Some Ranked Statistics

[Caduceus]

So today I learned that 1393+1159 < 923 +1011.
"IB has highest participation".

You don't know the difference between a percentage and absolute numbers?

Ok.

[Rapzel]

So today I learned that 1393+1159 < 923 +1011.
"IB has highest participation".

You don't know the difference between a percentage and absolute numbers?

Ok.

So if we had one (1) person play the old career beast master/mistress that was planned to get implemented here during a short period, and they had 10 games played you think it would be fair to say that that career had 100% participation in solo ranked?

[BluIzLucky]

Spoiler:

Ok, but this still doesn't explain why these classes would be over-represented in terms of having more weak players compared to other classes. I would expect roughly the same amount of comparatively bad players on every class. Maybe classes that have a reputation for being hard to play could see a higher percentage of unsuccessful players, but we actually see the opposite with IB and BG having the highest amount of participation for Order and Destruction respectively!

So what is going on here?

The statistics you provide here are useless. IB?
This how you should make statistic(I don't count draws, only for order tanks[Not gonna waste more time])
Win ratio by all order tanks that played ranked solo:
Kotbz: Win:1393 Lost:1159 54.5% win ratio
IB: Win: 923 Lost: 1011 47.7% win ratio
SM: Win:593 Lost:673 46.8% win ratio

Still don't get why this thread is not closed after first post:D Lets make data from nowhere and fun begin:D

So today I learned that 1393+1159 < 923 +1011.
"IB has highest participation".

The data is not bad bad, it's decent for the specific question of how well a class adapts to ranked, and to me it's a fairly simple explanation, which I put on second page, but here's the expanded version:

• Competitive players will more often pick the most competitive/meta classes

• Competitive/meta (or previously competitive) classes will more often have better gear, since they will have more spots in organized warbands/city

• Ranked Class preparedness - classes on the top of the list will have more similar ranked play-styles/specs to the main content, the ones at the bottom will have significantly different spec, tactics, and play-style in ranked

• Ranked Class Skill level requirements - all ranged classes will for example need to be good at positioning, pre-kiting, kiting, flee+potting while applying pressure back, none of which is important in the main content.
  Squishy MDPS will also need to know when to disengage, but this is closer to how their main content plays.
  Knowing burst rotation, while important on all DPS, the ones lower on the list relies on this heavily.
  
  Compare all this to for example KS/CH/IB/BG; can you punt and maybe guard swap, like in all other content? You good to go.

• Ranked equipment preparedness - understanding minimum required gear for your class, it's lowest for healers and tanks and highest for light armored DPS. And you'll also need 8-12 different pots (preferably blue) and linis, while the preparedness or lack thereof might be close between all classes, my guess would be more ranged will be unprepared simply because they take less heat in general

Of course, even after all that, you still have classes that are just less competitive, either by way of being more reliant on other classes/players (SM/BO and possibly some DPS classes) or just being weaker in ranked (ASW vs SL).

Spoiler:

So today I learned that 1393+1159 < 923 +1011.
"IB has highest participation".

You don't know the difference between a percentage and absolute numbers?

Ok.

So if we had one (1) person play the old career beast master/mistress that was planned to get implemented here during a short period, and they had 10 games played you think it would be fair to say that that career had 100% participation in solo ranked?

At this point in the conversation it's should be known that "highest participation" means "highest % continued ranked participation", despite that never being said explicitly by OP.
And yes it would be, since we are comparing % participation between classes, and have very low sample size for all classes anyway, the difference between Order being on top and destro is basically removing/adding 5 to the numerator on each class.

[Cimba]

This is false on both accounts.

The first information gathered is from inside a match, and not outside.

And queueing when your friends are queueing is not playing the system. Only queueing when your friends are queueing is.

Your definition of "playing the system" just isnt really usefull for what you're trying to do here. The Win ratios we currently see can easily be achieved with only "inside" information (whatever that means). Meaning you log off after the first loss/draw and only keep playing on a win streaks. As anecdotal evidence I had 2 or 3 evenings where I won all matches. Similarly I had 1 or 2 nights where I lost everything. If I had stopped playing after the first loss each time than my win ratio would probably be 1:4 or even higher with the few games I have played.

The second point isnt really false if you look at your original statement. You just changed/refined the meaning of your statement to falsify mine. In an honest conversation one usually does that by saying "Let me clarify..."

