A suggestion to improve racial variety in perpetual leagues

[Smeborg]

Please note this is not a nerf to Claw, P-On or M-Blow. Their effect remains the same as before, where a player has only one of these skills (or only wants to use one of them).

Instead, the proposal is to nerf (partially, by no means completely) the following 4 skill combinations:

Claw + P-On
Claw + M-Blow
P-On + M-Blow
Claw + M-Blow + P-On

For an overview of the basic numbers, please see my other post.

The game is complex. To see what happens to SPP acquisition rates, Death and Retirement rates, development strategies, play styles and other relativities will only be discovered by testing. But I have a good feeling about the proposed rules change, and that feeling got warmer when I did the numbers

All the best.

[dines]

I have to say if stopping one combo - the ability to roll 7 or higher on armour dice with a re-roll - ends the effectiveness of Chaos and Nurgle teams they need a serious re-think. One wonders how strength teams without claw have coped... It of course does not ruin them, just makes them have to think a bit more. A non stacked combo means you can't combine claw and MB on the same role and can't use either on a re-roll. This is a step up on the previous suggestion of simply no modifers applying if you use piling on (you can of course under both ideas still use MB on injury rolls).

And simply removing piling on damages everybody but especially non mutating bash teams who seem the most hard done by honestly under the current rule set at the TV levels we are discussing.

For the stats people out there what would be the effect of bringing back a mutation that affected injury rolls? So say a +1 or +2 to injury. It is another mutation killer teams could pick, would it be better to get that or the current claw first? (I would have thought the current claw as altough its less cas it is more chance of at least stunning the opposition.)

I don't have any data, but as far as i remember almost everyone took "claw" (+2 AV roll) before "razor sharp claws" (+2 inj. roll) in LRB4 FUMBBL. I guess the difference in cas production isn't that great, but in game "claw" is better due to more stun+ as you say.

[DoubleSkulls]

Smeborg, apologies for taking this off topic, but I think I was discussing the point of the solution: will nerfing teams make them less/more popular, or is perception more important than reality?

DS - those stats by Koadah for you:
http://www.cmanu.pwp.blueyonder.co.uk/b ... stats.html

Thanks!

So looking at games where the weaker team is 2m+ and including mirrors.

Then the 3 most popular teams are Chaos, Nurgle and Chaos Dwarves. Between them they have well over 50% of all the games played. Chaos and Chaos Dwarf are the two most popular by team count too.

Chaos are at a smidgen under .550 which is the upper threshold for tier 1. The sheer volume of teams and games played for those races seems to show that we can probably safely discount distortion because of one or two exceptionally good coaches/teams.

In comparision the top 4 teams (all Elves) have fewer games played than Chaos Dwarves alone, and as such are much more likely to be distorted by small numbers of coaches/teams.

In terms of performance Orcs and Dwarves are both pathetically low - 17th & 18th with records just under .410 - below the acceptable range for tier 1 teams. They are still quite popular (4th & 5th) in spite of their poor performance.

-------------------------------

So, for FUMBBL black box, I think its pretty easy to say that a very small number of teams are very popular and this causes a lack of racial diversity. I cannot recollect anyone saying they think this is a bad idea and that low levels of diversity should not be encouraged.

The suggestion is a small nerf to a common facet of the 3 most popular teams (CLPOMB access) which has a relatively small direct effect on other teams, although may have an indirect effect by reducing attrition levels. It should be noted that Chaos & Nurgle average around 4.4 casualties per game, more than a whole casualty more than highest race without Claw access (Ogres & Dwarves). Discounting the 1 goblin game 7 of the top 10 casualty causing teams have Claw access.

This leads to a concern that other Av9 teams would benefit too much - notably Orcs & Dwarves - and greater diversity would not be achieved, but merely a replacement of the current popular races with new ones. However the performance of those teams in the identified bracket is substantially below expectations, so an increase in popularity due to improved performance is unlikely to occur. However it should be noted they are already the 4th and 5th most popular races, so it is possible that reduced attrition will increase their popularity, so that diversity is not increased. The recommended nerf does have a small overall effect on all the "bash" teams, so it may be that any benefit from reduced attrition is offset by reduced deadliness on their own part. The concern seems valid and should be closely monitored.

A further concern was raised with the impact to the viability of the races dependent on CLPOMB. However the performance of these is at the upper end of the acceptable limits, so the risk would appear to be small that they would drop off sharply enough to be below acceptable bounds. It was questioned whether the current performance was due to the performance of a single individual, however with over 100 teams playing Chaos so far, it seems unlikely that any distortion would be overly significant, and discounting the results of the best teams/coaches in every race may have a similar affect to the stats.

QED?

[Tourach]

Well put mr. double skulls

[dode74]

So, for FUMBBL black box, I think its pretty easy to say that a very small number of teams are very popular and this causes a lack of racial diversity.

