we will use Jikker, a silver player of some repute who is by far one of the most pleasant people I have ever met and a pure joy to play with in game:

Jikker is a pretty standard silver level player, although her individual skill metrics tell the story that she could potentially rank higher if she cleans up a few mistakes that she commonly makes. She is a support main, which we will circle back to later, as this fact has some implications on her ranking. As a support her cs totals aren't inherently important, but we can see from her income totals in other roles that she is indeed a step above our first example. She is regularly capable of managing to average 4-5.5 cs a minute in laning roles, but can only manage about 2 per minute from the jungle. This shows that she is competent at an average level in lane, but has poor pathing in the jungle which leads to an inefficient use of her time in that role. Her lane phase trading habits are about that of an average player, as she manages to not get zoned off too hard and has the game knowledge required to play safely when she has to and simply acquire what income she can, although the metrics show she is not overly talented either and thus her impact on games is less than it could be if she was a superior player in the early stages of the game. We can see that she dies on average about 7 times per match. Thats a pretty average rate, nothing special, neither great nor awful. Her KDA, however, tends to rate in the 3:1 to 4:1 range! This means her decision making abilities are actually fairly strong in addition her team fight acumen. Coupled with her death rate per game we can see that this means she actually has very solid positioning skills. Furthermore, she tends to participate in about 55-60% of her teams kills in games, meaning shes is a very solid team player who is typically always on point, working with her team and coordinating their efforts in fights as they break out. We can see from her objective take down score of 55 towers taken in about 100 total ranked matches this season that she uses her strong teamfight abilities to transition into taking objectives, just not at an extremely high rate (which isn't surprising considering she is a support main, these champs don't have the best tower taking abilites), but a 50% -ish ratio from that role is a strong indicator that she understands the strategic elements of the game well enough to make a consistent impact in games for her teams. Lastly, she has a 48% win rate, which is a little lower than we would expect given what we see from the other individual skill metrics presented to us. Given that the metrics support her being a strong team player with solid awareness and decent game knowledge but average mechanical ability we would expect her to have MORE success in ranked play if she were to transition to a carry role rather than support where she is often hamstrung by her teammates who are probably slightly less capable players than she is as the support role is the role with the second least degree of influence over the outcomes of games. Should she ever decide to put more weight on her shoulders and play a carry role more often, the individual skill metrics support the idea that she could achieve a gold level ranking.

ACK! APtest has next to no games played this season! ZOMG! Hard to analyze your indiviudal skills if I don't have any data mate! Well we will analyze the 5 ranked games you have played to date and see what they say. You seem to manage an average cs rate, about 4-5 per minute, but significantly less than that as leblanc. So your laning skills are probably very average. Your cs rate in the jungle is particularly poor, but you manage a solid kill participation rate there, so while pathing is an issue for you along with efficient use of your time in the JG, you understand how to make an impact in the lanes and help your teammates. You die a little too much at about 8 deaths per game tho, so your decision making and positioning skills could certainly use some work. This is further confirmed by the less than 2:1 kda you manage as an average. Next, your Kill participation rate being about 40-45% means you lack some basic understand of team organization and teamwork. You don't find yourself in the right places at the right times often enough to make a significant impact in game. Learning to communicate and coordinate better with your team will significantly improve your success rate in ranked. And lastly, you manage only 2 towers taken in these 5 games for a 40% tower take down ratio, meaning your objective control and ability to prioritize the correct play in a game is lacking. From these individual metrics we would expect you to be a high bronze or low silver level player.

On to me, we will use oOBestEveNAOo as its the account I've been messing around on most recently, although I'm soon to change over to another smurf to attempt two tricking fizz and diana to diamond. Starting with my CS we can see some significant things, my cs rates in the jungle average better than 5 cs per minute, and often times I average 6 or more. However, my laner metrics show that I can barely manage about the same rate in lane. This means that while my jungling pathing is excellent, I am making extremely efficient use of my time in that role, and because the cs totals are so high it means I am ALSO depriving the ENEMY jungler of resources at a pretty decent clip in game as I steal many camps. When compared with my lackluster laning cs metrics though, one can see that my mechanical abilites in lane simply aren't good enough for diamond level play, I make too many mistakes in lane, trade poorly, and cost myself income. My positioning in lane is also likely subpar for a diamond level player. (Not a surprise here, I noticed this trend several years ago and quickly became aware that being 30 years old with poor reflexes was costing me matches when I was a solo lane main because of my poor early game performances in lane phase.) We can see that I die about 5 times per match, a tad on the high side for a diamond level player. I also manage a 3.5:1 KDA, so while I die a bit much (front line role, I often give myself up for my carries) my teamfight abilites are pretty exceptional and positioning in later phases is solid. Over the course of ranked this season I tend to participate in about 55% of my teams kills, so I am often in the right place at the right time, and I work very well with my team. From an objective standpoint I manage a tower takedown ratio of almost 90%! Thats astounding! Couple this with the average of 3.5 neutral objectives (dragons, baron) taken per match and you can see that I CLEARLY prioritize objectives correctly and coordinate my teams well to take them, leading to a high degree of success. And lastly we can see my 58% win rate is expected based on these individual skill criteria.