Anyway I hope the devs can upload the entire database of the season so people can do their analysis without need to extract it manually.

[Rapzel]

When we have all the amount of players we can calculate the amount of games played in total (all tanks)/6 (simplified by faction x tanks divided by 3, this is made possible by the 2/2/2 rule) as tanks AFAIK cannot fill the DPS role like DoK/WP/AM/Sham.
There by allowing us to get an estimate about the representation of different careers in ranked, from this set we can then proceed to calculate the participation of each career and set up rules to further analysis. This has not been done, there for I believe the % values presented are biased.

This approach does not magically remove the elephant in the room though as we lack the data about what setups are the most successful (highest win rate) ones.
Why e.g. is Lllara the only successful (success = more MMR) WL, or premetherion the only successful Magus?

Feel free to correct me BluIzLucky

[BluIzLucky]

Spoiler:

This is false on both accounts.

The first information gathered is from inside a match, and not outside.

And queueing when your friends are queueing is not playing the system. Only queueing when your friends are queueing is.

Your definition of "playing the system" just isnt really usefull for what you're trying to do here. The Win ratios we currently see can easily be achieved with only "inside" information (whatever that means). Meaning you log off after the first loss/draw and only keep playing on a win streaks. As anecdotal evidence I had 2 or 3 evenings where I won all matches. Similarly I had 1 or 2 nights where I lost everything. If I had stopped playing after the first loss each time than my win ratio would probably be 1:4 or even higher with the few games I have played.

The second point isnt really false if you look at your original statement. You just changed/refined the meaning of your statement to falsify mine. In an honest conversation one usually does that by saying "Let me clarify..."

Anyway I hope the devs can upload the entire database of the season so people can do their analysis without need to extract it manually.

Yep, was going to make the same point, I don't think queue stacking/playing the system is as common as people make it sound, but it's fairly easy to point to the Ranked discord to say otherwise.. either way I also don't think it's particularly relevant to this conversation.

I went from 1-10-2 to 18-12-4 on my AM this season lol.. it happens..

Spoiler:

When we have all the amount of players we can calculate the amount of games played in total (all tanks)/6 (simplified by faction x tanks divided by 3, this is made possible by the 2/2/2 rule) as tanks AFAIK cannot fill the DPS role like DoK/WP/AM/Sham.
There by allowing us to get an estimate about the representation of different careers in ranked, from this set we can then proceed to calculate the participation of each career and set up rules to further analysis. This has not been done, there for I believe the % values presented are biased.

This approach does not magically remove the elephant in the room though as we lack the data about what setups are the most successful (highest win rate) ones.
Why e.g. is Lllara the only successful (success = more MMR) WL, or premetherion the only successful Magus?

Feel free to correct me BluIzLucky

My post was/is about representation at the top 100-200 (where % class distribution matters more - and a lot of the same data quality issues applies to my post).
The % values for this discussion represent how well a new ranked player can play any given class in ranked (e.g. Sorc is hard to play so has low % here, but is well represented in top 100-200 so has a good % there). And yes, there's still a bunch of disclaimers/warnings about the data that I won't iterate here, but I think it's good enough to discuss what OP is trying to discuss without touching on class balance.

[Caduceus]

Your definition of "playing the system" just isnt really usefull for what you're trying to do here.

This part of the discussion was a reaction to people suggesting that the top of the leaderboard can be seen as a reflection of class balance.

What "I am trying to do here" is not necessarily directly related to class balance.

[Acidic]

The problem with the numbers given in this thread is they are single points and thus do not give the options to interpret the results easily as there are no trends to follow.

That said I did some playing with the median of rr for classes played more than 10 games and noticed a certain mele class very high while anither mele class was low.

This result as such only showed that there was differences in the average and not anything about why or balance. When doing the median with toons over 1k the result was very different look two mobile dps classes became noticeable high numbers

<No graph supplied deliberately as these numbers with the data in ranked does not support conclusions only supposition>

My interpretation from the numbers of certain classes (only was able to look at classes with a reasonable population in each bracket, so not exhaustive numbers) was that there is:

A) indication that the lower bracket players may not be swapping to optimal build, pots ...
B) certain classes have a high skill cap (perform significantly better with certain players and awful in other players hands)
C) certain classes have low scaling with player skill

Wheather or not B or C are balance issues or QoL on certain classes is upto others .

