Recent Topics

Ads

Some Ranked Statistics

Let's talk about... everything else
User avatar
kmark101
Posts: 482

Re: Some Ranked Statistics

Post#31 » Mon Aug 02, 2021 6:08 pm

Ranked died because certain individuals with extremely low skill que'd nonstop and if you accidentally got placed into the same team with them (since they literally been in que 24/7 wtf) that meant that you lose all night progress of your session... so eventually noone wanted to que because it sux to lose all progress due to some nolifer (on both sides), hence ques died off.

Not sure what is the solution, but maybe you shouldnt lose ALL your progress after 1-2 bad games and there should be a way to limit the participation, for example max 10 ranked games / account / day allowed, along this line.
Gryyw - Ironbreaker

Ads
Rapzel
Posts: 386

Re: Some Ranked Statistics

Post#32 » Mon Aug 02, 2021 6:12 pm

Caduceus wrote: Mon Aug 02, 2021 6:05 pm
Kabuco wrote: Mon Aug 02, 2021 5:12 pm
The underlined part refers only to its sentence, not the original topic.

Do you not understand my meaning, or pretending not to understand? I guess I shall err on the side of caution:

If you conclude that Ranked died because of a lack of players, wouldn't you want to know why:
A. a significant number of players is quiting Ranked after only a few games played
B. why certain classes are scoring much higher in this regard than others

Much like, in your example, if you had a game and 50% of your playerbase was quitting before beating the first level, it would probably also be a good moment to scratch behind one's ears to think about what is going wrong.
Though you have not provided anyone with relevant data about how many games each individual has played before they stopped playing, all you've provided is data that says that all careers have won at least one game in solo ranked.

Caduceus
Posts: 653

Re: Some Ranked Statistics

Post#33 » Mon Aug 02, 2021 6:19 pm

Rapzel wrote: Mon Aug 02, 2021 6:12 pm
Though you have not provided anyone with relevant data about how many games each individual has played before they stopped playing, all you've provided is data that says that all careers have won at least one game in solo ranked.

What my data clearly shows is that, for example, less than 30% of Witch Hunters that attempted Ranked stuck around to play more than 10 games (following my second measurement). Where that other ~70% went is not interesting to you?

Does it matter whether individual Witch Hunters stopped playing after 1 game or after 9 games?

The fact is that they stopped playing, and that Ranked is and has been suffering from a lack of players!
"I watched a snail crawl along the edge of a straight razor. That's my dream; that's my nightmare. Crawling, slithering, along the edge of a straight razor... and surviving." - Colonel Walter E. Kurtz

User avatar
diegomess
Posts: 216

Re: Some Ranked Statistics

Post#34 » Mon Aug 02, 2021 6:43 pm

eheemm can we have the rewards for the top 5 players or 6 of each class instead of 3?, like many ppl here stated it was kinda messy and many ppl had lucky and unlucky matches, can more ppl that actually tried be also rewarded? it wouldnt hurt right?
Chosen Dahaka RR88
BO Zamedi 80

Kabuco
Former Staff
Posts: 109

Re: Some Ranked Statistics

Post#35 » Mon Aug 02, 2021 8:08 pm

Caduceus wrote: Mon Aug 02, 2021 6:05 pm
Kabuco wrote: Mon Aug 02, 2021 5:12 pm
The underlined part refers only to its sentence, not the original topic.

Do you not understand my meaning, or pretending not to understand? I guess I shall err on the side of caution:

If you conclude that Ranked died because of a lack of players, wouldn't you want to know why:
A. a significant number of players is quiting Ranked after only a few games played
B. why certain classes are scoring much higher in this regard than others

Much like, in your example, if you had a game and 50% of your playerbase was quitting before beating the first level, it would probably also be a good moment to scratch behind one's ears to think about what is going wrong.
None of your presented data is a solid base to answer any of your questions.

for a it would be: How many games were played for each class as median average and with a diagramm with x=games y=players

for b it would be just the leaderboard itself
If you have a red name, you're on the wrong side!