Now, these examples are pretty basic, but you can go ahead and look up 100 other accounts and find similarities in performance and success that directly correlates to the same individual skill metric ranges displayed here. When I first began taking this game seriously about 4 years ago I did a lot of research and found that these individual skill metrics were of great value and much better indicators of skill than a person's displayed ranking. Just looking at BestEveNA being D5 doesn't tell you much, but looking at the metrics you can determine I could probably climb even a little higher with that account. Looking at xxwukong360elite's bronze 5 ranking doesn't tell us much, but seeing his individual skill metrics shows us EXACTLY what skills it is that he is lacking that prevent him from climbing higher.

Quite simply, individual skill metrics can tell you everything.

AptestSep 4, 20162 votes

I will write a more detailed response later. however just a small caveat:

if you intend to critique my gameplay, you should be looking at _MY _ gameplay: http://euw.op.gg/summoner/userName=aptest

jaymc1130Sep 4, 20161 votes

{quoted}

if you intend to critique my gameplay, you should be looking at _MY _ gameplay: http://euw.op.gg/summoner/userName=aptest

Agreed. I'm an idiot with 3rd grade reading comprehension level. You may now feel free to berate me mercilessly for my stupidity.

AptestSep 4, 20161 votes

I have already done so, and now that the horse has died beating it is not my style. I prefer to invest the rest of my response to the matter we are discussing.

Feel free to edit your post though i'd definitely appreciate feedback on my playing patterns, since you've already chosen to go through with a metrics analysis,

jaymc1130Sep 4, 20161 votes

{quoted}

I have already done so, and now that the horse has died beating it is not my style. I prefer to invest the rest of my response to the matter we are discussing.

Feel free to edit your post though i'd definitely appreciate feedback on my playing patterns, since you've already chosen to go through with a metrics analysis,

Nah, let everyone else see just how dumb I can be early in the mornings. It's only fair.

Feel free to respond to the concept of individual skill metrics when you would like. Its always fun for me to debate that particular topic.

AptestSep 5, 20161 votes

<> point 1: win % is a useless stat.

thought experiment. I.E. scenario with a much simpler situation that shows how things work in a much more complex environment:

lets assume league players are either Silver or Diamond. There are a lot of silver, but only few diamonds. If everybody gets matched to the same games, the silvers get 50% win rate and the diamonds get 80% win rate.

Example from reality: World of tanks.

This is because Silver VS Silver is 50% but 5 silver VS 4 silver + 1 diamond is win for team with diamond.

If we take diamond players out of general player pool and make game of only diamond VS diamond, than diamond player win rate becomes 50%.

We can generate any kind of win rate for any player, from 0 to 100, by arranging the teams accordingly. Put 4 diamonds + your player VS 5 silvers and have stomp. put him + 4 silvers VS 5 diamonds, stomp other way. Other win rate numbers can be arranged by mixing and matching Silver and Diamonds.

Conclusion: Win rate is only indicative of the relative skill of the player VS the people he has been matched with over the sampling period. Also, matchmaking determines win rate. Because of that, we need to already know how good these players are, in order to make use of win rate as a measuring statistics.

Another thought experiment. Lets say player oOBestEveNAOo makes new OTP account every season and plays it until it gets diamond. gets account through low level, buys Eve and some 450 champs and does placements. Gets put in high silver after placements and plays around 200 games in Silver , gold and Plat until reaches diamond around mid-season.

50 of those games are utter stomps in silver and low Gold, another 50 a decent climb through high gold, and 100 games are a plod through the mirk of plat trolling.

