Amazing Video here, very relevant to Bloodbowl!

Everything dealing with the video game developed by Cyanide!
User avatar
dode74
Posts: 7040
Joined: 11 December 2008, 11:18
Location: Nr. Reading, UK
Contact:

Re: Amazing Video here, very relevant to Bloodbowl!

Postby dode74 » 30 May 2017, 06:29

It seems strange that the people who have complained that the quality of the opposition you face will have a large effect on your ability to qualify are also complaining that poor opposition getting a handicap is a bad thing for the competition.
Image

triperis
Posts: 26
Joined: 06 January 2017, 06:31

Re: Amazing Video here, very relevant to Bloodbowl!

Postby triperis » 30 May 2017, 10:47

I believe this discussion could be relevant only if BB2 was not a dice game in the end of the day. If you want to show minor differences of skill, there should be way more reliable actions to show off skill. As long as everybody can roll a 6, or cas an important player, you will need to both tilt the probabilities towards your side AND be on a good streak.

I'd be interested to see an analysis how much winrates can differ between coaches who actually do not misclick, know how caging and column defense works, how to break cages and how 2-1 grind works and how to play around it. I.e. all the basic BB2 skills. Because it seems like a bunch of guys feeling entitled because they play people not knowing basics and think they are skilled enough for their skill to bring them to 100 % winrates.

From what I have seen, best PC coaches have like 80 % winrates including playing all the guys who do not know basics. This includes draws, but for simplicity let's assume it is pure win probability. Getting even basic 10-0-0 run with 80 % winrate becomes a 11 % chance. Because there are more than 10 people with 80 % winrates, at least 1 is likely to have that good record at 10-0-0. While the rest knowing they are NOT less skilled, blame everybody else. But in reality their skill is nowhere good enough to compensate pure luck. You'd need winrates closer to 95 % to start feeling entitled to prolonged winning runs. 95 % pure win winrates, Carl*.

This is partly why I speak negatively about principles of scoring system. People proclaim they want to award "the best teams", while in reality they reward the luckiest, because scoring system in no way accounts for certainty of the win record (aka luck). Not that we can realistically expect people getting statistically significant number of games. But certainty of the record can be accounted for, but isn't, aside very indirectly, through rewarding number of games.

User avatar
dode74
Posts: 7040
Joined: 11 December 2008, 11:18
Location: Nr. Reading, UK
Contact:

Re: Amazing Video here, very relevant to Bloodbowl!

Postby dode74 » 30 May 2017, 14:09

How is it "very indirect"? The scoring system directly accounts for record and number of games played: they are the two major factors.
Image

triperis
Posts: 26
Joined: 06 January 2017, 06:31

Re: Amazing Video here, very relevant to Bloodbowl!

Postby triperis » 30 May 2017, 15:23

Because assuming Poisson distribution the uncertainty scales as square root of number of games, not directly with games. That is getting wrong dependency, if one is willing to reward not the games played (which is definitely ok, we all want more people playing), but more certain records. Even then, there is no reasoning connecting how much certainty is enough and the scoring system. It was "let's see what sticks" calibration of coefficients showing how the previous table would have looked with new system. This is all well and good for crowd pleasing evaluating which records they "feel" are more worthy of qualification.

E.g. let's take 16-0-0 and 27-0-5 records with latter being marginally better under current scoring system. 16-0-0 record, while fairly uncertain, actually at one sigma means it can be 0.75 winrate at worst (assuming he was lucky, and it is a reasonable assumption). But in all fairness, same assumption of luck should be applied to the second record, too, which actually means the latter guy's winrate would be lower by almost 0.1 even if his record is more certain. The current formula could mimic the certainty quite well if it was square root of games played instead of integer number. But there were no considerations like this. Because there were no considerations of luck involved.

I do believe the current formula makes majority happy and ticks most if not all good boxes and a great compromise. I just do not appreciate comments it is made to reward best teams. It might have been in idea, replaced by direct number of games in design and considered "good enough".

