An ABLE System: toward league fairness....

[Hache]

Rank determines who you can challenge... that's all....

TV determines upper/underdog (same as rules)

Meaning...
Human is 7th in the ladder, TV 120.
Halfling is 1st in the ladder, TV92.
Elf is 6th in the ladder, TV150.

If human challenge Halfling and beat them, they gain less than if they beat Elf ?



I strongly think that all the math to calculate the ladder level should not take TV into account.
TV is a way to gauge the strength of a roster.
Ladder level is a way to gauge the strength of a coach.

You could even allow a coach to keep the same ladder level when using another team.

[MrBister]

The wiki for ELO is a bit over the top IMO.

Suffice to say that ELO ranking is used for almost all major ranking systems (perpetual you can say) in the world where two opponents/teams play a game. It was first developed for chess rankings.

The table for how many points you win in MBBL is based on ELO but uses ranges for whole points instead of the ELO formula. (Please correct me if I'm wrong, Galak.) NAF also uses ELO ranking.

The winner wins the same amount of points as the loser lose. A draw gives half the amount of points compared to a win. If both participants has the same rank and they draw, noone gets any points. (They are equally good, and should be if they have the same rank.)

How much the winner wins depends on if he's expected to win. The larger the gap in rank between the opponents the more likelier the highest ranked wins. And that makes him win less points, if he wins. If the underdog wins (not talking TV here) he wins more than he would if he met someone at the same rank. Since actually proven that he's proven to be better than his rank gives credit for, by beating someone higher in rank.

I hope this clears up ELO a little bit at least.

[Digger Goreman]

Rank determines who you can challenge... that's all....

TV determines upper/underdog (same as rules)

Meaning...
Human is 7th in the ladder, TV 120.
Halfling is 1st in the ladder, TV92.
Elf is 6th in the ladder, TV150.

If human challenge Halfling and beat them, they gain less than if they beat Elf ?

Unlikely to happen like that, but I DO get your point.... Time for re-thinking....

MrBister: I'll try to find the explanation for ELO on the MBBL.... The way you explain it sounds more reasonable....

[Digger Goreman]

Ok, I've found the url: http://www.midgardbb.com/MBBL2/LeaguePoints.html and am initially impressed! Thanks for pointing me in that direction!

I think, with the proviso of challenges only coming from below and not for more than 200k without approval (to keep someone from putting together a crappy little team for their friend(s) to beat up), we just might have the "Do All" system....

[Digger Goreman]

So the league "rules" would look something like:

ARKHAM BLOODBOWL LEAGUE EXPERIENCE
(A.B.L.E.)

Bloodbowl is a miniatures game of high fantasy football played on a game board.

“ABLE” is designed as a “bottom-up” environment, with lesser valued teams having more control over the schedule and flow of the “ladder” system.

Rules set: LRB 5 + Experimental Rules Review 2008; free download at:
http://graveyard-gothika.webs.com/blood ... trules.htm

Miniatures: ANY figure that may reasonably identify the position… including metal, plastic, Games Workshop and non-Games Workshop minis….

Schedule: technically, there is none. The ladder and challenge system allows for any number of games… play is pretty much at your leisure.

Team Entry into the ladder: All teams enter the ladder at 1,000,000 or less, i.e., teams start new. A team starts in the ladder at "100" rank value.

Playing a game, and issuing Challenges: you must be lower in rank AND not played the challenged team twice in a row. Challenging a team that is
over 200,000 in Team Value above you requires permission of the Commissioner. You may not play against one of your own teams.

Ladder Movement, Computing Rank, Win/Tie/Loss: See the "Strength of Schedule System", below.

Multiple Teams and Forced “Retirement”: Coaches are encouraged to play more than one team in the Ladder League. However, a team will be “retired” (ineligible to further play in the Ladder League) to “Open/Legacy” status if it has not been challenged in two months. Your team is retired, either for being “all that and a bag o’ chips” in greatness… or that pitiful.

The league point system is a strength of schedule system.
The League Point system that I'm using is used by the World Chess Organization and by Wizards of the Coast for the Magic the Gathering World League.

It works really well for longer term BB leagues (like 10 games), but I believe it will work for a 5 game league season also.

Every team starts with 100 League points, and your League point total can never go below zero.