OCRANA ...deal with it

Kabuco BW RR8x
Kynai WE RR8x

Caduceus
Posts: 653

Re: Some Ranked Statistics

Post#36 » Mon Aug 02, 2021 8:12 pm

Kabuco wrote: Mon Aug 02, 2021 8:08 pm
None of your presented data is a solid base to answer any of your questions.

No, it raises those questions.
"I watched a snail crawl along the edge of a straight razor. That's my dream; that's my nightmare. Crawling, slithering, along the edge of a straight razor... and surviving." - Colonel Walter E. Kurtz

User avatar
BluIzLucky
Posts: 689

Re: Some Ranked Statistics

Post#37 » Mon Aug 02, 2021 8:49 pm

Not attacking you, just trying to make you see another aspect of this debate.
Caduceus wrote: Mon Aug 02, 2021 3:57 pm I am not a professional statitician, nor do I claim to be, nor do I wish to be.

This is what you're getting, and if you want more or have it done better, you'll have to provide it yourself.
Right.. Didn't expect you to be, and I was being a bit pedantic and show-off-ish, sorry about that, my point was you use strong words that convey a confidence about the data that I would not personally have used, but whatever..
There are no "glaring issues" with my data at all, though no data or statistics are perfect and you can always poke holes in them.

It is just data. There is nothing faulty about its quality. Anyone can do the same measurements of the leaderboard. You would maybe have a point if I were drawing conclusions or pushing a certain narrative: I am not.
When I say it's bad data it doesn't mean you collected it wrong, it means that the data that is there can be misleading, inaccurate or
too inconclusive, and that's especially true if you just leave it to others to interpret.

Take it from someone who's cleaned and discarded millions of records.

For example you could say BWs are weak, but then you have MrGlass in vanq gear with an 80% win rate (and personally the best BW I've played with). If it was a class problem and not a player problem then that should be unlikely.. And the same is the case on basically all classes.


I already responded to Kabocu's suggestion to include only characters with 10 games or more played: it showed much the same picture.


Okay.. So you've just adopted someone else's bias.
Parameters are set according to what you are trying to show/conclude and will be biased. Not necessarily saying you have picked wrong, it's just an example in response to you saying the data is not biased.

Let people interpret the data themselves, discuss and share their opinions on it. I have no fear of such discussions and open debate. Do you?
My response to that would have been to loosen my position on the matter and edit my original post with some disclaimers and guidance, since I would feel responsible for making sure the debate is based on a general understanding and to limiting the spread of misinterpretations..

Yours is to say that's none of my business and ask if I'm against free speech? Okidoki, good example of open and honest debate..

If you posted a topic about climate change and people came to the conclusion that it isn't manmade either your data is bad or you've presented it in a way that misled people, you would have failed to lead the conversation. Not saying you've failed here, just letting you know why these things are needed.
I have shared my thoughts about it. The parameters I have picked show, in my view, a measure of how well classes are able to participate and I find the lower parts of the list worrying in that regard.
Cool, so you do have a conclusion (and in extention a narrative), anyway if by participate you mean..
Being able to compete then I disagree, all those classes have players I dread going against or I'm happy to see on my team, meaning in the hands of good ranked players with decent gear and spec, these classes are viable. And your data doesn't show this well.

If you mean they have a higher ranked learning curve then sure, we are on the same page and I think your data supports that :)
Last edited by BluIzLucky on Tue Aug 03, 2021 2:33 am, edited 1 time in total.
SM - Arhalien +80 | AM - Shaheena +80
ZL - Wildera +70 | BG - Blackcrow +70

User avatar
Amdus
Posts: 115

Re: Some Ranked Statistics

Post#38 » Mon Aug 02, 2021 8:54 pm

Rapzel wrote: Mon Aug 02, 2021 6:03 pm What mental gymnastics did you do this morning to get the "data" to provide anything closely resembling win rate?