In short statistics needs to be treated with heavy dose of sketasisim and alternative reasoning and investigations of trends needs to be done to understand what the numbers mean.

[Caduceus]

Spoiler:

The statistics you provide here are useless. IB?
This how you should make statistic(I don't count draws, only for order tanks[Not gonna waste more time])
Win ratio by all order tanks that played ranked solo:
Kotbz: Win:1393 Lost:1159 54.5% win ratio
IB: Win: 923 Lost: 1011 47.7% win ratio
SM: Win:593 Lost:673 46.8% win ratio

Still don't get why this thread is not closed after first post:D Lets make data from nowhere and fun begin:D

So today I learned that 1393+1159 < 923 +1011.
"IB has highest participation".

The data is not bad bad, it's decent for the specific question of how well a class adapts to ranked, and to me it's a fairly simple explanation, which I put on second page, but here's the expanded version:

• Competitive players will more often pick the most competitive/meta classes

• Competitive/meta (or previously competitive) classes will more often have better gear, since they will have more spots in organized warbands/city

• Ranked Class preparedness - classes on the top of the list will have more similar ranked play-styles/specs to the main content, the ones at the bottom will have significantly different spec, tactics, and play-style in ranked

• Ranked Class Skill level requirements - all ranged classes will for example need to be good at positioning, pre-kiting, kiting, flee+potting while applying pressure back, none of which is important in the main content.
  Squishy MDPS will also need to know when to disengage, but this is closer to how their main content plays.
  Knowing burst rotation, while important on all DPS, the ones lower on the list relies on this heavily.
  
  Compare all this to for example KS/CH/IB/BG; can you punt and maybe guard swap, like in all other content? You good to go.

• Ranked equipment preparedness - understanding minimum required gear for your class, it's lowest for healers and tanks and highest for light armored DPS. And you'll also need 8-12 different pots (preferably blue) and linis, while the preparedness or lack thereof might be close between all classes, my guess would be more ranged will be unprepared simply because they take less heat in general

Of course, even after all that, you still have classes that are just less competitive, either by way of being more reliant on other classes/players (SM/BO and possibly some DPS classes) or just being weaker in ranked (ASW vs SL).

Spoiler:

You don't know the difference between a percentage and absolute numbers?

Ok.

So if we had one (1) person play the old career beast master/mistress that was planned to get implemented here during a short period, and they had 10 games played you think it would be fair to say that that career had 100% participation in solo ranked?

At this point in the conversation it's should be known that "highest participation" means "highest % continued ranked participation", despite that never being said explicitly by OP.
And yes it would be, since we are comparing % participation between classes, and have very low sample size for all classes anyway, the difference between Order being on top and destro is basically removing/adding 5 to the numerator on each class.

The problem with the numbers given in this thread is they are single points and thus do not give the options to interpret the results easily as there are no trends to follow.

That said I did some playing with the median of rr for classes played more than 10 games and noticed a certain mele class very high while anither mele class was low.

This result as such only showed that there was differences in the average and not anything about why or balance. When doing the median with toons over 1k the result was very different look two mobile dps classes became noticeable high numbers

<No graph supplied deliberately as these numbers with the data in ranked does not support conclusions only supposition>

My interpretation from the numbers of certain classes (only was able to look at classes with a reasonable population in each bracket, so not exhaustive numbers) was that there is:

A) indication that the lower bracket players may not be swapping to optimal build, pots ...
B) certain classes have a high skill cap (perform significantly better with certain players and awful in other players hands)
C) certain classes have low scaling with player skill

Wheather or not B or C are balance issues or QoL on certain classes is upto others .

In short statistics needs to be treated with heavy dose of sketasisim and alternative reasoning and investigations of trends needs to be done to understand what the numbers mean.

Thank you for your replies. This was the type of conversation I was hoping for when I made this thread.

I think these are valuable insights, all of which offer logical explanations for why the numbers are the way they are.


One thing I'd like to address is that while a high skill cap on certain classes seems like a plausible explanation for why certain classes are seeing less participation than others, doesn't that mean that IB and Blackguard are expected lower on the list? In both measurements I have taken they are at the top in terms of participation for both Order and Destro. Are these classes less hard to play than is often assumed? Or is there some other explanation?