At the beginning of every game you compare the league point difference between the two teams on the following chart:

Point Difference League Points
0-10 16/16
11-32 15/17
33-54 14/18
55-77 13/19
78-100 12/20
101-124 11/21
125-149 10/22
150-176 9/23
177-205 8/24
206-237 7/25
238-273 6/26
274-314 5/27
315-364 4/28
365-428 3/29
429-523 2/30
524+ 1/31


Okay, the first number in the Points column is the points that the winner gets and the loser loses if the higher league point team wins. The second number is the points that the winner gets and losers loses if the lower league point team wins the match.

In case of a tie, each team gets half (round up if needed) of they points they should get for winning.

Example:
The Chaos All-Stars play the Bluebun Crammers (halflings). The All Stars have racked up 174 league points after 5 victories, the coach of the Bluebuns called him out to get the extra MVPs for his team and the All Stars accepted the league match after hearing the comments the halfling coach also made about his girlfriend.
The Halfling team has lost 5 straight games and now has league points of 24. The difference between the two teams is 150. On the chart this indexes to 9/23.

This means that if the All Stars Win:
All Star League Points = 174+9=183
Bluebuns LPs = 24-9 = 15

Should the Crammers pull off the upset win:
All Star League Points = 174-23 = 151
Bluebuns LPs = 24+23 = 47

Should the teams tie:
All Star League Points = 174+5 = 179
Bluebuns LPs = 24+12 = 36

Depending on how it plays out... I may relent on "forced retirement", though, really, what's the purpose in being an unassailable #1?

[Hache]

I like this ! Team Value is not taken into account, and a chart is way more easier to understand and use than a math function.

[Belmondo]

Hi all! We've been using this ranking system in our league. I like it. It works nicely but not perfectly and since I'm such a perfectionist, I'd like to raise a couple of questions:

-Is there a logic behing the first column numbers (the Point Difference)? Is there some kind of a finely-tuned equation working that I'm not clever enough to appreciate? Why aren't the increments more simple and maybe stable as well, say:

Point Difference League Points
0-20 16/16
21-40 15/17
41-60 14/18
61-80 13/19
81-100 12/20
101-120 11/21
etc.

-Would the point difference between two teams really ever go up to the multiple hundreds in a normal size&length league? The table could (and therefore should) end at maybe 200 ranking difference. We play maybe 5-25 games per coach this season (multiple teams per coach are allowed). Right now the max difference is a bit over 60. Lookin at Digger's league it's been as high as 150, but that looks like a pretty relentless halfling coach.

-Draws/Ties. It's very possible to be first in the league by draws alone: just play a lot of games and don't lose more than you win. That just doesn't seem right. I like it that the draws give you some points: it encourages gaming to see that on average, playing will move you up (1W+1D+1L => increase in ranking, ceteris paribus). But since playing games also gives you the benefit of developing your team in this game (you can't develop your team in Chess, although it would be more fun) maybe draws should give you slightly less points than they do: say 1/4 instead of 1/2 of the points for a win? (Or maybe even some points for the underdog and zero points for the overdog - although I do think that defending your position has some value and should give some points. In boxing, draw=champion wins. Boxing is not an ELO system but it's pretty close to Blood Bowl.)

I've been playing around with numbers, trying to adjust the system for our next season, but would really appreciate help with these issues.

[Warpstone]

Digger, the MBBL system works with a points reset at the beginning of each season. All teams get put back to a 100 rating for their initial match of the season and then the ELO factors start to create spreads. If your ladder doesn't have a fixed term, then you may want to consider reducing the "K factor" once teams hit certain thresholds (i.e. 200 points). This is because the degree to which a team benefits from a result should be scaled to how far ahead of the pack they are in terms of experience and sheer volume of matches played. If unchecked, your powerhouse teams will reach ratings that are almost untouchable largely because they've been around longer.

An example would be that teams that have passed 200 league points now start earn a slightly smaller fraction for an equal TV win (i.e. 16 pts x 3/4= 12 points) so that they are still in striking range of a younger team at 150 LP. You can find out more about K factor at: http://en.wikipedia.org/wiki/Elo_rating ... e_K-factor

[Alamar]

If you're a ladder overdog && tie I like the idea of counting the game as "half a loss" instead of as "half a win". This way an overdog loses points for a tie so they are forced to win to keep growing in the league.

Example: The DoomDivers with a Ladder rank of 100 challenge the Unbeatables with a Ladder Rank of 225 there's a 125 point difference:

-- If the Unbeatables win they go from 225 to 235 [+10 points] ... The DoomDivers lose 10 points to 90.

