27 coaches online • Server time: 13:19
Forum Chat
Log in
Recent Forum Topics goto Post Secret League Old Wo...goto Post Creating a custom to...goto Post ramchop takes on the...
SearchSearch 
Post new topic   Reply to topic
View previous topic Log in to check your private messages View next topic
Poll
Is CLAWPOMB really a problem?
Yes, absolutley
55%
 55%  [ 464 ]
No, Chaos Dwarfs Disagree
20%
 20%  [ 174 ]
Still Haven't Decided
8%
 8%  [ 75 ]
Pie!
15%
 15%  [ 127 ]
Total Votes : 840


xnoelx



Joined: Jun 05, 2012

Post   Posted: May 04, 2016 - 09:51 Reply with quote Back to top

You may want to add a little more explanation, tussock. DElves +4.0% for example. Does that mean DElves at 1800 TV or over win 4% more games than a rookie team, or than teams under 1800 TV? Is 1800 TV even the comparison?

NB: I think you're right, I'd just like that to be clear.

_________________
Image Nerf Ball 2014
bghandras



Joined: Feb 06, 2011

Post   Posted: May 04, 2016 - 10:41 Reply with quote Back to top

@tussock
Did you mean that Nurgle win 60%-65% over 2000 TV?

_________________
Image
plasmoid



Joined: Nov 03, 2009

Post   Posted: May 04, 2016 - 14:28 Reply with quote Back to top

Hi Guys,
while developing NTBB2014+ I worked on the Black Box stats (pre-scheduler fix).
If you want to take a look, you can see them here:
http://www.plasmoids.dk/bbowl/NTBB2014x.htm

The relevant table is at the very bottom of the page.
I should note that when only 1 number is present in a cell, then it is just the mean, which tells is very little. It is just there for reference.
When a cell spans multiple columns, 2 numbers are listed. This is not just the mean, but the confidence interval (95 CI). These are the relevant numbers. And I've done the math for stretches where peformance was suspiciously high or low.

To me, the most problematic thing about these stats is that the way that balance was defined, the relevant meta is just "whatever games are played". Since around 30% of the games played at high TV are either Chaos or Nurgle, it stands to reason that they drag Down eachothers stats.

The way I see it, the most interesting points that can be deduced about TV1700+ play is that:
a) all bashy teams crash performancewize. Except Nurgle and Chaos.
b) Elfs (and to some extent Skaven) are doing very well.

One possible explanation is that CPOMB & MM creates a meta, where survival is everything. By extension, people pick survival skills rather than anti-elf skills, and as a result, AG4 blodgers thrive.
The non-Claw Bashy teams then crash performancewize, because they don't stand up very well to the Claw teams, and without anti-elf skills they don't stand up very well to the elves either.

Cheers
Martin

Oh - PS: According to the "official" BBRC definition of lifetime performance, then splitting up stats the way I have done is totally irrelevant.
thoralf



Joined: Mar 06, 2008

Post   Posted: May 04, 2016 - 16:05 Reply with quote Back to top

MattDakka wrote:
50% is the perfect theoric balance, trying to get as close as possible to it is a reasonable goal.


We could push this even further: a perfect balance is a configuration where all (or most, or the most) games between two teams lead to OT and then to a coin toss. We could extend this to BB games with an infinite board, infinite menz, etc. Under that kind of definition, we don't even know if Chess is balanced. All we know is that if an ideal game of Chess doesn't lead to a draw, either White or Black wins.
bghandras



Joined: Feb 06, 2011

Post   Posted: May 04, 2016 - 16:37 Reply with quote Back to top

I think really smart chess players also know that if chess is not balanced, then white wins. At least that is what one of them told me in the past.

_________________
Image
zakatan



Joined: May 17, 2008

Post   Posted: May 04, 2016 - 18:02 Reply with quote Back to top

plasmoid wrote:
The way I see it, the most interesting points that can be deduced about TV1700+ play is that:
a) all bashy teams crash performancewize. Except Nurgle and Chaos.
b) Elfs (and to some extent Skaven) are doing very well.