The "data" consists of every played character that's won at least one game over the total amount of characters of that career that's participated in solo ranked. I.E if 100 Slayers have won 1 game but lost 100 games they would have 100% here while if 4 DoK out of 10 won all their games and the rest lost their games we would get 40%.
First and foremost it wasn't me who wasted precious and unrecoverable time collecting the data, it was Caduceus. Why are you quoting the data as if it was some sort of conspiracy theory when it's in fact real and well documented? You can even open the leaderboard ingame an see it by yourself, so stop pretending this is something made up. Also, ironic that you have chosen the slayer as an example when this class is the one with the highest win rate disparity.

Ads
User avatar
agemennon675
Posts: 500

Re: Some Ranked Statistics

Post#39 » Mon Aug 02, 2021 9:09 pm

Thank you very much for sharing this data, unfortunately we dont have a win/rate tab on the ranked leaderboard so this is really helpful and not suprising at all for someone who play alot of different classes. It will probably be igonred but its always nice to see a dataset on the forums.
Destruction: 40-BG / 40-DoK / 40-Chosen / 37-Mara / 37/Sorc / 36-SH / 36-Choppa / 24-Shaman / 16-WE
Order: 40-SW / 40-SM / 40-WP / 40-WL / 39-Kotbs / 38-BW / 33-AM / 22-WH / 16-RP / 12-Slayer

Rapzel
Posts: 386

Re: Some Ranked Statistics

Post#40 » Mon Aug 02, 2021 10:34 pm

Amdus wrote: Mon Aug 02, 2021 8:54 pm
Rapzel wrote: Mon Aug 02, 2021 6:03 pm What mental gymnastics did you do this morning to get the "data" to provide anything closely resembling win rate?

The "data" consists of every played character that's won at least one game over the total amount of characters of that career that's participated in solo ranked. I.E if 100 Slayers have won 1 game but lost 100 games they would have 100% here while if 4 DoK out of 10 won all their games and the rest lost their games we would get 40%.
First and foremost it wasn't me who wasted precious and unrecoverable time collecting the data, it was Caduceus. Why are you quoting the data as if it was some sort of conspiracy theory when it's in fact real and well documented? You can even open the leaderboard ingame an see it by yourself, so stop pretending this is something made up. Also, ironic that you have chosen the slayer as an example when this class is the one with the highest win rate disparity.
Caduceus wrote: Fri Jul 30, 2021 8:15 pm a. had a positive MMR
b. had 0 MMR but at least a single victory.
1. Where is the win rate calculated? The "data" collected is a collection of characters with positive MMR (I.E character needs to have drawn or won a game) or characters with 0 MMR and at least 1 win, how do I calculate the win rate from this data? Please explain.
2. You don't think it's weird that when the numbers are adjusted to contain only characters with OVER 10 games played, there's 1 career that has above 50% win rate, IB.
3. The data is meaningless, it does not say anything as several people have already stated.
4. Why I put it in quotation marks is because setting up arbitrary rules to rule out parts of the data set does not make it data about ranked, it's a clamped example of someone trying to prove some "bias".
5. It was an example, there's not 100 slayers that have lost 100 games and won 1 game on the leaderboard, I thought making it obviously exaggerated would make it clear it was an example, it apparently failed.

As I stated earlier in this thread, "23. Bright Wizard: 13/37 = 35.1%" and then you look at the BW that have 1 played game and you end up with having to remove 17 BW from the set, resulting in 11/20 (1 BW won their game and one drew I think), and all of the sudden the % value is BW = 55%, what does this mean? It has no value what so ever, it's useless "data".
Another example would be taking Sorc vs BW, there's 2-3 Sorcs above 1k MMR and 0 BW above 1k MMR, from this we can conclude that Sorc > BW right?
Of course not, because there's more parameters that we need to take into consideration, such as rSH being the dominating ranged career in ranked and Order having fewer viable setups than Destro (albeit the available ones may be stronger).

Who is online

Users browsing this forum: Trickx and 13 guests