It is a game of dice (you need luck) and grind (just because you can ignore bad results by deleting teams to hit those streaks). No need to believe it is supposed to be fair to each individually ;)

User avatar
VoodooMike
Posts: 1614
Joined: 14 July 2009, 07:44
Location: Calgary, Alberta
Contact:

Re: Amazing Video here, very relevant to Bloodbowl!

Postby VoodooMike » 30 May 2017, 15:32

I dont agree with that at all, i just want less nonsense in the way of me qualifying. Getting tougher matchups if said matchups are available just makes it more of a grind and very few people have time to commit that many hours to qualify consistently.
Which would be a valid complaint if your ranking was not automatically going up simply by playing games. CCL incentivises playing more games rather than sitting on one's laurels and stopping play just to hold an artificially high rating (that may have been achieved through luck... which is what happened with the WC qualifiers).

In modern CCL you get a ranking "bonus" for playing more games, but as you continue to play them it prioritizes matching you with other players that are similar performers which, in turn, lets us find out which of you two high performers is the better performer and thus deserves to go on to qualify.

Under full TVPlus having win streaks would no longer be important in qualifying - the system would be able to figure out who the best coaches were via the performance distributions. It would be considerably more accurate than the current system... the main issue seems to be that people have trouble wrapping their head around why.

Skill is not a big enough deciding factor of succes as far as qualifying is concerned...
If you truly believe that then you have no real dog in the fight as to how the qualifications are done.

VoodooMike, that's your opinion mate :).
It's more than a simple opinion, "mate".. it's supported with unimpeachable logic and data. What folks like you have is "feelsies" about the topic, which you imagine is just as valid because all opinions are created equal. Fact is... they're not.

Mine is that skilled players should not face an artificial mechanical (TV) handicap in order to produce an ideal 50% win rate across the population. That trend towards dumbing down the difficulty for less skilled players is exactly what the video was discussing. There is no other reason to force players with even 1 win to be penalised by matching them against higher TV.
TVPlus does not "dumb down" the difficulty of the game, it simply ensures that every match is a challenge suitable to the performance abilities of both teams and coaches... that means a single win is easier for a below average player, but a single win is harder for an above average player. That would be an issue if we couldn't tell the two apart because of that, but we absolutely CAN tell them apart in that system... we know who the skilled people are... we know who the unskilled people are.

If the unskilled people want things to be harder they can choose not to spend their inducement cash.. so the idea that it's "dumbed down" for them against their wishes is silly. What you really mean is that the chickenshit so-called "skilled players" are afraid of the game being harder because then they might have to do more than phone in their game against the average opponent.
Anything that attempts to handicap skill in a prize competition is ridiculous.
CCL's current use of the TVPlus rating matching does not "handicap skill" in any way. Inducements are the handicapping system, and it uses TV for those. You're whining about nothing.

Full TVPlus on the other hand does use handicapping for each match but does not rank you the same way. Your complaint would be valid if it did rank the same way we currently rank, but it does not. Instead it ranks in the way those coin-operated strength tests measure you.... it sees how hard you can push against resistance and maintain that push, ignoring the quick push at the start in favour of the endurance.

The TV based win% systems we've used thus far are very vulnerable to that quick push at the start, which is why there has been a need to set minimum numbers of games played to try to find a proper ranking. In the WC qualifiers you had to have 20-0-0 to get in... and we all know a 100% win rate in BB is based heavily on luck, and cannot be maintained indefinitely. What we really want to know is what success rate you can maintain in the long term... and TVPlus can do that effortlessly... the existing systems struggle to find work-arounds to do that.
Friendly Reminder: Correlation does not equal Causation - tattoo it on the inside of your eyelids if it'll help.

User avatar
VoodooMike
Posts: 1614
Joined: 14 July 2009, 07:44
Location: Calgary, Alberta
Contact:

Re: Amazing Video here, very relevant to Bloodbowl!

Postby VoodooMike » 30 May 2017, 15:53

Because assuming Poisson distribution..
Which we cannot assume due to how those records are calculated and how the games within the environment are distributed.

