i think its time to rebuild Keystones from the ground up.

agbudar·Jan 20, 2017·55 votes·2,937 views

there are too many offending keystones that warp the game in a very bad way. thunderlords makes burst champions overburst into obscenity. fervor makes atack speed champions complete cancer once they get warlords bloodlust makes any and all weak adc laning phase obsolete.

CotC is a MANDATORY pick right now with the absurd amount of burst mages and assasins have when played right but gets abused to every and all diver champs with a 0.5 sec stun in their kit. (for the love of god make it bonus health not max health)

and since when is it OK to burst tanks? that was suposed to be their strong point being able to survive burst but not dps. yet most genuine tanks get blown to bits

with the exception of who also happens to deal maxhp% damage (yay) is considered a tank at this point with how he builds even though he is a juggernaut.

are now super mobile juggernauts because of CotC

67 Comments

EndlessSorcererJan 20, 201727 votes

Rework masteries as a whole.

Having two offensive trees and a sub-par defensive/utility tree is rather poor design.

I miss the days where 3% damage amplification was the pinnacle of the offensive tree.

agbudarJan 20, 20172 votes

yea thats probably also a good idea the amount of damage multipliers that you can get vs defensive multipliers is stunningly lopsided towards damage.

OnceNFuturePonyJan 20, 20171 votes

Helps to balance out how completely broken tanks are. ohh wait no it doesn't.

EndlessSorcererJan 20, 20177 votes

That's more of an issue with how quickly games end. Games are often decided long before ADCs and other late-game carries can actually get their items.

Basically, tanks are strong in the early/mid-game and many games never get beyond that point

If the damage amplification in masteries were lower, squishy carries may actually be able to reach their late-game (since games wouldn't snowball quite as hard) and not be instantly killed in teamfights.

LankPantsJan 21, 20174 votes

Basically, tanks are strong in the early/mid-game and many games never get beyond that point

Just to add to this the tanks in the game are actually lategame scaling tanks (Sej/Amumu/Shen/Malphite) are currently pretty god damn terrible.

It's not like every tank is overpowered, it's mostly just a few top lane tanks who have insane durability steroids.

ChaoticPineconeJan 21, 20173 votes

Tanks dont scale into "lategame". Lets be real - they are mid/mid-late game champions before ADCs each 6 items and shred them. Always have been.

But tanks who rely on their ult and teamfight seem to be just fine. Champion.gg lists Sej at a 50.8% winrate, Amumu at 51.8%, and Malphite at 53%. They dont seem strong because they cant take over a game like Vi can, but they seem to be just fine.

LankPantsJan 21, 20176 votes

But tanks who rely on their ult and teamfight seem to be just fine. Champion.gg lists Sej at a 50.8% winrate, Amumu at 51.8%, and Malphite at 53%. They dont seem strong because they cant take over a game like Vi can, but they seem to be just fine.

Champ.gg's statistics are basically a crock of shit. They don't check out with other statistic sites, their sample size is abysmal and Riot has even gone as far as to confirm that the statics they list are inaccurate. This is why you should always cross check results.

What do reputable statistics sites say?

op.gg, which Riot suggests using has the following numbers Sejuani: NA:49.5% (outlier) EUW:46.1% KR:47.98% Amumu: NA:46.82% EUW:46.25% KR:45.49%

League of Graphs Sejuani:47.1% Amumu:49.8%

Lolalytics Sejuani:48.68% Amumu:50.17%

For Sejuani the only site that reports her above 50% is the notoriously inaccurate champ.gg, the only other site that comes anywhere near a 50% winrate is op.gg for NA specifically and this result can be explained by poor sample size. There are less than 1.5k games of Sejuani available for the region, that's their entire sample of plat+ games. The two larger regions have better sample size and are more reflective of what's actually happening. her average disregarding the obvious outlier is around 46.5%, even with the outlier it's only about 47%.

For Amumu it's harder to say. Riot has cited op.gg as reliable but it does not agree with what other sites say. I'm willing to bet he's somewhere between the 50% given by Lolayltics and LoG and the 46% given by op.gg, we'll call the average somewhere around 48%.

These numbers are terrible. Amumu has dropped around 5% from the 53% he rested at last season, he was balanced at that number. Sejuani has dropped from around 52% all the way down to 46.5%, that's a 5.5% drop. Both of these are giant drops in winrate and show a huge loss in power as well as two very underpowered champions.

Tanks dont scale into "lategame". Lets be real - they are mid/mid-late game champions before ADCs each 6 items and shred them. Always have been.

http://www.leagueofgraphs.com/champions/stats/sejuani http://www.leagueofgraphs.com/champions/stats/amumu http://www.leagueofgraphs.com/champions/stats/vayne

Compare the winrate/game time graphs.

Sej and Amumu especially are late game scaling tanks. They have methods on their kits to shut down DPS for a long time and have enough damage to kill a carry. They can almost kill an ADC within the lengths of their CC. In the late game Sej and Amumu basically get to have the best of both worlds from Bruisers and Tanks, they have the CC and Durability of a tank and the damage of a fully fledged Bruiser.