[BluIzLucky]

Spoiler:

Spoiler:

So today I learned that 1393+1159 < 923 +1011.
"IB has highest participation".

The data is not bad bad, it's decent for the specific question of how well a class adapts to ranked, and to me it's a fairly simple explanation, which I put on second page, but here's the expanded version:

• Competitive players will more often pick the most competitive/meta classes

• Competitive/meta (or previously competitive) classes will more often have better gear, since they will have more spots in organized warbands/city

• Ranked Class preparedness - classes on the top of the list will have more similar ranked play-styles/specs to the main content, the ones at the bottom will have significantly different spec, tactics, and play-style in ranked

• Ranked Class Skill level requirements - all ranged classes will for example need to be good at positioning, pre-kiting, kiting, flee+potting while applying pressure back, none of which is important in the main content.
  Squishy MDPS will also need to know when to disengage, but this is closer to how their main content plays.
  Knowing burst rotation, while important on all DPS, the ones lower on the list relies on this heavily.
  
  Compare all this to for example KS/CH/IB/BG; can you punt and maybe guard swap, like in all other content? You good to go.

• Ranked equipment preparedness - understanding minimum required gear for your class, it's lowest for healers and tanks and highest for light armored DPS. And you'll also need 8-12 different pots (preferably blue) and linis, while the preparedness or lack thereof might be close between all classes, my guess would be more ranged will be unprepared simply because they take less heat in general

Of course, even after all that, you still have classes that are just less competitive, either by way of being more reliant on other classes/players (SM/BO and possibly some DPS classes) or just being weaker in ranked (ASW vs SL).

Spoiler:

So if we had one (1) person play the old career beast master/mistress that was planned to get implemented here during a short period, and they had 10 games played you think it would be fair to say that that career had 100% participation in solo ranked?

At this point in the conversation it's should be known that "highest participation" means "highest % continued ranked participation", despite that never being said explicitly by OP.
And yes it would be, since we are comparing % participation between classes, and have very low sample size for all classes anyway, the difference between Order being on top and destro is basically removing/adding 5 to the numerator on each class.

The problem with the numbers given in this thread is they are single points and thus do not give the options to interpret the results easily as there are no trends to follow.

That said I did some playing with the median of rr for classes played more than 10 games and noticed a certain mele class very high while anither mele class was low.

This result as such only showed that there was differences in the average and not anything about why or balance. When doing the median with toons over 1k the result was very different look two mobile dps classes became noticeable high numbers

<No graph supplied deliberately as these numbers with the data in ranked does not support conclusions only supposition>

My interpretation from the numbers of certain classes (only was able to look at classes with a reasonable population in each bracket, so not exhaustive numbers) was that there is:

A) indication that the lower bracket players may not be swapping to optimal build, pots ...
B) certain classes have a high skill cap (perform significantly better with certain players and awful in other players hands)
C) certain classes have low scaling with player skill

Wheather or not B or C are balance issues or QoL on certain classes is upto others .

In short statistics needs to be treated with heavy dose of sketasisim and alternative reasoning and investigations of trends needs to be done to understand what the numbers mean.

Thank you for your replies. This was the type of conversation I was hoping for when I made this thread.

I think these are valuable insights, all of which offer logical explanations for why the numbers are the way they are.


One thing I'd like to address is that while a high skill cap on certain classes seems like a plausible explanation for why certain classes are seeing less participation than others, doesn't that mean that IB and Blackguard are expected lower on the list? In both measurements I have taken they are at the top in terms of participation for both Order and Destro. Are these classes less hard to play than is often assumed? Or is there some other explanation?

From what I can tell, people call them hard to play because they have a lot of buffs/debuffs to manage to play "optimally", but really they only have a couple of critical ones, IB has Oathbound, kneecapper and armor debuff, BG has crimson death and armor debuff, in return BG has best ST punt in game and IB is the best guard buddy, which makes up for any perceived difficulty.

Got a low lvl BG with most of the skills, and just feels smoother than SM, free infinite AP, very good dmg, mechanic is rarely felt and doesn't bug, would guess they still feel easier/better at lvl 40.. IB might be a different story.

So between good punts, good burst dmg and good protection, it should be fairly easy to land a single win with these classes.