-- If the Unbeatables only tie the Gobbos then they FALL from 225 to 214 [-11]. The Gobbos gain +11 for the effort to 111.

I like the idea of a sufficiently high overdog losing more points with a tie than they would gain with a win!!! It also keeps the ladder ranks from getting too inflated with nothing but draws pumping you up.

The idea also gives nice symmetry ... win/lose/draw the winner & losers get the same delta in points applied to them and the net result is usually 0 [unless someone has a ladder rank near 0 to start with ...]

[mattgslater]

Haven't read the MBBL rules, so apologies if this is a stupid question. But I've read your posts on the system in action, and I think I understand it.

Have you thought about reducing the penalty for losing, so that teams that play more games will naturally climb the ladder, even if they go through a losing streak? I got this buddy with a 1-1-14 Necromantic team that is just gnarly. Nobody would be stupid enough to face it with a rookie team; he's only been shut out once, and would ruin anybody were it not for his ridiculous Keystone Kops retrieving squad (for some reason, his SH/Block/KOR Ghoul is a classic bum; he routinely fails his pickups, but gets lots of Cas and MVPs, and the other Ghoul keeps getting killed, so no matter how often he disappoints he still has a job).

[Alamar]

A team that was 1 win, 1 loss, 14 Ties [please tell me he didn't lose 14 times] shouldn't go too far with a system that rewarded an overdog tie with "half a loss".

As far as challenging starting teams you could allow a starting team to refuse a challenge against a veteran team without any penalty -- sounds like a good idea to keep league bullies from getting gimmes by challenging folks with no chance.

[mattgslater]

A team that was 1 win, 1 loss, 14 Ties [please tell me he didn't lose 14 times] ...

Yes, 14 losses in 16 games, with TDs in 15 games. To be fair, four of the losses and the tie are to my teams; he's perhaps the worst coach in the league, but he'd still be a competitive tier-two guy down at the game store. One time he lost 5-0, but most of his games are (1-2) to (2-4), with a 2-2 tie and a 2-1 win. That's what happens when you have an effective quick-strike defense, poor coverage skills, and a really s**tty offense (again, just chance; Nuffle seems to like his Ghoul squad, killing one over and over, while letting the other one soak up MVPs and fail, fail, fail... but only on offense).

But you can't get into his backfield without getting through either Blodge/SS or Blodge/SF/ST4, and he has an all-Guard line with ST4 ends, and two Ghouls (***hole bum and victim-in-waiting) in protected positions. Would you go up against that with a rookie team and 900k inducements?

[Belmondo]

If you're a ladder overdog && tie I like the idea of counting the game as "half a loss" instead of as "half a win". This way an overdog loses points for a tie so they are forced to win to keep growing in the league.

Example: The DoomDivers with a Ladder rank of 100 challenge the Unbeatables with a Ladder Rank of 225 there's a 125 point difference:

-- If the Unbeatables win they go from 225 to 235 [+10 points] ... The DoomDivers lose 10 points to 90.

-- If the Unbeatables only tie the Gobbos then they FALL from 225 to 214 [-11]. The Gobbos gain +11 for the effort to 111.

I like the idea of a sufficiently high overdog losing more points with a tie than they would gain with a win!!! It also keeps the ladder ranks from getting too inflated with nothing but draws pumping you up.

The idea also gives nice symmetry ... win/lose/draw the winner & losers get the same delta in points applied to them and the net result is usually 0 [unless someone has a ladder rank near 0 to start with ...]

I'm starting to like this idea of negative points for overdog draws... My problem is that I really feel you shouldn't get past this particular opponent of yours in ranking with just a draw.

Eg.:
League overdog: 161 points
Challenger: 150 points (difference 11 points)
Draw means Overdog (-8) drops to 153 and Challenger (+8) would get past him to 158. Without winning him. That doesn't feel right. You need to beat the champ to be the champ!

* One possibility is to change the first Point Difference scale from 0-10 to at least 0-19. That would make the Overdog always unreachable with just one draw.
(Overdog 169 -9 =>160
Challenger 150 +9 => 159)
Although for the occasional situation where teams have the same or almost the same Ranking (under 10 point difference) this would make draws very lucrative for the champ (+8 points) and could result in boringly safe play from his part.