ChaoticPineconeJan 21, 20171 votes

{quoted}

First of all, all the sites have some deviance. I dont know how you can call the site I like to use "crock full of shit" when every other site cant get a unified answer either.

Until Riot actually posts their own numbers or openly comes out with a message saying that champion.gg is wrong, I'm going to say that you didnt show anything. All you did was post some Rioter's tweet. He showed a little chart, but I want to know if he took the data within the same parameters as champion.gg does - it looks like a standard global winrate graph by patch.

And a Rioter's personal comments on his personal Twitter is different from Riot itself.

For Sejuani the only site that reports her above 50% is the notoriously inaccurate champ.gg, the only other site that comes anywhere near a 50% winrate is op.gg for NA specifically and this result can be explained by poor sample size. There are less than 1.5k games of Sejuani available for the region, that's their entire sample of plat+ games. The two larger regions have better sample size and are more reflective of what's actually happening. her average disregarding the obvious outlier is around 46.5%, even with the outlier it's only about 47%.

We have major deviance between all the sites. LeagueOfGraphs put her at 47.1% at plat+, with a global winrate of 51%. Lolaytics states "Sejuani Jungle has a win rate 49.04% for Platinum+ players" but has a global winrate of 48%. Op.GG is too much work since they dont give any global stats - if you want to calculate her global winrate, then please do.

I'm not sure why you think NA has "poor sample size". Her pickrate is at a 1.65% on NA, which is slightly lower than EU (1.7%) and a bit higher than KR (1.11%). In terms of number of picks, it goes EU>NA=KR. However, the accuracy of sample size is tied with population - meaning that pickrate is the best indicator for a more accurate sample size. If anything KR is the one with the poor sample.

It seems as if you are just dismissing anything that you dont agree with. You state that champion.gg has poor accuracy, but the sites you listed have just as much deviance from each other. You say NA has a poor sample size and decide to omit the "outlier" in your winrate analysis, but her Pickrate is 33% higher than in Korea. Then you end by saying that Sej has a 46.5% winrate, which makes no sense - youre the one who linked to LeagueOfGraphs and Lolayics, which both state that her winrate is significantly higher.

(I wont go through Amumu since your analysis has the same problems).

Edit - so as a conclusion, I'm going to say that none of the sites are perfectly accurate. If we take out the outliers (LeagueofGraphs and champion.gg) or average out all the data we have, she still sits around 48-9% winrate.

LankPantsJan 22, 20173 votes

Until Riot actually posts their own numbers or openly comes out with a message saying that champion.gg is wrong, I'm going to say that you didnt show anything.

Riot doesn't do things like this, a single Rioter is the best you'll get.

First of all, all the sites have some deviance. I dont know how you can call the site I like to use "crock full of shit" when every other site cant get a unified answer either.

Welcome to statistics. Nothing is ever a unified answer because samples inherently vary from each other. If you want to understand why champ.gg data should be looked at with scepticism I'll do a simple calculation to show the variance of champ.gg as well as the other sites from the mean.

I got the number for thee op.gg average multiplication by finding the total number of games per region and dividing that by total games, this is a proportional average which gives more value to larger sample sizes, the games are valued evenly rater than the regions, a win from NA is not worth extra because there are fewer games.

op.gg average: 49.58x0.235+48.01x0.298+45.84*0.467=47.37

Average of all sites=(47.37+47.1+48.71+50.83)/4=48.50

Absolute Variance from 4 site mean

LoG=48.50-47.1=1.4
op.gg=48.50-47.37=1.13
lolalytics=48.71-48.50=0.21
champ.gg=50.83-48.50=2.33
sum of absolute variance=1.4+1.13+0.21+2.33=5.07
average variance=5.07/4=1.27

You can already see from this that champ.gg varies a lot more than the other sites, but there's one more thing you can do to show how far of an outlier it is, you take the sum of absolute variance and compare it to the average variance on 3 sites.

Average of 3 sites=(47.37+47.1+48.71)/3=47.73

Absolute Variance from 3 site mean

LoG=47.73-47.1=0.63
op.gg=47.73-47.37=0.36
lolalytics=48.71-47.73=0.98
sum of absolute variance=1.97
average variance=1.97/3=0.66

As you can see there's far less variance between these 3 sites than there is if you add in champ.gg, this is typical of outliers.

I'm not sure why you think NA has "poor sample size". Her pickrate is at a 1.65% on NA, which is slightly lower than EU (1.7%) and a bit higher than KR (1.11%). In terms of number of picks, it goes EU>NA=KR. However, the accuracy of sample size is tied with population - meaning that pickrate is the best indicator for a more accurate sample size. If anything KR is the one with the poor sample.

