FUMBBL :: Online Blood Bowl League

Hitonagashi

Joined: Apr 09, 2006

Posted: Oct 10, 2011 - 00:37

Garion wrote:

. But would anyone use it other than the odd fling coach? Probably not, but i may be wrong?

Just as a point, I would. I prefer to play 500 TV down with Lizardmen than at TV...I think they are much better at that range, even against massed claw.

[edit] at least, I'd like the option to Smile

. I wouldn't want it constantly, but I'd enjoy the choice of really having a blast.

Carnis

Joined: Feb 03, 2009

Posted: Oct 10, 2011 - 01:24

Lizzies are an exception to the rule though. Getting the 7th saurus star *really* benefits the team. A lot of the stars are like that, but most teams dont get as big a difference from it.

I sort of get dode's point that it may be too strict. But the other end of the stick really is, widening it to say 20% would make a really big difference in what is a "safe" TV to avoid mass clawpombs in B, while it would hardly have much effect in R. Except allows for more induced stars & more picking.

There's also the point of counterpicking. Like playing up 200-440k against a coach to get eldil + hubris every game etc. Strict range means no stars at all, I think on the whole the range could be wider for R. After all, you cant be forced to take any games if you dont want to in R.

Anyway if it were up to me I dont see why R could not have a wider range. More wizards + stars would be seen, especially at eg. 1500 vs 1800's the 15% seems a bit strict, really.

Tigga

Joined: Dec 21, 2007

Posted: Oct 10, 2011 - 01:33

What interests me is the fact that matching based on TV is seen to give "fair" games. If you're going for fair games match based on a team Elo value (+ racial mods) and there we have it. Every game is a challenge, you're not stuck playing a coach of CR 140 whose team have lost the last five when you're 165+ with a five game win streak, and each game you have a roughly evens outcomes. That sounds like a fun system to me.

Of course a pure team Elo system falls down on a number of counts; mainly due to the teams that don't necessarily do all that they can to win. If you win your first couple of games then get tied against a 2300 TV CPOMB monster team that plays to maim things aren't going to be fun.

For this reason I think the best system would be to have a system where team Elo rank and number of games played are combined. It'll match based on Elo, but with the restriction that your opponent must have played 50%-150% the number of games you have (so 10 gamer could play 5-15 gamers). Once you're over 30 games this restriction can be lifted (ie. team is fully developed). Numbers are fairly arbitary and could be altered... just an example.

There are two main reasons I don't play Cyanide MM. One is because I do not enjoy the TV min-maxing that comes about by matching by TV (also a reason I don't play [B]), the other is that a fair proportion of games you start, watch your opponent's first couple of turns and think "I'm going to have to roll really badly to not win this". Even if that happens only 1/4 games it's enough to annoy me as I feel I'm wasting my time with those games.

harvestmouse

Joined: May 13, 2007

Posted: Oct 10, 2011 - 01:54

Carnis wrote:

Anyway if it were up to me I dont see why R could not have a wider range. More wizards + stars would be seen, especially at eg. 1500 vs 1800's the 15% seems a bit strict, really.

It simply would do more damage than good. There are far too many people who would try to take advantage of a more lenient system, this is online gaming afterall. Once a % take advantage, another % do the same to maintain a status quo.

The ones who suffer are the newer coaches, who don't know better and the coaches who aren't prepared to barter for a game or don't have time to barter for a game.

In the end we lose a large % of these from the ranked environment.

harvestmouse

Joined: May 13, 2007

Posted: Oct 10, 2011 - 02:01

Tigga wrote:

What interests me is the fact that matching based on TV is seen to give "fair" games. If you're going for fair games match based on a team Elo value (+ racial mods) and there we have it. Every game is a challenge, you're not stuck playing a coach of CR 140 whose team have lost the last five when you're 165+ with a five game win streak, and each game you have a roughly evens outcomes. That sounds like a fun system to me.

Of course a pure team Elo system falls down on a number of counts; mainly due to the teams that don't necessarily do all that they can to win. If you win your first couple of games then get tied against a 2300 TV CPOMB monster team that plays to maim things aren't going to be fun.

For this reason I think the best system would be to have a system where team Elo rank and number of games played are combined. It'll match based on Elo, but with the restriction that your opponent must have played 50%-150% the number of games you have (so 10 gamer could play 5-15 gamers). Once you're over 30 games this restriction can be lifted (ie. team is fully developed). Numbers are fairly arbitary and could be altered... just an example.
.

