LRB6 stats pulled from fumbbl's box div:

[Carnis]

Still early, but this is how it's looking. Interesting in it imo is how badly dwarves do at TV 1800+. But this is probably due to coach bias & the abundance of MB/Claw/PO, which the dwarves are not fitted to counter at all.

This is what I'm seeing

All Games

http://www.cmanu.pwp.blueyonder.co.uk/b ... -02-06.gif

Where at either team is 1800+

http://www.cmanu.pwp.blueyonder.co.uk/b ... -02-06.gif

[Smeborg]

Hi Carnis -

For those of us who are not FUMBBLERS , please could you give us some background on the "box division"? Thanks.

In general terms, the dominance of AG teams at TV1800+ does not surprise me. It is consistent with what I have experienced in leagues in 5th and 6th Editions.

The fall of Dwarfs and Norse is remarkable (base stats to TV1,800+), as is the rise of Underworld. The fall of Elfs leads me to question the stats (because I have found them to be one of the very best teams). Are you able to put any figures on your view of "coach bias"?

All the best.

[James_Probert]

The 'Box' Division, is a Division, (well, Duh) in which the only way to get a game is to put your teams into a 'BlackBox' at certain activation time periods. Then, once the activation period is up, there is a sheduler, that goes through all the teams that have been activated, and tries to find the best possible set of games, such that as many people who have activated as possible, get as 'good' (which means close in TV) a game as possible. There is also a limit on the gap between teams. I think there's some 'racial factors' in the sheduling system as well.


tl;dr: BlackBox, put teams in, get match out, play match, repeat.

[Greyhound]

so the matchmaking is actually biased by selecting certain match ups?
That's actually disruptive (from a stat point of view) since it would artificially pair teams more with some races often than others?
also if it is matching TV, then we don't have a typical attrition system where the skavens use inducements and the orcs and dwarfs face constantly the mighty wizard. right? or did I miss something.

[duckwing]

Not to mention that goblins and halflings don't get to use their ace cheap inducements.

[James_Probert]

well, i'll answer point 3 first: No one has to face "the mighty wizard" atm, as unfortunately, the wizard is one of the few inducements still no included in the FFB client.

back to points 1+2: I can't remember how it works, at the moment it may simply be raw TV, but back in LRB4 Box, the sheduler had a (very small) weighting for racial factors ( https://fumbbl.com/FUMBBL.php?page=stats&op=rankings ) which were based on the results of (I think) all 'competitive' games played on fumbbl. so for example, you may well find your amazons playing dwarves, but you'd be more likely than not to have a slight advantage, according to conventional measures...

There are some very good (and long) posts by Christer on the fumbbl forums about exactly how the sheduler works (or, atleast, used to)

[Joemanji]

A small weighting for racial matchups is not going to be a problem over a large enough sample. Unfortunately the sample (particularly for TV180+) isn't yet large enough to make these stats 'gospel'. Can't imagine how Norse can be the second worse team for example, or Underworld the 5th.

[spubbbba]

Don't forget that some of these teams were transferred and converted from FUMBBL's old LRB4 Blackbox division so that would have an impact as well.

I'd say waiting a good 6 months would give more reliable data. Even better will be when the [L]eague division transfers and we have data from those that start with all rookie crp teams.

That will allow comparison of how the teams do with big TV differences though I'm not sure how you'd show the difference between a fling team being 200TV underdog, equal TV and 150TV overdog against various races.

[Purplegoo]

I'd say waiting a good 6 months would give more reliable data.

Indeed. Super-early days.

[Ullis]

The 1800+ figures have such low numbers of matches for certain teams that at least some of them are probably the work of a single coach. At least the Underworld at 1800+ suggests this.

But I'm amazed that Slann have such a low winning percentage over all games. It's probably due to them being new for many coaches. At least I've seen some very strange builds (such as low starting rerolls and lots of blitzers but no catchers) over on Fumbbl.

[Ullis]

And yes, the matches are randomly based on TV. Based on my own experience, the number of coaches is high enough to give very even matches. I've played 13 games on the Blackbox so far and I've seen inducements (one set of babes) once in my own games. On the matches I've spectated I remember seeing one extra training.

[GalakStarscraper]

Happy to have more data. But as I said with Cyanide's MM ... the rules were not meant to be used on leagues that did only TV matchings for match-ups. Its not organic and it ignores that certain inducements were added in to help weed down those higher TV teams.

That doesn't mean its not fun and a great way to play BB ... but its not a statement on balance in my opinion.

Tom

[mubo]

Surprised to see vamps so low, in my experience they can be pretty competitive with a good coach. Mind you, I guess they are terrible in the hands of a poor coach, so maybe it balances.

Not sure that we need inducements to estimate team strengths, surely inducements are a device to even up a game where teams are mismatched, rather than an integral part of the game? Don't get me wrong, I think they add loads to the game, but if you have a large enough pool they're not needed. This is why I'm not so keen on flings being v. v. weak and propped up to weak with good value inducements (i.e. the chef). If they were able to play 100k below consistently, I'd expect that 25% to creep up a bit.

I'd love to see some stats adjusted by coach CR, (or BBR). I understand there's not enough data to do this yet, but I'd maybe expect to see that a team like vamps has a very wide range, whereas results of orcs (say) are less dependent on coach ability. Maybe- just a pet theory.

[Smeborg]

So as I understand it, the stats will be biased towards teams that perform well against teams with similar TV. Since this is not a criterion used to establish game balance in 5th and 6th Eds (which aim rather to restore balance when TVs differ), I am inclined to agree with Galak that the stats are interesting, but not particularly useful regarding balance.

I am not surprised at the high performance of Underworld at TV1800+. Underworld are a team that get (relatively) better with development. They are also cheap, so at TV1800 would likely have 50 points (or more) of skills, with a full roster. 3 can-openers (Claw/M-Blow), high-performing Throwers and multiple 2-Heads Gobbos would be standard at this TV. Because of low AV and Stunty, they dislike damage-causing inducements being used against them (Saws, Wizard, Bribes).

All the best.

[Overhamsteren]

Also tackle seems sparse.