That's hardly surprising really. In a see full of clawpomb sharks, elves triumph and die. Elven trips to 1800+ TV are usually successful and short, and the general population is quite sturdy and bloodthirsty, so it's just natural that they're equipped to bang their heads against each other more than to counter the occasional elf. In any case, it's not survival skills what they equip.

You should really join the box for a couple hundred games. There are things you can't feel from raw data.

_________________
Image
thoralf



Joined: Mar 06, 2008

Post   Posted: May 04, 2016 - 18:26 Reply with quote Back to top

bghandras wrote:
I think really smart chess players also know that if chess is not balanced, then white wins. At least that is what one of them told me in the past.

Of course, and the Correspondence Chess data corroborates this. However, there's still no logico-mathematical proof (at least to my knowledge) that the starting position in Chess ain't a mutual zugzwang.

My indirect point was to show that searching for balance could lead to some degenerate results.

Martin wrote:
The way I see it, the most interesting points that can be deduced about TV1700+ play is that:
a) all bashy teams crash performancewize. Except Nurgle and Chaos.
b) Elfs (and to some extent Skaven) are doing very well.

To have a really interesting meta, we'd need a third archetype. Lizardmen may fill that gap, but one race ain't enough. We'd need the 3AG teams too.

When I played MTG, the abstract meta was a combo-control-blitz triumvirate. From seasons to seasons, a different archetype would dominate. This rotation was related to the card pool more than randomness.

It may not be possible to recreate the same dynamic, but to me, that would be the most interesting balance to seek.


Last edited by thoralf on May 04, 2016 - 23:01; edited 1 time in total
bghandras



Joined: Feb 06, 2011

Post   Posted: May 04, 2016 - 19:31 Reply with quote Back to top

thoralf wrote:
To have a really interesting meta, we'd need a third archetype


This is the winning post of the thread. I fully agree. (minimum) 3 archtypes are needed for any resemblance of balance.

_________________
Image
Wreckage



Joined: Aug 15, 2004

Post   Posted: May 04, 2016 - 19:57 Reply with quote Back to top

I can't believe Dode, caused an actually productive debate when he left. Although he would probably remark that you're doing it completely wrong and your entire approach is biased in multiple ways. Is there a name for that? Somebody who initiates interesting conversations that actually can only develop without him? Troll perhaps? Dunno. Smile
Endzone



Joined: Apr 01, 2008

Post   Posted: May 04, 2016 - 20:04 Reply with quote Back to top

So basically we don't have a working third archetype because clawpomb makes a joke of AV9 teams. Sounds to me like a balancing defensive strength skill is needed.
licker



Joined: Jul 10, 2009

Post   Posted: May 04, 2016 - 20:05 Reply with quote Back to top

bghandras wrote:
thoralf wrote:
To have a really interesting meta, we'd need a third archetype


This is the winning post of the thread. I fully agree. (minimum) 3 archtypes are needed for any resemblance of balance.


Aren't there three archtypes (at least)?

Problem is they are not evenly matched at different TV ranges.

But as opposed to a CCG BB cannot really (and likely should not really) have expansions where the archtypes get reset. You can do this withing private leagues already anyway, but it doesn't really make sense to me to do this generally.

The other issue generally with applying a R/P/S system to BB is that then the bad matchups are even worse, but the games aren't done in 10 minutes as they are for CCGs. Unless you enjoy concessions.

Really to me it's more a question of how individual coaches choose to build their teams rather than just how certain skills or rosters define the matchups.

High TV M teams can utterly destroy elf/AG teams. *IF* they are built to do that. Most of them are built with the generic cpomb all menz mentality though, and those builds suck vs. high TV elfs compared to what a balanced, or anti-elf built M team does.