Now oOBestEveNAOo has a lot of wins in his record. He can't reach 60% win rate now but he's got 40 out of 50 wins from silver and 30 out of 50 from gold. so his account says 59%.

Conclusions:

Win rate measures your performance against past opponents. Win rate is only relevant when you already know the level of said past opponents. Win rate is only useful as a measurement of skill when the level of said past opponents is fixed over the sampling period (this does not happen). Win rate is indicative not of current skill level, but of having been miss-matched by the matchmaker and having had to win / lose games to get the matchmaker to work properly.

I.E. you have a positive win %, i know that you have climbed.

jaymc1130Sep 5, 20161 votes

{quoted}

<> point 1: win % is a useless stat.

thought experiment. I.E. scenario with a much simpler situation that shows how things work in a much more complex environment:

lets assume league players are either Silver or Diamond. There are a lot of silver, but only few diamonds. If everybody gets matched to the same games, the silvers get 50% win rate and the diamonds get 80% win rate.

Example from reality: World of tanks.

This is because Silver VS Silver is 50% but 5 silver VS 4 silver + 1 diamond is win for team with diamond.

If we take diamond players out of general player pool and make game of only diamond VS diamond, than diamond player win rate becomes 50%.

We can generate any kind of win rate for any player, from 0 to 100, by arranging the teams accordingly. Put 4 diamonds + your player VS 5 silvers and have stomp. put him + 4 silvers VS 5 diamonds, stomp other way. Other win rate numbers can be arranged by mixing and matching Silver and Diamonds.

Conclusion: Win rate is only indicative of the relative skill of the player VS the people he has been matched with over the sampling period. Also, matchmaking determines win rate. Because of that, we need to already know how good these players are, in order to make use of win rate as a measuring statistics.

This is a good point to make, and it certainly makes it clear that win rate by itself is a poor, EXTREMELY POOR, measurement of skill.

My question to you would be this: Why is our entire ranking system based off of win rate then? That just doesn't make sense does it? You pointed out just how flawed using win rate alone is as the only metric. It should be common sense that the system based on that cannot function properly.

As an aside however, lets focus on the "we need to know how good these players are in order to use win rate as a measuring stick". Precisely. The best use of win rate as a metric is to see what a player is capable of doing versus average level competition (silver range). There are a lot of average players, their performances can be easily quantified, and thus playing against a known quantity at a superior level should be a relatively good indicator of skill. Lets say that every account had to play its 10 placement matches versus silver opponents. And the next 15 games as well. Thats a sample size of 25 games. Lets assume that EVERY individual skill metric in this sample size is measured before actually placing a player into any tier. The performance rate of diamond and up players versus this range of players will be pretty consistent. For example, in that 25 game stretch, for me personally, I always manage a 5:1 or better kda, massive cs per minute totals, participate in close to 30 kills a match, take more than one tower per game, average more than 1 multikill per game, and a win rate of better than 70%. Without fail I accomplish this in that 25 game stretch, everytime. Which is pretty average for a diamond player playing versus silver to gold 1 talent. Using that sample size versus a known quantity is a much more accurate "measuring stick" than using win rate in a vacuum as we currently do. Bronze players will perform very poorly versus that level of competition, silvers will perform at the exact level of the competition, gold will perform slightly better, plat will perform better, diamond will perform much better, and master tier plus will perform at an exceptional level.

You are 100% correct though, win rate in a vacuum, the system we currently use, will never, ever, ever be accurate or fair for ranking individuals on the ladder.

Another thought experiment. Lets say player oOBestEveNAOo makes new OTP account every season and plays it until it gets diamond.

I actually do indeed do this, its a fun challenge for me to one trick new champs each season.

gets account through low level, buys Eve and some 450 champs and does placements. Gets put in high silver after placements and plays around 200 games in Silver , gold and Plat until reaches diamond around mid-season.

Typically I'll place in gold, unless I just get really unlucky in placements. Where you place determines how many games you have to grind to hit diamond, placing in silver as opposed to gold typically adds about 20-30 extra games for me to hit diamond play.

50 of those games are utter stomps in silver and low Gold, another 50 a decent climb through high gold, and 100 games are a plod through the mirk of plat trolling.

Now oOBestEveNAOo has a lot of wins in his record. He can't reach 60% win rate now but he's got 40 out of 50 wins from silver and 30 out of 50 from gold. so his account says 59%.