That's pickrate, not sample size. The KR server has more players than NA, so you wind up with a larger set of players at a lower pick rate. Champs with approximately equal pick rates on NA and KR end up with almost 2x as many games on KR. You can work out total games by taking any number on the site, IE there's 1406 games of Sej with smite+flash at an 98.74% pick rate. 1406/0.9874=1424 games approximately. Doing the same thing for KR there's 1804 at 99.54%, 1804/99.54=1812. A larger amount of total games by over 400, although if this was in exact opposition to what other results say you'd still question it because it's not a huge sample.

It's easy to think about how small sample sizes work when you think about an extreme example. Let's say 97% of people like icecream, an overwhelming majority. I go out and ask a random person if they like icecream. 97% of the time my sample will still be representative, only 3% of the time it won't represent the population.

The KR sample is a supporting sample, it's not defining in and of itself, it's not a big enough sample. The fact that it lines up supports the other samples. LoG also has to be considered as this because they don't release their sample sizes, which is one thing I hate about that site.

The NA sample is conflicting data, either it is wrong or the rest of the data is skewed or wrong. It's obvious just at a basic level which of these is true, but you can look at it more deeply. The sites that store medium term winrate (op.gg and LoG) both show Sejuani falling pretty steeply. She's falling in winrate on all 3 large servers according to op.gg, there's one point of data that conflicts with this and that's this patches NA data. In the other two regions Sejuani's winrate has fallen every single patch. In NA it's fallen or stagnated every other patch. With that much going against the result something is up with it.

Edit - so as a conclusion, I'm going to say that none of the sites are perfectly accurate. If we take out the outliers (LeagueofGraphs and champion.gg) or average out all the data we have, she still sits around 48-9% winrate.

LoG is the opposite of an outlier. It's the closest single set of data to the mean and is the median. The definition for an outlier is falling outside 2.5x the interquartile range from the quartile figure. The quartile figures are the KR op.gg data (lower) and the lolayltics data (upper), I counted the points for op.gg separately because this needs at least 5 points. The interquartile range is 0.7, therefore data is an outlier if it's above 1.75 above lolayltics data. That means anything above 50.54 is definitionally an outlier. Champ.gg is at 50.84, it's the only data I can find outside of LoLking which falls into the actual outlier range. It can be worse though, LoL king is at 52.7%, which is just fucking hilarious. Even if you used na.op.gg as the upper quartile LoLking would still be an outlier.

I was na.op.gg an outlier, that was a poor choice of words. It's not an outlier, rather a statistically skewed sample.

ChaoticPineconeJan 22, 20171 votes

{quoted}

Riot doesn't do things like this, a single Rioter is the best you'll get.

And sorry if I dont take the word of a single Rioter whom I've never heard of, who didnt post this information on the Boards or any Riot-endorsed medium.

I got the number for thee op.gg average multiplication by finding the total number of games per region and dividing that by total games, this is a proportional average which gives more value to larger sample sizes, the games are valued evenly rater than the regions, a win from NA is not worth extra because there are fewer games. op.gg average: 49.58x0.235+48.01x0.298+45.84*0.467=47.37

Oh no - youre not off the hook that easily. There are more than 3 regions. Youre taking the 3 most populous regions and ignoring the rest - you took about 65% of the players and presumed to have it indicate 100% of the population.

Average of all sites=(47.37+47.1+48.71+50.83)/4=48.50

Absolute Variance from 4 site mean

LoG=48.50-47.1=1.4

op.gg=48.50-47.37=1.13

lolalytics=48.71-48.50=0.21

champ.gg=50.83-48.50=2.33

sum of absolute variance=1.4+1.13+0.21+2.33=5.07

average variance=5.07/4=1.27

You can already see from this that champ.gg varies a lot more than the other sites, but there's one more thing you can do to show how far of an outlier it is, you take the sum of absolute variance and compare it to the average variance on 3 sites.

Average of 3 sites=(47.37+47.1+48.71)/3=47.73

Absolute Variance from 3 site mean

LoG=47.73-47.1=0.63

op.gg=47.73-47.37=0.36

lolalytics=48.71-47.73=0.98

sum of absolute variance=1.97

average variance=1.97/3=0.66

As you can see there's far less variance between these 3 sites than there is if you add in champ.gg, this is typical of outliers.

Okay, you can do math. Of course when you take out the top-number there is less variance - thats how variance is calculated. That doesnt mean LoG is more accurate. It just means from the 3 sites you handpicked, there is a lower varience among them. There are sites that you decided to ignore, because it doesnt "fit" with your data.

That's pickrate, not sample size. The KR server has more players than NA, so you wind up with a larger set of players at a lower pick rate. Champs with approximately equal pick rates on NA and KR end up with almost 2x as many games on KR. You can work out total games by taking any number on the site, IE there's 1406 games of Sej with smite+flash at an 98.74% pick rate. 1406/0.9874=1424 games approximately. Doing the same thing for KR there's 1804 at 99.54%, 1804/99.54=1812. A larger amount of total games by over 400, although if this was in exact opposition to what other results say you'd still question it because it's not a huge sample.