In CCL seasons your record is not strictly made up of legitimate games, there is a concession rate of around 20% overall, but that's a rate that starts higher and decreases over the course of the season. Likewise, the average skill level starts low and increases over time as lower performing teams/coaches bow out. Either one of those facts violates basic poisson assumptions.
But in all fairness, same assumption of luck should be applied to the second record, too, which actually means the latter guy's winrate would be lower by almost 0.1 even if his record is more certain.
Not "would" be... "could" be.

Even if I don't agree that we can use the model you're suggesting, I don't totally hate the idea of using the low end of the probable record for qualification... but I don't think that's something that the players themselves could relate to, and it'd be hard to display to them in a way that wouldn't lead to a riot.
I just do not appreciate comments it is made to reward best teams.
And yet, it is exactly that. Keep in mind that you're talking about ranking by "what might have been" rather than what actually is... and ranking by what actually is is exactly what is being done. We have no way to properly calculate the contribution of "luck", which is the aspect you're objecting to, so going by performance with well-defined scoring systems, is almost certainly the best way to go.
Friendly Reminder: Correlation does not equal Causation - tattoo it on the inside of your eyelids if it'll help.

User avatar
dode74
Posts: 7040
Joined: 11 December 2008, 11:18
Location: Nr. Reading, UK
Contact:

Re: Amazing Video here, very relevant to Bloodbowl!

Postby dode74 » 30 May 2017, 16:59

I don't totally hate the idea of using the low end of the probable record for qualification...
That is something I looked at while researching TrueSkill, and I rejected it because we want to reward actual record for the season, not the potential record for a longer season.
Image

User avatar
VoodooMike
Posts: 1614
Joined: 14 July 2009, 07:44
Location: Calgary, Alberta
Contact:

Re: Amazing Video here, very relevant to Bloodbowl!

Postby VoodooMike » 30 May 2017, 17:24

That is something I looked at while researching TrueSkill, and I rejected it because we want to reward actual record for the season, not the potential record for a longer season.
Except that the potential record is a measure of their actual ability assuming we admit that the actual record, at least in part, is based on chance. The reason I wouldn't use that method given the choice is that it is difficult to present to the players in a way that they could relate to and understand... hard to tell them "yeah you have a better record, but through calculations we can't really explain to you in a way you'd understand, we think you're a weaker player so... no prize for you".

We've already seen that people have a hard time with that just via the ranking formula that penalizes for fewer games played. I wouldn't recommend tempting fate by getting even more esoteric.
Friendly Reminder: Correlation does not equal Causation - tattoo it on the inside of your eyelids if it'll help.

User avatar
dode74
Posts: 7040
Joined: 11 December 2008, 11:18
Location: Nr. Reading, UK
Contact:

Re: Amazing Video here, very relevant to Bloodbowl!

Postby dode74 » 30 May 2017, 17:39

assuming we admit that the actual record, at least in part, is based on chance
Of course, but most things have an element of luck in them. Real sports certainly do. I don't see that as a downside. What we're really saying is "this team had the best season by this measure" rather than "this is the best team".
Image

User avatar
VoodooMike
Posts: 1614
Joined: 14 July 2009, 07:44
Location: Calgary, Alberta
Contact:

Re: Amazing Video here, very relevant to Bloodbowl!

Postby VoodooMike » 30 May 2017, 17:57

What we're really saying is "this team had the best season by this measure" rather than "this is the best team".
Well, all rankings are saying "this is the best team by this measure"... if we gauged them by how nice their hair was they'd be the best team by that measure. I think it falls to the people making that system to demonstrate that it approximates "this is the best team overall" lest people not consider it a worthwhile system.

As it stands I think the existing system does a sufficient job of that considering the obstacles.
Friendly Reminder: Correlation does not equal Causation - tattoo it on the inside of your eyelids if it'll help.


Return to “General”

Who is online

Users browsing this forum: No registered users and 1 guest

cron