I'm not 100% sure what your point is, but I do agree that TV matching does cause some problems.

Fumbbl does have a sort of elo reating CR and BWR? for box. However this isn't used at all for match making. The idea wasn't popular for box, and for ranked would simply limit the amount of matches that can be played. Also both divisions are heavily based on 'coaching skill'. Fumbbl's elo system does penalise coaches for playing down, however you do have the problem of how much do coaches care about CR compared to getting a win.

Wallace

Joined: May 26, 2004

Posted: Oct 10, 2011 - 02:02

Personally, I would prefer a wider limit than 15% in Box. I'd be happy with no limit at all! However, that's just a personal preference. If I was suddenly put in charge of making the rules around here I'd leave it like it is. For the mainstream open divisions the restrictions make sense and are probably close enough to optimal to leave alone.

For those who want a different approach there are open groups in League that exist or could be made to exist to satisfy more niche desires.

Tigga

Joined: Dec 21, 2007

Posted: Oct 10, 2011 - 02:57

harvestmouse wrote:

However this isn't used at all for match making. The idea wasn't popular for box

It's popular with me though!

I guess I've been slightly spoilt by online RTS games. These days most of them rank each player and then match you up against somebody of similar rank. This means that most games are close matches. I am of the opinion that FFB matchmaking systems should work in a similar way; with a small concession to the team building process limiting games with a youngish team to other youngish teams. We'd get a lot more close games like this. Bad coaches would win more (because they play against other bad coaches) and good coaches would win less (because they play against other good coaches). All coaches face challenging games which will both test them and give them a chance. I propose doing it based on a team ranking, but you could (although I think it'd be worse) do it on coach ranking.

No real issue with the way [R] is handled. To be honest I don't really see much reason to put many rules on a league where you can pick and chose your opponents.

pythrr

Joined: Mar 07, 2006

Posted: Oct 10, 2011 - 05:09

this rule is required to stop another German picking scandal.

just saying.

_________________

KhorneliusPraxx

Joined: Jul 28, 2005

Posted: Oct 10, 2011 - 05:22

with the inducements the %15 rule is unnecessary...there, I said it!

_________________
Hopefully my pulsating avatar will remind you to always take +1 Strength...ALWAYS!

pythrr

Joined: Mar 07, 2006

Posted: Oct 10, 2011 - 06:12

inducements do not counter TV difference entirely, and therefore picking will resume.

look, i said it too

_________________

Wallace

Joined: May 26, 2004

Posted: Oct 10, 2011 - 06:19

As I said above, I'm not in favour of changing the rules in the Divisions, but at the risk of some shameless self promotion, some people reading this thread might be interested in this

Smeesh

Joined: Oct 29, 2005

Posted: Oct 10, 2011 - 08:13

Well Schduling by win rate would be really nice. Regardless of TV. But that will cause a problem especially at the beginning. The same players will play each other more often in a row. So there sould be a bias against plaing the same matches for some short period.

Greetings

Smeesh

Last edited by Smeesh on %b %10, %2011 - %08:%Oct; edited 1 time in total

dode74

Joined: Aug 14, 2009

Posted: Oct 10, 2011 - 08:15

@Woodstock

Quote:

So start with the basics, the skills and the lack of traits. Fix the cause, not the symptoms.

First, nobody has shown what is a symptom and what is a cause. Second, as I said before, this is the ruleset we have to work with. Until LRB 7 (or CRP 2 or whatever) happens, CRP is a constraint we have to work with.

Quote:

It shows lack of testing and a wrong mindset. You buff one skill, and nerf its counter into the ground. You make a passive skill counter an active skill. Fend is not like tackle that works both ways.
The only inducements worth are wizard and chainsaw, and both of them still rely on good dice.

Testing was carried out in the Vault. I don't have access to that data but I understand it worked. Personally I don't see an issue with CPOMB, but that's a whole different matter.
Inducements are worthwhile - they do the job they are designed to, as I stated before using data from the box.

Quote:

The matchmaker was based on TR/TS yes, but that should not really matter as that as well was limited back then, nor was it a secret of how things worked. Galak even claims that 'LRB 5.0/6.0 takes what FUMBBL had and made it even more balanced.". How is that not saying LRB6 should be perfect for FUMBBL?