It's easy to think about how small sample sizes work when you think about an extreme example. Let's say 97% of people like icecream, an overwhelming majority. I go out and ask a random person if they like icecream. 97% of the time my sample will still be representative, only 3% of the time it won't represent the population.

First of all, we dont know if the numbers we have are a sample. I was under the impression that since the sites use Riot's own API, they are representative of the population.

Secondly, pickrate is more indicative than playrate. I dont know why youd think otherwise. A sample size of 100/1,000 people more accurate than 200/10,000, even if one sample is larger than the other. Sure, Sej could theoretically be played 2x as times on the KR server than the NA (she wasnt - it was about 1000 for both) but the pickrate, not the playrate is the best indicator. Even if there are more people playing on the KR sever, we are talking about winrates here - the winrates get adjusted based on the pickrate and the overall number of games played. So again - I dont know why you trash NA's data, just because there are less games.

The NA sample is conflicting data, either it is wrong or the rest of the data is skewed or wrong. It's obvious just at a basic level which of these is true, but you can look at it more deeply. The sites that store medium term winrate (op.gg and LoG) both show Sejuani falling pretty steeply. She's falling in winrate on all 3 large servers according to op.gg, there's one point of data that conflicts with this and that's this patches NA data. In the other two regions Sejuani's winrate has fallen every single patch. In NA it's fallen or stagnated every other patch. With that much going against the result something is up with it.

It might be because different servers have different metas. For example, op.gg lists Caitlyn as NA's most played ADC in challenger for the past month. On KR, they list her as 8th most played. On NA, Ryze is the 9th most played; on KR, he is 25th.

Different regions have different power picks. In NA, Sej has a better record than in KR. But to find her overall power, you have to take them all into consideration - not just ignore NA because she as a higher winrate on NA. That doesnt mean NA's data is "skewed" or "wrong".

LoG is the opposite of an outlier. It's the closest single set of data to the mean and is the median. The definition for an outlier is falling outside 2.5x the interquartile range from the quartile figure. The quartile figures are the KR op.gg data (lower) and the lolayltics data (upper), I counted the points for op.gg separately because this needs at least 5 points. The interquartile range is 0.7, therefore data is an outlier if it's above 1.75 above lolayltics data. That means anything above 50.54 is definitionally an outlier. Champ.gg is at 50.84, it's the only data I can find outside of LoLking which falls into the actual outlier range. It can be worse though, LoL king is at 52.7%, which is just fucking hilarious. Even if you used na.op.gg as the upper quartile LoLking would still be an outlier.

Which goes back to the sites you used. Depending on which sites you used, the varience changes. Of course LoG is close to the mean and median when you only take 3 data points, and one of the sites posts similar numbers about LoG. If I decided to run your varience and deviation tests with LoG, Champion.gg, and Lolking youd find that LoG has the highest deviation. If you used LoG, Lolalytics, Op.gg, Champion.gg, and Lolking, youd find that Lolking and LoG have the highest deviation.

Doesnt show me why LoG is the end-all, be-all of winrates.

I was na.op.gg an outlier, that was a poor choice of words. It's not an outlier, rather a statistically skewed sample.

And reiterating my point before, I dont think the data is a "sample". Na.op.gg has over 1000 champions played today by challenger players. Once you take into account that there are only 200 challenger players (whom I'm guessing didnt all play today) and that Challenger games often have Master players in it as well (from the streams I see, its often 2-3 Challengers in a game with 7 Masters), thats a large number of data points for a "sample".

LankPantsJan 22, 20172 votes

And sorry if I dont take the word of a single Rioter whom I've never heard of, who didnt post this information on the Boards or any Riot-endorsed medium.

I'm sorry if I trust his word over yours.

If you want to read some of the stuff he's said here's his profile. He's on the live balance team, he has a fair idea about this sort of stuff considering it's his job.

Oh no - youre not off the hook that easily. There are more than 3 regions. Youre taking the 3 most populous regions and ignoring the rest - you took about 65% of the players and presumed to have it indicate 100% of the population.

op.gg doesn't store statistics for many regions. The only other two regions are EUNE and BR, all other regions redirect you to Korea. These regions have small samples close to the average, so they wouldn't change very much anyway.

It might be because different servers have different metas. For example, op.gg lists Caitlyn as NA's most played ADC in challenger for the past month. On KR, they list her as 8th most played. On NA, Ryze is the 9th most played; on KR, he is 25th.

Different regions have different power picks. In NA, Sej has a better record than in KR. But to find her overall power, you have to take them all into consideration - not just ignore NA because she as a higher winrate on NA. That doesnt mean NA's data is "skewed" or "wrong".

Yes, that explains why Sejuani's winrate was plumiting over the last two patches on NA, EUW, BR, EUNE and KR and continued to fall this patch in every other region. Use Occam's fucking Razor.