If you're limiting to 2k+ TV I don't think you can say that at all without checking for a correlation between games played by a team overall and games played by a team at 2000+, otherwise you run the risk of there being another factor (such as large variations in TV by certain teams meaning that there are teams which don't get to stay at high TV for long and are therefore under-represented in the 2000+ stats.
That said, on looking for a correlation between games played overall and games played at 2000+, we find r=0.87, so a strong correlation.

Looking at the data for all TVs you find that the teams popularity by games played puts them in the order (all TVs, mirrors included):
Chaos
Chaos Dwarf
Nurgle
Dwarf
Orc
Chaos Pact
Necromantic
Norse
Dark Elf
Undead
Human
Skaven
Wood Elf
Lizardmen
Khemri
Amazon
Slann
High Elf
Underworld
Elf
Vampire
Ogre
Goblin
Halfling

When you check the correlation between games played and win% it comes out as r=0.52. With 24 teams there are 22 degrees of freedom, so anything over 0.404 (for 95% confidence interval) can be considered significant. HOWEVER, if you remove the bottom 4 teams (tiers 2 and 3, I think) so we can look at the successful teams (as you suggested), what we see is that r=0.08. We're now down to 20 teams, so 0.423 is the threshold for significance. What we can say is that there is no significant correlation between games played and win% overall for the tier 1-1.5 teams.

Looking at some of the other stats provided, we can see a significant correlation between games played and cas caused (r=0.61 for all teams, r=0.72 for T1-1.5) and a significant negative correlation between games played and cas taken (r=-0.73 for all teams, r=-0.62 for T1-1.5). One of the things we can see from that is that among the top tier teams causing cas would appear to be more important than taking them. If you look for a correlation between games played and cas difference (home cas - away cas) then there is a significant correlation there as well (r=0.81 for all teams, r=0.76 for T1-1.5).
To summarise:

Code: Select all

All games, mirrors included
Games played vs  All teams  T1-1.5
Win%             0.52       0.08
Cas caused       0.61       0.72
Cas taken       -0.73      -0.62
Cas difference   0.81       0.76

Now, to look at the 2000+ games (mirrors included) in the same way, the order of teams by games played is:
Chaos
Nurgle
Chaos Dwarf
Dwarf
Orc
Necromantic
Dark Elf
Khemri
Chaos Pact
Human
Undead
Wood Elf
Norse
Skaven
Vampire
Lizardmen
Slann
High Elf
Elf
Amazon
Underworld
Ogre
Goblin
(no flings at 2000+)

That's pretty similar to the all-TV list. The correlations are as follows:

Code: Select all

2000+ TV, mirrors included
Games played vs  All teams  T1-1.5
Win%             0.22       0.06
Cas caused       0.57       0.75
Cas taken       -0.33      -0.70
Cas difference   0.54       0.81

From both sets of data (2k+ and all TVs) it is clear that popularity (games played) correlates with both causing cas and not taking them, rather than the success of the team in terms of winning. It is therefore this which needs to be addressed if popularities are to be equalled.

Based on that I agree much of your statement above. That said, where you talk about Orcs and Dwarves you say "an increase in popularity due to improved performance is unlikely to occur", but I would argue that, as they are 4th and 5th in both lists they don't need to improve in popularity at all. If anything, the low AV teams need to increase in popularity without increasing in win%. Looking at what correlates with popularity what is required is some way of reducing the casualties to low AV teams without increasing their overall win%, and a reduction of casualties by high AV teams without reducing the casualties to them - i.e. a relative survivability nerf to higher AVs. Would your solution do that?

[DoubleSkulls]

Very nice... with Orcs & Dwarves my point was that they are already quite popular and we do not want them to become more popular. So the concern would be that reducing CLPOMB would encourage more Orcs & Dwarves, so instead of increasing the diversity, we'd just end up replacing Chaos & Nurgle with Orcs & Dwarves and the other teams wouldn't get played more anyway.

Its good to see that the stats seem to back up the theory that Black Box is most popular with people who don't want to take cas, but do want to dish them out - rather than people who want to win as much as possible.

[dode74]

Its good to see that the stats seem to back up the theory that Black Box is most popular with people who don't want to take cas, but do want to dish them out - rather than people who want to win as much as possible.

Yep. It also makes a case that perhaps the issue is not the rules but the league dynamic. I'll see if I can dig out some stats for cas for OCC to compare it with, but I suspect the lack of motivation to win in B changes things such that survival (not taking cas) and team building (which is aided by causing cas for SPP) become the drivers. It's not really possible to say that until we have some stats from league structures where winning is the motivator.

[Hitonagashi]

Its good to see that the stats seem to back up the theory that Black Box is most popular with people who don't want to take cas, but do want to dish them out - rather than people who want to win as much as possible.