He (and Jervis in his quote) was largely talking about how the inducement system made things work for a perpetual environment. He also says he is a big fan of house rules. I am suggesting that a house rule is needed, but that the current one can be improved upon.

@harvestmouse
Regarding your page 2 post:
I'll come to TV as a measure of team capability at the end.
Currently it is still possible to sit like your predator fish, particularly as a low-TV dwarf team. The same happens on Cyanide for the same reasons.

Quote:

If this isn't working then there is a problem with the official team evaluation,

That, or our understanding of it. Bear in mind that much was sacrificed on the altar of making the game uncomplicated.
Regarding your page 3 posts:

Quote:

It simply would do more damage than good. There are far too many people who would try to take advantage of a more lenient system, this is online gaming afterall.

Again, I'm not suggesting no limitations, merely that this one may not be the right one.

@Garion

Quote:

and TS was essentially the same as TV with a few nobs on, and i dont really think you answered my post tbh.

As I said above, I'll come to TV vs TS in a bit. I did miss the last part of your post above:

Quote:

Open leagues are where people pick up quick games. I personally want fair ones in ranked and box, and not to be a 15% disadvantage if his figure is correct.

The key point is "if his figure is correct". I don't think that TV is an accurate measure of team capability, as I hope to show at the endd.

@VoodooMike

Quote:

Wouldn't everyone's goal when creating a competitive game be to minimize the effects of everything but player skill?

Not necessarily. The most popular sports and games have a luck factor in them. This element of the unknown adds to the (subjective) excitement of the game. Otherwise we'd all be playing synchronous chess Smile

Quote:

What you seem to be saying is that inducements are a deeply important aspect of Blood Bowl and that people miss out by not using them more.. and that's where the real difference of opinion may lie.

That is exactly what I am saying, and it is backed up by Galak's statement quoted in the OP of this thread.

Quote:

Any time you say something is "boring" you're really just using "because I think it is, so it is" as your reasoning. If you find min-maxing boring then by all means, don't do it - but you finding it boring doesn't mean it is objectively boring.

It is entirely subjective, as you say. It does appear that the mood here is such that min-maxing appears to be considered boring, although I may have entirely misread that and look forward to hearing from those who think that min-maxing adds to the game.

Quote:

Obviously plenty of people disagree or they wouldn't be doing it.

I'd suggest that people play here despite it because FUMBBL is the best available option. Best available and best possible aren't the same though, and as good as FUMBBL is I think this is an area which can be improved upon.

@Hitonigashi

Quote:

Dode, I think it's quite simple really....it is a houserule.

More than happy to bow down to CRP page 24 if I am told to. Surely questioning the intent of such rules is allowed though?

Quote:

The judgement of Christer is that there are more fair games produced by enforcing a 15% limit than there are by allowing it not to.

I'm sure that is the case. I'm not advocating no house rule though.

Quote:

There could be a more complicated formula, but crucially, R is a division that you propose games against each other. In order to do that, you have to be able to tell at a glance whether your team is legally allowed to play each other. FUMBBL is run from all Christers hard work, and as a result, any more complicated feature would require a new set of interfaces and a lot more work on his behalf. 15% is a very easy limit to comprehend, and allows enough inducements without being unbalanced.

TS was FAR more complicated, and it worked well. I'm not saying that Christer should change things, merely debating whether the current system's effects are all intended. As I asked before, did you guys really intend for 95% of your games to take place within 150TV of each other, for over 50% of your games to be so closely matched by TV that even Babes can't be taken?

@pythrr

Quote:

inducements do not counter TV difference entirely, and therefore picking will resume.

Nor are they intended to, but in [B ] the underdog win% is meaning at 43% at a 230TV difference. Even at that "high" a difference the game certainly still seems competitive.

TV as a measure of team capability
Lots of people are arguing that TV is a measure of team capability (I'm going to call that TC to prevent confusion with TS as FUMBBL measured it). I think it's pretty easy to show that this is not the case.