Which goes back to the sites you used. Depending on which sites you used, the varience changes. Of course LoG is close to the mean and median when you only take 3 data points, and one of the sites posts similar numbers about LoG. If I decided to run your varience and deviation tests with LoG, Champion.gg, and Lolking youd find that LoG has the highest deviation. If you used LoG, Lolalytics, Op.gg, Champion.gg, and Lolking, youd find that Lolking and LoG have the highest deviation.

LoLking is a joke amongst players for a reason, their statistics aren't even consistent with their own site. Look at the graph for highest winrate in ranked, Swain is the top winrate at 54.4% and a 4.3% pickrate. Now go to Swain's profile, it says 55.0% win rate with a 1.2% pickrate. You can't trust that site.

Doesnt show me why LoG is the end-all, be-all of winrates.

LoG isn't the be-all end-all of the stats I listed, it's the worst. They don't list their sample sizes at all, I don't know how accurate they are. I just know that they're the most average of all the sites, they are closest to the mean and they are the median.

And reiterating my point before, I dont think the data is a "sample".

The data can be one of two things, it's either a sample or the population. If it's the population that means that every single game on the server played at plat+ is included. I can tell you that's not how these sites work, it's even in the name. They're statistics sites, not a database.

thats a large number of data points for a "sample".

Yes, it is. That's why op.gg is a good site. I wouldn't be surprised if they use a different method to pick up all challenger games, it's a more digestible number but I can tell you plat is a sample.

Lolaytics, who have in past specifically stated their data is a sample have 4,395 games of Sejuani in NA specifically, a larger sample than na.op.gg. If you're wondering what that does for region specific winrate, it's 47.9%, so you're arguments about different metas on different servers (which is BS by the way) aren't really supported.

ChaoticPineconeJan 22, 20171 votes

{quoted}

I'm sorry if I trust his word over yours.

You dont have to trust my opinion to be suspect to his. If this is a Riot-endorsed opinion, I want something more - either a post on the Boards or something to indicate that a certain site isnt accurate. Someone posting on a non-Riot affiliated twitter where he has no accountability doesnt give me confidence that the information he is spreading is correct.

If you want to read some of the stuff he's said here's his profile. He's on the live balance team, he has a fair idea about this sort of stuff considering it's his job.

It might be his job, but the medium by which he is providing the data is suspect. If its his job, why doesnt he post things like that on the Boards? When he posts things on his private twitter, he isnt representing Riot as a company. When a Rioter posts things on the boards, they are. The repercussions for posting something like that are different depending on the medium and the capacity they are posting in - if they decide to post something on their private page, it makes me wonder why they arent doing the same on a Riot page.

op.gg doesn't store statistics for many regions. The only other two regions are EUNE and BR, all other regions redirect you to Korea. These regions have small samples close to the average, so they wouldn't change very much anyway.

This has no basis in fact.

In fact, EUW (about 20% of the population) seems to have the lowest winrate by far around 45.8%. KR and EUNE (40% of the population combined) have her at 48-48.8% winrate. NA (15%) has her at 49.8%.

You actually have no idea about other regions. Instead, youre using the data from 3 of the 4 given regions on op.gg, including an absurdly low EUW winrate. Yet NA is the one that you decide to omit in your analysis.

Yes, that explains why Sejuani's winrate was plumiting over the last two patches on NA, EUW, BR, EUNE and KR and continued to fall this patch in every other region. Use Occam's fucking Razor.

Has nothing to do with the current discussion. Are we talking about past patches or talking about now? Way to bring in off-topic, irrelevent things into the discussion.

LoLking is a joke amongst players for a reason, their statistics aren't even consistent with their own site. Look at the graph for highest winrate in ranked, Swain is the top winrate at 54.4% and a 4.3% pickrate. Now go to Swain's profile, it says 55.0% win rate with a 1.2% pickrate. You can't trust that site.

LoG isn't the be-all end-all of the stats I listed, it's the worst. They don't list their sample sizes at all, I don't know how accurate they are. I just know that they're the most average of all the sites, they are closest to the mean and they are the median.

LoG is the most average of the sites you used to calculate the average.

The data can be one of two things, it's either a sample or the population. If it's the population that means that every single game on the server played at plat+ is included. I can tell you that's not how these sites work, it's even in the name. They're statistics sites, not a database.

Statistical analysis can also be derived from using the population database.

Statistics is generally used for inferring population data from sample data, because thats the most efficient way to conduct research. But it also encompasses analyzing data, not just the data-collection methods. You can have statistical analysis on a database.

Yes, it is. That's why op.gg is a good site. I wouldn't be surprised if they use a different method to pick up all challenger games, it's a more digestible number but I can tell you plat is a sample.

Lolaytics, who have in past specifically stated their data is a sample have 4,395 games of Sejuani in NA specifically, a larger sample than na.op.gg. If you're wondering what that does for region specific winrate, it's 47.9%, so you're arguments about different metas on different servers (which is BS by the way) aren't really supported.

