Write leagues system from scratch

[HHC]

So if you have a high overall it won't effect the amount of points you gain or lose because you are only taking into account the overall of your opponent??

Yessir.

Could you in theory go pro hunting to just gain points off higher ranked overall players?

You can, but is that a bad thing? Will be hard enough to win 50% of those games. I think it's pretty balanced, noobs are still worth quite a few points, while at the same time, pro players may not be shy anymore when it comes to playing fellow leets 

What are the point differences like if random00 loses games against players with a lower overall/season?

It depends on where random is at in the season. At the very start of the season he won't lose many points (as it would be 1000 vs 750). Later on, when his season rating is at 1400 or so, he may lose a whole bunch (as it would be 1400 vs 750 then; the number of points he loses then is fairly similar to the points he loses in the current TUS system).

Shouldnt in theory i deserve to win more points off random than csongi because he is of a similar skill level (from his overall being closer to randoms)??

Depends on how you look at it. Your performance is more special perhaps, as it isn't as expected.
But when you start from scratch each season it is inevitable. For example, it may be more worthy for Australia to beat Brazil at the World cup than it is for Spain, but if you give Australia 6 points for it, and Spain only 2 points you'll get a very weird world cup don't you think 

The way I see it is that you and Csongi are equally good when you both manage to beat Random. You should not be higher in the standings for achieving the same result as Csongi. It's perhaps a performance that is worthy of more praise, but it doesn't make you a better player than Csongi is. And that's why I feel you both should get the same amount of points.

(this is only at the very start of the season btw, if Csongi has a higher score this season than you, he won't get as much points added anymore).

[Chicken23]

Yea i agree with the football example. Also because the points u win/loss come from the overall of your oppenent there is no handicap on those players having a greater overall at the start of the season?

You'll see the greater loses of points as your seasonal catches up with our overall but that is fair because thats how your preforming over the season!

May question tho is it hard for a player like zippo, random, dulek to gain points over a period of time compared to average lower allrounders? Is there any form of handicap for anyone with a greater overall? Or is there any kind of reward for someone with a greater overall? I still think that your overall should be worth something?? Players have done alot to deserve that. Shouldnt atleast 2 spots of overall make the playoffs if they have done the necessary games and winning %?

But with the seasonal points calculation you are likely to see them reaching a playoff spot anyway because of the seasonally points they will gain from making the 50 games and winning % anyway?

therefore not making my concern an issue?

[HHC]

Are you one of those people who finish every sentence with a question mark?  


It is true that the overall ranking has no effect on P/O quali's anymore. This is why I proposed a TUS global ranking in which the overall ratings in the leagues are counted together, as to form a kind of World Ranking. If this does get implemented it will be quite an honour to be on top of those rankings and it gives players something to work on in the long term, over many seasons.
As well as bind all of the seperate leagues and cups together into a global system (and thus prevent the ongoing desintegration).

[franz]

Chicken:

having a high overall rating doesn't hurt your own seasonal performance. you're confusing this way with how TUS currently is.. and that's the important difference. there is no 'seasonal disadvantage of getting fewer points from most players' -> you start at 1000 season rating like everyone else and gain/lose points according to other people's overall rating.  so no, you don't have to only stick to playing amongst other high level players -> you're even free to play 'noobs' if you want since losing won't hurt your seasonal rating as much compared to overall rating now and winning gets you more points compared to overall rating now.

The only real benefit of having a high overall ranking at the start of a new season.. is that other high rated players will want to play with you.. so it's kind of an indirect benefit that you attract those other high overall rated players -> giving you a nice boost if winning vs them.  anyway, I guess this isn't all that bad actually.. since you should want to play other high rated players anyway.

And Chick, regarding your comment about "is it hard for a player like zippo, random, dulek to gain points"... the answer is no because there isn't really a handicap for having high overall rating (this has been said many many times now, but worth repeating). is there any reward for having a greater overall you ask?  well, my quote above shows that there is indirect benefit, but it's a healthy benefit because it encourages high level players to play each other. I really don't think it's healthy for the league if someone with huge overall rating easily wins 80 games vs noobs to qualify for something like PO's. they should still have to work for a high seasonal rating, but this should not be hard at all if their true skill level is measured by their overall rating -> because then they should be winning vs great players and getting the appropriate amount of points to reach high in the new standings. so yes, your last line should be accurate -> "with the seasonal points calculation you are likely to see them reaching a playoff spot anyway because of the seasonally points they will gain from making the 50 games and winning % anyway"

[DarkOne]

I don't understand why comparisons with football keep being made. They're using a completely different competition format where there's a set amount of games. Worms leagues don't have a set amount of games.

We need a system where you don't just get points because you are active regardless of your results; there should be a minimum requirement of games played for playoffs, but whether you're eligible for playoffs or not should depend on the results (this means not only the percentage of wins, but also the opposition against which you reached that result). And perhaps your rating shouldn't depend mostly on the last 10 games you've played, but rather on the entire collection of games you've played through the season; the effect of beating Random in an elite at the start of the season is entirely lost after about 10-20 games, but if your last elite was against Random, you can see a huge boost in your rating.

[Rok]

No post here in a week. By the look of it, we have a full detailed of HHC plan and Mablak's suggestion for a system like NNN's.

I'm not done with this thread yet (if that matters anyhow ), I just don't have time for worms at the moment.

I don't understand why comparisons with football keep being made. They're using a completely different competition format where there's a set amount of games. Worms leagues don't have a set amount of games.