First, different skills on the same player:
Black Orc Blocker 1: Block, MB, Guard, Tackle - 160TV
Black Orc Blocker 2: Thick skull, Strong Arm, Sure Hands, Shadowing - 160TV
Quite simply BOB 1 is likely to be more useful than BOB 2. I'm not saying BOB 2 is useless - there will be situations where his particular mix of skills saves the day, but he is less likely to be as useful as often as BOB 1.
Many people counter this with "you can't take bad coaching into account". My contention is that you could label sub-optimal (TV-wise) skill pick as "bad coaching", but that doesn't change the fact that the TVs are still the same and do not accurately reflect the capability of the player. "Good coaching" in this case would actually constitute min-maxing.

Second, the same skill on different players:
Shadowing on a Gutter Runner or Necro Werewolf - 20TV
Shadowing on a Dwarf Lineman - 20TV
Same cost, but the skill is massively more likely to work on the former players. Again, there may be cases where the dwarf uses the skill effectively, but that is far less likely than for the former players. The same skill creates a different level of capability dependant on the player it is taken on.

Third, capability of teams at different TVs:
Chaos vs Amazons at 1000TV
Chaos vs Amazons at 2400TV
I know which team I'd rather be in each case, and that's because of TV efficiency (the opposite of bloat).
Zons don't need many more skills to be hugely effective, so above a certain number of skills extras will tend to have diminishing returns. That's not to say that high TV Zons are ineffective, and I am sure that there are some players who make them work well, but the tendency is for them to become inefficient as they take unnecessary extra skills.
Chaos, on the other hand, start with no skills, have generally good stats (certainly nothing below average), and have a wide range of very effectively combining skills to choose from. As such they don't suffer from that same "diminishing returns" effect and can keep adding skills to players without the same inefficiency (or bloat) creeping in.

Finally, TV vs TS
Garion stated earlier:

Quote:

TS was essentially the same as TV with a few nobs on

I don't think it is, and I took a glance at the [B ] data to see if that was the case, since both TV and TS are listed there. Very quickly I found a 1000TV goblin team with a TS of 172, a 1240TV Nurgle team with a TS of 187, a 2030TV Dwarf team with a TS of 246, and a 1200TV goblin team with a TS of 97. From just that glance, as well as a look at how TS is actually calculated, I don't think your description is fair at all.

Matching
So, given the above, I don't think that anyone can reasonably say that TV is a fair figure for TC. It's closer at low TV, certainly, because those are (and have been) easily adjusted through playtesting, but it isn't possible to say that matching by TV gives a "fair" game, if fair is defined as "as close to a 50:50 chance as possible for each team".

Now, I'm not saying that 15% is a bad figure. It's not. But the vast majority of games don't take place at even that much difference. Below is a graph showing all [B ] games to 30 Sep (~57k games) and the TV differences involved (x-axis) as a percentage of all games played (y-axis):

The black line is at 150TV - 15% at 1000TV. Only 1.5% of games take place at such a difference, with <5% of games taking place above that. My suggestion would be to smooth this out - increase the mode TV difference from the extremely TV-even matchups (which I think have shown are not actually even due to deficiencies in TV as a measure of TC) you see there to something which allows for more use of inducements, more flexibility in team builds, and more variety (which is still subjectively a good thing).

Sorry for the Wall of Text, but you left me a lot to answer while I was sleeping Smile

Last edited by dode74 on %b %10, %2011 - %08:%Oct; edited 1 time in total

RandomOracle

Joined: Jan 11, 2004

Posted: Oct 10, 2011 - 08:20

dode74 wrote:

Garion stated earlier:

Quote:

TS was essentially the same as TV with a few nobs on

I don't think it is, and I took a glance at the [B ] data to see if that was the case, since both TV and TS are listed there. Very quickly I found a 1000TV goblin team with a TS of 172, a 1240TV Nurgle team with a TS of 187, a 2030TV Dwarf team with a TS of 246, and a 1200TV goblin team with a TS of 97. From just that glance, as well as a look at how TS is actually calculated, I don't think your description is fair at all.

TS is bugged with regards to LRB 6. For example, splitting foul appearance into two skills now means the skill is essentially counted twice. Thus, you can't really see how TS worked by looking at teams in the new rules.

dode74

Joined: Aug 14, 2009

Posted: Oct 10, 2011 - 08:27

True, but the fact that some are above and some are below what they would otherwise be suggests a weak correlation. That, and looking at how TS is calculated, which in no way resembles TV or TR.