Whats the sample on na.op.gg? Why does Lolaytics have closer winrates to na.op.gg compared to LoG if its BS?

See, this is the problem with statistics - people see what they want to see. You jump to the conclusion you want and cite the data as your basis, but dismiss other explanations of the same data based on conjecture and theory while using suspect methods of analysis to "dismiss" my points.

"There are three kinds of lies: lies, damned lies, and statistics."

MagikarpIsBossJan 21, 20171 votes

Aw fuck dude, I always used champion.gg. I only stopped because I learned that their reccomended builds are complete shit (which is what I mainly used it for). I see builds that are "Highest winrates" with 2 titanics and 1 ravenous hydra (on camille a few weeks ago) and some other builds like that.

LankPantsJan 21, 20173 votes

now you get 5% bonus damage 2/3rds of the way down the tree, but tanks getting 6% bonus HP was broken and needed to be kneecapped because, to quote Riot 'squishy champs can't use it'. Masteries are a fucking joke right now.

GayFluffyUnicornJan 21, 20171 votes

Ofc it is a joke when the whole tree is about DAMAGE DAMAGE ON TOP OF EVEN MORE DAMAGE YAY. I wish there would be way more utility options like stormraider. It would be good on a lot of champs but why take it when you don't need to hit run and instead can hit and kill 9/10 times. I also don't like the fact that the percent pen is in a tree only used by adcs (including the likes of yasuo) with maybe two mages taking deathfire which is subpar on most champs but when the enemy has zac jungle, leona bot and maokai top i'dlike to have more pen but i will lose lane most likely on a bursty champ without TLD.

CancerBasedGodJan 21, 20171 votes

The defensive tree is possibly the strongest. While it has a lot of weak options it also has fearless and CotC, which is what combines for those super safe dives. You get a shield and an armor boost? That's the problem.

LankPantsJan 21, 20176 votes

Ferocity is the strongest tree and it's not even close. The first tier of the tree is unconditionally good, which is as good as it gets for first tier. The second tier is versatile with 3 very strong options. The third tier has an unconditionally good and a conditionally strong mastery. The 4th tier is basically a 5% increase in damage. The 5th tier is once again unconditional, but this time it's amongst the strongest masteries in the game. Ferocity also has 2 of the top 4 Keystones, better than either of the other trees making it more flexible.

You get a shield and an armor boost? That's the problem.

No one takes increased resists at any point in the tree on the vast majority of champs. 5% increased armour/MR is the Rammus mastery, it's only really good on him and other people just ignore it. Fearless would be a very strong mastery if Insight didn't exist, it turns out Insight just winds up being better. Legendary Guardian VS Swiftness is closer, the general consensus is that Swiftness edges it out on most champions however, I personally run LG, but I'm also a Sejuani main and Sej has some inbuilt antisynergy with Swiftness and some extra synergy with LG since she tends to be a health stacker.

The thing about Resolve is that you spec the whole tree for 2 masteries. The only things you want if you're in Resolve are your Keystone and Insight. The first 3 tiers of Resolve are basically hot garbage, the only good mastery in them is Runic Armour and even that's situational. Tier 5 is also pretty weak compared to the pen options that the other trees have. The whole tree is currently weaker than Ferocity except for CotC and Insight, you could argue that Fearless is also on par but it's just weaker than Insight so it never ends up mattering.

CancerBasedGodJan 21, 20171 votes

Poppy says hi. Look just because the resolve tree is conditional doesn't mean it's not strong. Good tanks like Poppy and Rammus destroy with it. It lets divers like Hecarim dive safer and earlier. It is a very strong tree. Cunning is honestly the one I'd say is weakest. The third tier is strong, and Thunderlords is strong, but the rest is fairly conditional.

LankPantsJan 22, 20174 votes

Poppy says hi. Look just because the resolve tree is conditional doesn't mean it's not strong. Good tanks like Poppy and Rammus destroy with it.

I'm sorry to break it to you buddy, but Poppy runs Insight. There isn't a rank where Poppy picks Fearless more than Insight. Even Rammus, the champ with the best logical scaling with Fearless takes Insight more at all ranks.

It is a very strong tree.

It's not though. It has the weakest first and second tier, the second tier is garbage compared to both of the other trees. The third tier is fair if you can use runic armour, if you can't it's by far the weakest in the game. The fourth tier has Insight, which is the strongest non keystone mastery in the game, so that's one strong point for resolve. The fifth tier is hot garbage. CotC is a saving grace for the tree, it's one of the few good masteries.

I think after the CotC gutting on the PBE you'll start to see tanks running 0/18/12 or 12/18/0 with Thunderlords. It's just not going to be worth a full spec into Resolve anymore. Personally I'm liking the look of 12/18/0 on Sejuani next patch.

Cunning is honestly the one I'd say is weakest. The third tier is strong, and Thunderlords is strong, but the rest is fairly conditional.