Conclusions:

Win rate measures your performance against past opponents. Win rate is only relevant when you already know the level of said past opponents. Win rate is only useful as a measurement of skill when the level of said past opponents is fixed over the sampling period (this does not happen). Win rate is indicative not of current skill level, but of having been miss-matched by the matchmaker and having had to win / lose games to get the matchmaker to work properly.

I.E. you have a positive win %, i know that you have climbed.

Again, yes. This is all relevant and for the most part accurate. Which is why it makes absolutely zero sense to use win rate as the SOLE metric being measured for the ladder. You cannot illustrate better just why Riot's current system is garbage.

Win rate ONLY means something when it is taken into account using context from other individual skill metrics. We currently uses a ranking system that ranks players according to the eyeball of the Mona Lisa painting as opposed to using the WHOLE Mona Lisa painting to rate an individual.

We really need to start using the whole big picture, not a tiny pixel that happens to be part of it.

AptestSep 7, 20161 votes

<> Point 1.1 Win rate as a statistic for analisys (what you used) is not the same as win-based skill measuring (what riot uses)

{quoted}

My question to you would be this: Why is our entire ranking system based off of win rate then? That just doesn't make sense does it? You pointed out just how flawed using win rate alone is as the only metric. It should be common sense that the system based on that cannot function properly.

Beause while win rate By itself is very poor, win rate when already know the level of your opponents is very good.

The current system uses a "cheat" to know the level of your opponents before hand and only then uses your win rate VS them to make conclusions about your own level of skill.

Quite obviously this results in some errors. But the method is such that errors become progressively smaller.

There are a lot of average players, their performances can be easily quantified, and thus playing against a known quantity at a superior level should be a relatively good indicator of skill.

Assuming you have a pre-existing knowledge that these players are in fact silver. If you do have this knowledge by some other method than you can use that other method to measure all other players as well and other metrics become redundant, and if you do not than the other metrics help you how ?

You are 100% correct though, win rate in a vacuum, the system we currently use, will never, ever, ever be accurate or fair for ranking individuals on the ladder.

This is your mistake. You believe the system we currently use is win rate in a vacuum. It is not.

Win rate ONLY means something when it is taken into account using context from other individual skill metrics. We currently uses a ranking system that ranks players according to the eyeball of the Mona Lisa painting as opposed to using the WHOLE Mona Lisa painting to rate an individual.

We really need to start using the whole big picture, not a tiny pixel that happens to be part of it.

Reality is the exact opposite of how you describe it. "winning" based skill measuring takes into account all the aspects of one's skill including the unobservant variables. "metrics" based skill measuring takes into account only a small subset of easily measured ideas.

SecondChanceHeroSep 7, 20161 votes

{quoted}

<> Point 1.1 Win rate as a statistic for analisys (what you used) is not the same as win-based skill measuring (what riot uses)

Win rate is a metric that can be inferred from other metrics with a fair degree of reliablity. It's the least important metric to analyze.

Beause while win rate By itself is very poor, win rate when already know the level of your opponents is very good.

Even this isn't very good tbh. Win rate can be influenced by a huge number of factors, many of them outside a players direct ability to control, such as afks or intentional feeders.

The current system uses a "cheat" to know the level of your opponents before hand and only then uses your win rate VS them to make conclusions about your own level of skill.

Quite obviously this results in some errors. But the method is such that errors become progressively smaller.

Actually the opposite is true. The errors become progressively greater over time unless you are skilled enough to regularly overcome games you have an afk or a troll and win them.

Assuming you have a pre-existing knowledge that these players are in fact silver. If you do have this knowledge by some other method than you can use that other method to measure all other players as well and other metrics become redundant, and if you do not than the other metrics help you how ?

This is your mistake. You believe the system we currently use is win rate in a vacuum. It is not.

Actually thats exactly what the system is. Win rate in a vacumm. Thats what all ELO systems are. They were designed this way on purpose for individual on individual competition because every action a player takes in such a game determines the outcome, so the single most important metric is the metric that captures everything you do as a player. It just happens that for single player on single player competition this can be distilled to that one metric.

Reality is the exact opposite of how you describe it. "winning" based skill measuring takes into account all the aspects of one's skill including the unobservant variables. "metrics" based skill measuring takes into account only a small subset of easily measured ideas.