Orcs and Dwarfs also nominally build to throttle elfs (with lots of tackle and mighty blow and guard), but they are useless against the cpomb spammers, though they don't really have an alternative build to counter cpomb spam anyway.

I guess I don't understand what concept you may have in advocating 'a third archtype' though.

I also think BB is basically doing fine as it is though. What's not doing fine are large open divisions with either MM or challenges, but that's not so much an issue of the rule set as it is of the mentality of the people involved. Though I do think some roster tweaks would improve the game across most metas.
ArrestedDevelopment



Joined: Sep 14, 2015

Post   Posted: May 04, 2016 - 20:15 Reply with quote Back to top

The following is subjective opinion:

I actually think using box as a medium for looking at team win rate% (lifetime or otherwise) is a rather cack-handed way of doing things. Because while it may be a blind match-making system, it is also subject to its own biases (mostly generated by the user base).

If we revisit bghandras' example of the chaos team that loses its first X games and then wins its last X games before being mng'd/permed to retirement - in box that team probably retires for most users after such a miserable start.

The above is true of any race for some people. But it should also be noted that with agility teams especially, the only people who tend to play them to (and at) high TV range tend to be very good coaches (generalisation and anecdotal opinion, easily dismissed) and thus even without the "blanket blodge facing tackle-less cpomb" scenario (which incidentally, is not actually as common as people seem to think) - there are a fair few wins for high TV elf teams against high tv bash (esp. nurgle/chaos) where the difference in coaching skill is marked. Now you might say "so what?", but the ability of the people playing the races also creates a bias point in your data - if the only people who play Elves in box also tend to be very good at the game in general, then the Elvish win% is subject to error immediately.

There's also draw "bias" - anecdotally, I have several teams who have faced absolutely nothing but bash teams in box, and others who have had a relatively bash-free ride. What's the difference? Mostly activation time. This is mostly a product of the new scheduler I'd imagine, but under current circumstances, living in Europe is going to be a benefit if you want to play as an agility team or non-claw bash team.

Basically, I understand why people look at box data for conclusions about team win% and so on, but I think it is deeply, deeply flawed to treat what it shows as sacrosanct, unless you want to talk about extrapolating the Stanford prison experiment to online gaming Wink

_________________
Image
koadah



Joined: Mar 30, 2005

Post   Posted: May 04, 2016 - 20:28 Reply with quote Back to top

Wreckage wrote:
I can't believe Dode, caused an actually productive debate when he left. Although he would probably remark that you're doing it completely wrong and your entire approach is biased in multiple ways. Is there a name for that? Somebody who initiates interesting conversations that actually can only develop without him? Troll perhaps? Dunno. Smile


Once you cut out the pedantic quibbling over definitions, who owns the IP etc you might actually be able to discuss something worthwhile.

_________________
Image
O[L]C 2016 Swiss! - April ---- All Stars - Anniversary Bowl - Teams of Stars - 13th March
bghandras



Joined: Feb 06, 2011

Post   Posted: May 04, 2016 - 21:59 Reply with quote Back to top

The "theory" was about high TV, where there are 2 directions to go
- hard cpomb bash
- agile dash all in
Game theory suggests that to reach an equilibrium in a metagame, at least 3 archtypes are needed (unless it is just a coinflip).

_________________
Image
licker



Joined: Jul 10, 2009

Post   Posted: May 04, 2016 - 22:05 Reply with quote Back to top

bghandras wrote:
The "theory" was about high TV, where there are 2 directions to go
- hard cpomb bash
- agile dash all in
Game theory suggests that to reach an equilibrium in a metagame, at least 3 archtypes are needed (unless it is just a coinflip).


Well then I still maintain that those are not the only 2 archtypes present.

The issue feels to be more one of racial diversity (which I am all for) rather than roster design, which can still be accomplished with only 'two' races being used.
Display posts from previous:     
 Jump to:   
All times are GMT + 1 Hour
Post new topic   Reply to topic
View previous topic Log in to check your private messages View next topic