Cunning has probably the strongest first tier. Assassin is solid for second tier. The strongest third tier. The second best fourth tier. The best fifth tier master in Precision. Cunning is undoubtedly the second best tree.

KeevalroyJan 21, 20175 votes

Ooorrrrr...we revert to the mastery system in season 5...the one that didn't break everyone and everything it touched.

SoryInAdvanceJan 21, 20174 votes

the one that made csing on mages easier with just auto attacks because autos dealt scaling magic damage. miss that.

watch the bushJan 21, 20175 votes

Why do you want to rework keystones? Just fucking remove them, stop over complicating it. They're not needed and never were.

KeevalroyJan 21, 20175 votes

[{quoted}](name=watch the bush,realm=NA,application-id=3ErqAdtq,discussion-id=bYfYiUOV,comment-id=000a,timestamp=2017-01-21T05:31:40.317+0000)

Why do you want to rework keystones? Just fucking remove them, stop over complicating it. They're not needed and never were.

"B-b-b-but muh meaningful choices doe." - Riot's defence.

watch the bushJan 21, 20173 votes

Lol, I was disappointed with them even adding keystones in the 1st place. Season 5 had its fair share of balance issues, but it was never as bad as 6 and 7. Friend's don't even play anymore cus Riot loves to add things that aren't needed.

GayFluffyUnicornJan 21, 20172 votes

at first i was really hyped because you actually had a choice between harass/poke in lane in form of Deathfire and burst on TLD until they nerfed it into the ground so that TLD is the only viable option on basically every midlaner except swain and maybe cassio. Well brand too and malz. Instead of nerfing the abusing champs accordingly riot chose to destroy it for every champ except those 4.

agbudarJan 22, 20171 votes

i dont think riot will drop them completely though.

i just hope they step away from damage, more damage. or stupid insane tankyness.

things like windspeaker blessing and storm raider surge are imo well designed and not really broken.

but because half the keystone break/warp the damage in the game so much they are far too niche to be used.

Sexy Jack RabbitJan 21, 20174 votes

More bitching about Poppy, Darius, Lee, Vi and Hecarim. Are we done with this circle jerk? We know they're unbalanced.

ProfDrDeathJan 21, 20173 votes

Poppy benefits from CotC -> Which will be nerfed. Darius -> Will be nerfed. Lee, Vi, Hecarim benefit from the general too high impact of the jungle -> Which Riot already said will be nerfed.

AirKingNeoJan 21, 20174 votes

Thunderlord's accounts for barely any damage. Honestly make it restore mana/energy instead.

Fervor has no interactions with Hurricane anymore except for the 25% of AD on the bolts. It isn't an onhit anymore.

Warlords is worthless above half health. Honeslty it giving 20% is a bit excessive. Try 15-17% or make it 10-20% with level scaling.

CotC is getting nerfed. It should be changed to include soft CC so it can be better balanced and adds more variety.

Deathfire touch should be removed and replaced because it only works with DoTs. Windspeakers should be removed because it only works with heal. Stormraider's is worthless because TLD is just better, but if you used my change then it would be viable. Bond of Stone is worthless and needs to be removed and replaced. Grasp of the Undying needs to not deal damage but just heal you, so remove damage and raise healing.

Keystones add some variety to play but some are obscure, thus should be replaced. Bruisers and ADCs usually choose between Warlords or Fervor. Mages and assassins always go TLD because why have speed from Stormraiders when you've already wasted cooldowns just to proc it. Tanks should have a choice between Grasp and CotC. However masteries like Deathfire, which requires a DoT or be Jhin/MF, Windspeaks, which requires a heal/shield, or bad masteries like Stormraiders and Bond of Stone need changes.

We all know they aren't being removed. Rather we come up with better ideas to fix them then just say, "No, they can't exist at all!"

EDIT: People don't understand how the game works. RIP this comment.

Automated RivenJan 21, 20179 votes

stormraiders is far far far from worthless. I legit quit reading after that.

AirKingNeoJan 21, 20171 votes

Uh yes it is. Mordekaiser is the only champion who uses it. That says a lot. Why would I need a speed boost if I already used by entire combo?

PacothegintJan 21, 20171 votes

Nasus can use it very well

AirKingNeoJan 21, 20171 votes

but he doesn't.

3 champions use stormraiders.

One because he lacks mobility, one because it's available for him, and the other is because he deals so much damage he doesn't need TLD.

PacothegintJan 21, 20172 votes

except he does http://champion.gg/champion/Nasus

AirKingNeoJan 21, 20171 votes

High playrate, low number of games on Stormraiders. People are going all kinda of keystones on Nasus.

Sp33d Zer0Jan 22, 20171 votes

Ryze. Hec. Morde. Nasus. Garen. Darius. Seen it on more than a couple Singed. Khazix

All people I have personally, and fairly reguarly, seen abuse storm raiders

AirKingNeoJan 22, 20171 votes

Ryze is broken enough to have enough damage to use it.