* Another possibility for eliminating this/these is giving less points for draws. Only 1/4 of the points for draws means that even if you penalize the overdog for draws you'd never get past him with just a draw.
Ie.
161 points -4 => 159
150 +4 => 154.
You'd still get some points for trying, but winning would be much more valued.

Hope I'm making sense, it's hard to articulate this without knowing math in English.

[Alamar]

How about this? ....

-- If you are an overdog by more than 16 points [or whatever number works] and you tie then you "lose" half a game and the opponnent "wins" half a game in terms of points.

-- If you are an overdog by no more than 16 points [or whatever number that works] and you tie then it's a true draw with NEITHER team gaining or losing points.

Obviously you could get more complicated with the system, introduce formulas, etc. but this isn't a bad way to go.

Then again the prior system where you trade off who is number one may not be bad either as it opens up possibilities for lots of smack talk & changes of position

[Digger Goreman]

Interesting, and timely, to see this come back up... as I've been wrestling (mightily) to get the metagaming aspect out of league play.... The Ladder was a (pardon the pun) "step up", but still suffered from "gaming the system".... As Sun Tzu and Von Clausewitz would readily agree, no competent general seeks a fair fight!

I have a sneaky suspicion that the challenge system in the rulebook is actually the way to go IF you have a regular meeting place/day that ALL coaches come to.... Not a chance for my league.... Real Life often intervenes....

So, variable schedules... perpetual teams... (human?) tendencies to seek advantages... all folks still want to have fun.... What to do...?

First, obviously (and the easiest), let them schedule their own games!

With many apologies to people who don't like formulas, I think we (those interested) have to dive into the design of the game, as expressed by former member(s) of the BBRC.... Similar to the rulebook's "collapse of the NAF", travelling teams, let's have each coach with a stable of teams.... The teams will each have a rating composed of their "average strength of schedule", "current TV", and "real rating"/(worth?).... This real rating (or whatever better name comes up) can be used on a composite ladder if the commish wishes....

The Assumptions and Formula!:
The TV that you have available for the pitch/field, should determine the outcome (coaching strength being an uncontrolled variable).... Intentionally dodging the interminable arguments of player/skill worth, LRB6 has values for each of these.... In short, the first number of importance is the Team Value (don't forget to add the value of all journeymen!).... Add to that, 1/3rd of any inducements taken (this is the agreed upon, design value of inducements, per BBRC).... Divide your opponent's value by your value.... Now, multiply that quotient by the quotient of: your opponent's tier value, divided by your tier value....

Tier values, as expressed by Galakstarscaper, for LRB6.0
Tier 1 (45%-55%) All teams not rated below.
Tier 1.5 (40%-50%) Chaos Pact, Slann, Underworld
Tier 2 (35% - 45%) Vampire
Tier 3 (30%-35%) Halfling, Goblin, Ogre

The "Tier Value" is the average expected win % of each tier....
Tier 1 = 50%
Tier 1.5 = 45%
Tier 2 = 40%
Tier 3 = 32.5%

So it's:
(((opponent's TV)+(1/3 inducements taken))/((your TV)+(1/3 inducements you took))) times ((opponent's team's average expected win %)/(your team's average expected win %))

In the ExCel sheet, the formula is:
=((F3+(1/3*G3))/(D3+(1/3*E3)))/((IF(H3=1,0.5,(IF(H3=1.5,0.45,(IF(H3=2,0.4,0.325))))))/(IF(I3=1,0.5,(IF(I3=1.5,0.45,(IF(I3=2,0.4,0.325)))))))

D is Your TV
E is Your Inducements Taken
F is the Opponent's TV
G is the Opponent's Inducements taken
H is Your Tier
I is Your Opponent's Tier

In summary:
the only way I feel you can compare Goblins to Elves (Apples to Oranges) is to take a look at the strength of schedule they play... all quantitative design parameters compared.... This is comprised of Team Value, 1/3rd Inducements, and expected winning percentage, by design.... Without any more refined numbers, this should give you a truer worth of your team that can be compared to any other team....

Possible refinement:
(Especially in a league with a dominating coach/coaches), Substitute the coach's actual win percentage, for each individual team, as the Tier rating....

Hopefully the formulas meet/met the assumptions and, all math aside, this will prove useful to present and future leagues....

The basic spreadsheet is in the attachment....

TeamSheet Strength of Schedule.xls