This is only true for single player games. As I state previously. Using ALL skill metrics is the way to go when ranking individuals on a ladder in a team game. Say, for example, we were to base a baseball player's ratings on W/L record alone. Why, then every player on the Kansas City Royals are the 25 best players in all of baseball! They won the most games and the world series! But we KNOW this is not true. So whats a better way to rate baseball players? WAR. Wins above replacement player. This metric takes into account every action every player takes at every point in his career and EVEN goes so far as to include the CIRCUMSTANCES surrounding the actions that took place. Sabermetrics can determine the value of a base hit in the 2nd inning with no one on and no one out in a tied ballgame relative to the value of a base hit in the 8th inning with a runner on 2nd with 2 outs in a game where you trail 4-3.

What league of legends needs in order to rank individuals accurately on the ladder is a similar metric. We need a "WAR" for league players.

The original intention when I came up with Champion Mastery 4 some odd years back was for it to be this kind of system. The thing I proposed took into account not only things like KDA, farm rate, objectives taken, wards placed, kill participation and the like, but ALSO wanted to take into account WHEN and WHERE these things happened in the game and how much the timing and location of actions factored into winning or losing a match. Riot went ahead and used my concept, dumbed it down, ignored the when and where parts (even though that information is already tracked for match history purposed. You can see exactly what happened when and where in the Match History Overview on the web), and implemented a half assed version of what I had proposed.

AptestSep 7, 20161 votes

<> point 1.2: Using the Win rate statistics, in context with other metrics

I've made prior a list of what the win rate statistics can and can not measure. I am making the assumption that you've agreed with me on that list, as you have not disputed it.

The gist of it is this:

Win rate is indicative not of current skill level, but of having been miss-matched by the matchmaker and having had to win / lose games to get the matchmaker to work properly.

how does that settle with "in context" analysis of win rate by you?

...46%, which you would expect given these other individual skill metrics.
...she has a 48% win rate, which is a little lower than we would expect given what we see from the other individual skill metrics presented to us. Given that the metrics support her being a strong team player with solid awareness and decent game knowledge but average mechanical ability we would expect her to have MORE success in ranked play if she were to transition to a carry role rather than support where she is often hamstrung by her teammates who are probably slightly less capable players than she is as the support role is the role with the second least degree of influence over the outcomes of games. Should she ever decide to put more weight on her shoulders and play a carry role more often, the individual skill metrics support the idea that she could achieve a gold level ranking.
...And lastly we can see my 58% win rate is expected

2 of of 3 cases, nothing new is learned. This supports the idea that win rate is useless. Lets examine the second case though.

the case for Jikker is this: if win rate as a statistics has meaning, than it conflicts with the rest of your metrics. The metrics say: this player should have more success. Win rate says: what metrics say not realized.

I havn't seen in your analisys of Jikker's games any support to win rate being worth looking into, in or out of context. I think that whenever you add it to your analytic narrative of a player, you are doing so in error.

SecondChanceHeroSep 7, 20161 votes

{quoted}

Snip

I think I already addressed most of this in the previous response so I'll limit this one to clarification of using win rate for Jikker and others.

The win rate part is actually INFERRED from the other statistics we can measure. You can draw direct correlations between how players perform and what their win rate is likely to be based on those individual skill metrics. This isn't to say that the inference is perfect by any means, there are factors that aren't easily quantifiable that affect winning or losing such as how well you communicate with your team to create and execute a game plan. In Jikker's case specifically we can see from the host of other metrics that this is actually one of the areas she's weaker in. If she were better at organizing her teams then the other individual skill metrics support the idea that her win rate, and thus her ranking, could improve some.

Again, its a matter of taking into account EVERY piece of data available to us in order to find an accurate ranking for players as a "ranking" by definition infers a player has a particular skill set. The unfortunate thing about the systems we use is that many skill sets are not fairly represented in a players ranking. For example, in my case, my mechanical abilities as a player are VERY poorly represented in my diamond+ ranking. My ability to communicate with and organize a team is so exceptional that it overshadows my general mechanical ineptitude. If we were to include a better representation of those metrics I would probably rank lower than I do currently and I feel it would be a more fair and accurate representation of my skill set.

Mojihito666Sep 3, 20162 votes

You are vocal minority.

jaymc1130Sep 3, 20164 votes

{quoted}

You are vocal minority.

Trolls are always safe to ignore. They post the most inane nonsense.

Congratulations on identifying yourself for us and thanks for saving many others the trouble.

MelinethSep 3, 20161 votes

I just had a ranked game with two fresh level 30s who were duoing together on my team. Neither had a full rune page, needless to say, my team got crushed.