Hec goes CotC now but he's one of the few who'd actually use it.

Morde needs mobility.

Nasus goes all different keystones.

Garen doesn't go stormraiders.

Darius doesn't go stormraiders.

Singed doesn't go stormraiders.

Kha is broken enough if he wanted to go it he could because he deals so much damage TLD is useless.

GayFluffyUnicornJan 21, 20171 votes

Lol. Must be silver- A lot of jayce uses it, ryze, morde, darius, nasus. There are a lot of champs who can utilise it. You need it to dodge skillshots, CC, trade better before your enemy can hit you, you can use it to follow up when he flashed/ghosted. Stormraider is one of the best and healthiest keystones. Also TLd accounts for barely any damage? do you play midlane? There you notice how broken it actually is.

HigeJan 21, 20172 votes

In all honesty, thurnderlord's is on the cunning tree though it works as a ferocity keystone, I like your idea though.

As for reworking keystones, riot said they would add more trees to the mastery system in order to remove thunderlord's from cunning and to make people have more a bit more variety from game to game

ProfDrDeathJan 21, 20172 votes

Ferocity is the sustained damage tree. Thunderlord's is burst...

VoidreyJan 20, 20172 votes

why i'm not not supportive of this option, what about just removing them altogether? just an idea ive been seeing around on the boards lately

101100111000Jan 21, 20172 votes

The old mastery system was better designed to be honest, it just had some subpar options that could have just been buffed, and some other options added.

ArakadiaJan 21, 20171 votes

I think Stormraiders should stay, its pretty healthy and creates some really cool builds like Surge Kench.

Weasel KenseiJan 21, 20171 votes

I always ignore the suggestions of people who can't manage to capitalize the first words of sentences.

Just like I ignore that guy with pants on his head and shoes on his hands telling me about my future.

Sure they may both be right about what they're saying, but the appearance of the source suggest otherwise.

TL(azily written)DR.

agbudarJan 22, 20171 votes

english is not my first language, so please get over it.

Weasel KenseiJan 22, 20171 votes

I clicked on your name to skim through your various forum post. You seem to have a full grasp on English, and how to use the shift key. Whatever language you come from, you seem to be able to use comas and periods.

Using the shift key to make the first letter of a sentence capitalized is a standard rule in English, now you know. While people often dont like it when I tell them I dont read their all lowercase post, what they often dont realize is Im not the only one. Other people ignore it too because it looks so lazily written they too assume its not worth their time. Food for thought.

XoziJan 22, 20171 votes

{quoted}

there are too many offending keystones that warp the game in a very bad way. thunderlords makes burst champions overburst into obscenity. fervor makes atack speed champions complete cancer once they get warlords bloodlust makes any and all weak adc laning phase obsolete.

CotC is a MANDATORY pick right now with the absurd amount of burst mages and assasins have when played right but gets abused to every and all diver champs with a 0.5 sec stun in their kit. (for the love of god make it bonus health not max health)

and since when is it OK to burst tanks? that was suposed to be their strong point being able to survive burst but not dps. yet most genuine tanks get blown to bits

with the exception of who also happens to deal maxhp% damage (yay) is considered a tank at this point with how he builds even though he is a juggernaut.

are now super mobile juggernauts because of CotC

I have an Idea, how about riot actually uses the roles they have listed in the store to define Masteries? Support champions have support masteries to choose from, Tanks have armor based masteries, ADCs have ADC based masteries. Just a larger variety of them in each category. This would also stop A LOT of cheesing. And prevent certain champion classes from breaking a mastery that was just buffed to compensate a weaker one.

JabberwöckyJan 22, 20171 votes

I just wish the champs could be simplified....

ADCS, two types, the kind that do hard poke, but lose to anyone who has DPS, and one that is more AA based for longterm DPS, but cant poke for shit.

Tanks, Two types again, has shields/healing/aoe to survive forever as a meat shield under turrets, and another that is more CC AOE based with massive resistance/hp.

Supports, Healers, who whole objective is to heal/provide buffs/debuffs, Kill laners, who if get a kill, become a threat, it not, their useless all game. Tanks, as above.

Brutes/Junglers Just a beefier tank, who doesnt have as much resistance but does more damage.

Assassins, whose whole job is to get in, do massive amount of damage, and get out, not tank, not kill 3 people (except like kat).

Mages, of course, like above, either assassin type, or a poke type.

Sp33d Zer0Jan 22, 20171 votes

{quoted}

fervor makes atack speed champions complete cancer once they get warlords bloodlust makes any and all weak adc laning phase obsolete.

But that isn't how that interaction works at all. You don't get extra fervor from hitting multiple people with bolts from Runaan's, it just gives a little extra damage onto your bolts (25% of what you get from fervor). And warlords is almost mandatory, not because of laning phase, but because unless you are an AD with a very strong synergy with fervor, you HAVE to get it because with a triple crit build (almost the only viable build for every ADC) you have 0 life steal to deal with tanks mid game.