That's a bit of a jump. The stats back up the theory that the more a team plays in the Black Box, the more it causes casualties. Given the other fact that the most popular teams are those that get more lethal with age, that's not entirely a surprise. From my experience, the majority of Chaos players in the Box get quite indignant if you tell them they are concentrating too little on the ball; at least for them, they believe this is the best way to win a game.

Nice work Dode.

My random suggestion: If Tackle is used to bring the target down, PO cannot be used. Kinda makes sense from a fluff perspective, and would drastically boost the survivability of elves.

[dode74]

The stats back up the theory that the more a team plays in the Black Box, the more it causes casualties.

True for 18 of the 24 teams (not for Amazons, DE, HE, Flings, Slann and Vamps). You can also say that the more a team plays in B the fewer casualties it takes - true for 22 of the 24 teams (not Gobs or Flings).

The stat you are talking about is games played per team vs cas caused, which comes out at 2000+TV as 0.472, so a significant correlation, but not as significant as the survivability stats. For all games the correlation is stronger at 0.69.

The fact remains, though, that the correlation between a T1 team's popularity (games played) and winning (win%) is low, whereas the correlation between popularity and causing/not taking cas is high. Chaos players can get as indignant as they want, but those are the facts.

[Hitonagashi]

Aye, I agree with you on the effect. What I was complaining about was the fact that Doubleskulls phrased it as they don't want to win. I'm saying they do, they just aren't very good at it...

[Hitonagashi]

*double post*

I would wager the only players that play high TV non hyperbash are strong players (as weaker players will get trashed)...so the lack of a correlation could be due to the fact that weaker elf players don't get the chance to play teams for a long number of games.

Is a correlation the same in reverse? So if long running teams cause more casualties per game, does that imply that teams that cause more casualties per game end up being played for longer periods of time?

[dode74]

I would wager the only players that play high TV non hyperbash are strong players (as weaker players will get trashed)...so the lack of a correlation could be due to the fact that weaker elf players don't get the chance to play teams for a long number of games.

Speculative
And I'm not sure which correlation you are referring to for elf teams?

Is a correlation the same in reverse? So if long running teams cause more casualties per game, does that imply that teams that cause more casualties per game end up being played for longer periods of time?

There is no indication cause and effect, it's just a correlation. We can't say whether it is the causing cas which means they play more games, or playing more games means that cause more cas.

[Jimmy Fantastic]

I read Doubleskulls post and more than a few times he refers to his proposed "small" nerf.
It is a MASSIVE nerf to Claw as it means no-one will ever take it because it would only take effect on 5/36 knockdowns against Av9 - narrow or what !?
PO may be taken as a late skill on a Tackle MB player but effectively this proposed nerf deletes PO and Claw as viable skills, Claw being the only skill that makes Chaos or Nurgle worth taking over Orcs.
Horrible horrible idea!
Maybe saying 2 skills could be combined could be an ok idea, Chaos would have Claw MB and other teams could have POMB and sometimes Chaos could use PO too.
Saying that ppl in Box don't pick races because they don't want to win is just ignorant and doesn't even deserve a response quite frankly.

[DoubleSkulls]

Aye, I agree with you on the effect. What I was complaining about was the fact that Doubleskulls phrased it as they don't want to win. I'm saying they do, they just aren't very good at it...

I didn't mean they don't want to win. Intuitively I'd expect people to take teams that maximises their chances of winning, and they aren't. They are picking teams based on toughness and lethality.

Also Dode was referring to a previous conversation (in a thread about halflings) where I pointed out the correlation between popularity and success is very weakly indicated at tournaments for all the teams below average. So small boosts in effectiveness of races in the bottom half probably won't have much affect on their popularity. Its a shame the NAF doesn't track casualty stats (although it could, most games don't have them entered), it would be interesting to see whether the correlation was a consistent one with tournaments - where presumably resurrection should make it less important.

[Hitonagashi]

I didn't mean they don't want to win. Intuitively I'd expect people to take teams that maximises their chances of winning, and they aren't. They are picking teams based on toughness and lethality.

It depends how you define winning.

The only non tournament Box stat I care properly about is the sprint: http://www.cmanu.pwp.blueyonder.co.uk/b ... printId=19
That stat is for Nov/Dec (it's for 2 months at a time, best run of 16 games over 1750, if you drop below 1750, games aren't counted till you go above it again). In that list, 10 of the top 20 were Chaos or Nurgle, and that's fairly common along all the past sprints; that Chaos/Nurgle/CD/Pact make up 50% of the top teams list.

In July/August, 15 out of the 20 top teams in the Box (by that ranking) came from the clawmbpo teams. Coaches look at that list, and see what results are possible with them.