|
There is a topic on reddit where a master's league player is asking about people's bans, picks, and common bans they see in an attempt to write an article about how to draft. The idea of looking at common bans as a way of seeing what is valued in your personal meta as well as the general meta struck me as interesting.
My procedure
I copied the front page of hotslogs filtering for each league. I removed all heroes with less than a 30% popularity (which is an arbitrary cutoff), then sorted the remaining heroes by how often they are banned. Note that there should be a correlation between banning and popularity, since all bans contribute to popularity (as far as I know)
Here are the results:
https://docs.google.com/spreadsheets/d/1SOrazph-hd3Wv8DZqo1VJ9KcxmdTiTYLowdriXCdd5k/pubhtml
Simple Observations:
* There is something approximating a stable ban-meta in heroleague right now, with the same 10 high-popularity characters getting banned across all leagues.
* Some characters are only feared in low leagues (Raynor, Valla, Medic, Tracer, LiLi). This shouldn't be a surprise.
* Some characters are only feared at higher leagues. Brightwing, Anub, Sonya, and Uther are on this list.
* Illidan is ignored in platinum, but feared at all other levels of play. I'm guessing this is an aberration of my arbitrary 30% cutoff.
The list of cross-league banned heroes:
Falstad Rehgar Kael'thas Li-Ming E.T.C. Zagara Muradin Thrall Sylvanas Johanna
|
What is popularity? % of games the hero is either picked or banned? And why is it above 100% in some cases?
|
Bans in heroleague are mostly pointless, people fret about them too much. Bans only gain good value when you know certain preferences or compositions of the opponnent or when some combo's exist that you really want to break up.
First ban especially is just as wash usually. With kerrigan now it has some uses as that should definitely be the ban if you aren't picking first but before the patch when no really OP hero existed it didnt matter much. Especially with people hardly prepicking or prepicking stuff that shouldn't be high priority at all (li-ming most of all) the ban becomes quite the crapshoot. Second ban is usually just a healer or tank because HL drafts fill up with assassin/dps picks first anyway.
|
How is Xul not in there lol. Most annoying hero ever.
|
Err... Where's Kerrigan? She's the #1 popularity ban on HL right now.
|
On July 22 2016 20:36 RouaF wrote: How is Xul not in there lol. Most annoying hero ever.
On July 22 2016 21:28 Random_0 wrote: Err... Where's Kerrigan? She's the #1 popularity ban on HL right now.
Check out the google docs spreadsheet I linked(it is color-coded for ease of use). Seeing the spreadsheet is probably more useful than the condensed tldr list, since you can see what the meta is for your specific league [and maybe what the meta is for the league you'd like to be in]
https://docs.google.com/spreadsheets/d/1SOrazph-hd3Wv8DZqo1VJ9KcxmdTiTYLowdriXCdd5k/pubhtml
There, you'll see that Kerrigan is not a factor at Silver league--- her popularity is below 30%, so she doesn't even factor into this list. I suspect this is because Kerrigan requires a modicum of skill to play, and maybe a little bit of teamwork to make dominate. I have no idea why xul is not at 30% participation in silver league: he's easy to play AND super dominate. Chalk this up to "silver league has no idea what they are doing", I would totally draft xul almost every game if given the chance.
|
On July 22 2016 19:47 Markwerf wrote: Bans in heroleague are mostly pointless, people fret about them too much. Bans only gain good value when you know certain preferences or compositions of the opponnent or when some combo's exist that you really want to break up.
First ban especially is just as wash usually. With kerrigan now it has some uses as that should definitely be the ban if you aren't picking first but before the patch when no really OP hero existed it didnt matter much. Especially with people hardly prepicking or prepicking stuff that shouldn't be high priority at all (li-ming most of all) the ban becomes quite the crapshoot. Second ban is usually just a healer or tank because HL drafts fill up with assassin/dps picks first anyway.
Well, I think bans are significant because they give a clue of what people either think are powerful or what people really dislike playing against. This list contains a lot of characters with extreme disruption (eg wailing arrow, sunder, mosh, gust, chain bomb, or just the random burstyness due to resets of li-ming). It also contains characters which are difficult to deal with (johanna is not easily stopped if she wants to get to your back lines, crosspushing zag is currently a monster). The exception to all of this might be muradin, but reverb hurts melee assassins so much that maybe he makes sense too.
It will be interesting to see if this trend continues in future balance patches. I guess if Raynor is ever super OP again he might crop up to the top of the ban list, but somehow I doubt a hero like Tassadar ever would across all leagues of play.
|
On July 22 2016 15:52 dpurple wrote: What is popularity? % of games the hero is either picked or banned? And why is it above 100% in some cases?
I am not sure how it works after bans were implemented, but my discussion with Ben [owner of hotslogs] before then revealed this formula: (this may not be exact--- it's the formula I proposed and he just said "yes" without more explanation, so maybe the formula is just similar but not exact)
a = (# of times a character is picked or banned) b = total number of heroes picked in a period
Popularity = a / (b/10)
However, games frequently have people from multiple leagues in them. So let's say we have a sample size of four games where one game had this draft:
Chen Li-Ming Zag Brightwing Sylvanas
vs
Thrall Muradin Lunara Li-Li Kael'thas
So let's look at zagara. Let's assume we have three other games were zag wasn't picked. We would expect popularity to be 25% (she was in 1/4 games). For our formula, a = 1, b = 40. So, Popularity = 1 / (40/10) = 1 / (4) = 25%. Perfect!
However, I am filtering by league. There were 6 plat players and 4 diamond players in my game. Let's assume that distribution was the same for all 4 games:
Platinum popularity: a still equals 1 b = 6 * 4 = 24 platinum players.
So Popularity of zag among platinum players = 1 / (24/10) = 1/2.4 = 41%.
For diamond players, it would be 0%. [no diamond players drafted her in this hypothetical.]
Now, let's cut things back even more. What if we only consider my game?
a still equals 1 b now equals 6.
So popularity of zag in that single game is 1 / (6/10) = 166%
That was the explanation I proposed to Ben, and he said it's why popularity can go above 100% when filtering by league. The way popularity is calculated gets minorly broken when people draft heroes in mixed-league format games. More specifically:
Any time a > b/10, it will go over 100%. I believe this means that players in that league are picking the hero more than players in surrounding leagues when they are in the same game. In the case of the numbers from this spreadsheet, platinum players value KT so much they steal KT picks from diamond players, and gold players value KT so much they steal picks from bronze players.
|
|
|
|