We need a system where you don't just get points because you are active regardless of your results; there should be a minimum requirement of games played for playoffs, but whether you're eligible for playoffs or not should depend on the results (this means not only the percentage of wins, but also the opposition against which you reached that result). And perhaps your rating shouldn't depend mostly on the last 10 games you've played, but rather on the entire collection of games you've played through the season; the effect of beating Random in an elite at the start of the season is entirely lost after about 10-20 games, but if your last elite was against Random, you can see a huge boost in your rating.

Yes, indeed. One tweak we can still do is to adjust ratings after a series of games rather than after every single game as it is now. ATM, I'm still not convinced by. The system that Mablak mentioned (table tennis thing) could be promising for this league, though.

One more thing, @HHC, franz et. al.: Either use the overall rating in the seasonal calculations fully (in a way similar like we do it now), or don't use it AT ALL. Fiddling with some magic numbers without any statistical or mathematical background is meh and wrong and all...

Cya again when RL let's me...

[franz]

One more thing, @HHC, franz et. al.: Either use the overall rating in the seasonal calculations fully (in a way similar like we do it now), or don't use it AT ALL. Fiddling with some magic numbers without any statistical or mathematical background is meh and wrong and all...

uhh what? why such forceful language? you make a strong statement about using overall rating in seasonal calculations fully or NOT AT ALL and you don't back yourself up with any explanations as to why.

you must not fully understand the idea if you describe it as "fiddling with some magic numbers." then you claim we're trying this idea "without any statistical or mathematical background" as if this has never been tried before... -> except it has. there have already been comparisons to WoW's arena rating system, so no.. this isn't foreign territory.

[Chicken23]

Will there be no change to overall calculations? So if random00 did lose to a noob he will still lose alot of points in his overall.

I hope this is the case, then you will get a more accurate stats for overalls and how much you would win off random00 if he went on a losing streak.

He loses to a noob. Loses 25 seasonal. But loses 60 overall. So players playing random wont gain as much if he has a some bad results due to his lower overall.

[HHC]

Will there be no change to overall calculations? So if random00 did lose to a noob he will still lose alot of points in his overall.

I hope this is the case, then you will get a more accurate stats for overalls and how much you would win off random00 if he went on a losing streak.

He loses to a noob. Loses 25 seasonal. But loses 60 overall. So players playing random wont gain as much if he has a some bad results due to his lower overall.

That is the case. The overall and season scores are calculated seperately. The overall ones the old way, the seasonal ones the new way.

[Rok]

One more thing, @HHC, franz et. al.: Either use the overall rating in the seasonal calculations fully (in a way similar like we do it now), or don't use it AT ALL. Fiddling with some magic numbers without any statistical or mathematical background is meh and wrong and all...

uhh what? why such forceful language? you make a strong statement about using overall rating in seasonal calculations fully or NOT AT ALL and you don't back yourself up with any explanations as to why.

you must not fully understand the idea if you describe it as "fiddling with some magic numbers." then you claim we're trying this idea "without any statistical or mathematical background" as if this has never been tried before... -> except it has. there have already been comparisons to WoW's arena rating system, so no.. this isn't foreign territory.

Sorry if it sounded harsh, it wasn't meant to. I really don't have time to explain myself, because writing a long post takes me several hours (I like to do some research  ). In short, from what I read about WoW system, It looks roughly like HHC system, but it really isn't. That 'par' number in HHC equations - where does this come from, what does it represent?, etc... this is what I haven't understood reading this thread. Rating systems were developed by people with PhD's who surveyed databases of hundreds of thousands of games before they came up with things like 'expected score' etc. Without being disrespectful towards HHC's work, but I'm sceptical.

[HHC]

That 'par' number in HHC equations - where does this come from, what does it represent?, etc... this is what I haven't understood reading this thread.

In a zero-sum system like the old TUS the amount of points won by the winner and lost by the loser of a match is the same. Winner gets +18, loser gets -18. The difference is 0, hence zero-sum. This is what I called the par value.
In my system there's a discrepancy between points won by winner and those lost by loser. Winner may get +29, while loser gets -12. The sum here is no longer zero, but +17. This sum is what I referred to as the par value. Perhaps it's not the appropriate term, but I thought it was pretty clear what it stood for.

Rating systems were developed by people with PhD's who surveyed databases of hundreds of thousands of games before they came up with things like 'expected score' etc. Without being disrespectful towards HHC's work, but I'm sceptical.

Modesty is a virtue. But what these PhD's have come up with is an ELO system meant for continous play. In CPU games this is not always applicable, thus new systems had to be created. I'd go for one of these, were it not that Worms is a totally different game as well. Worms has schemes, gametypes of which the ratings tend to differ A LOT. Someone may be 2000 in BnG, but only 700 in Rope.
So in short, like I said in the article, you can't use a system already thought up by PhD's. You gotta come up with something new, to suit it for TUS.
The WoW arena thing Flow mentioned is a good system because it incorporates seasonal resets into the former ELO-system. But since WoW doesn't have seperate schemes it is not fully suited for Worms either.

I know you're sceptical, so is everybody else I think. But unless you try out these things, there's no way of knowing what works and what does not. We can always tweak it later on if necessary.

[darKz]

I don't know how much work coding the HHC system would be. If it's not too much, why not try it out? It does sound interesting at least.

[Chicken23]

And perhaps your rating shouldn't depend mostly on the last 10 games you've played, but rather on the entire collection of games you've played through the season; the effect of beating Random in an elite at the start of the season is entirely lost after about 10-20 games, but if your last elite was against Random, you can see a huge boost in your rating.

Are you saying HHC's plan doesn't cover this situation well? Becuase of the potential point difference of beating random during the start of end of a season?

What is the solution for this HHC, franz et al because beating random in elite at the start of the season is going to be just the same as the end. But the points u gain from this may vary  due to when you play him, isnt that a little unfair?

[franz]

uhh what? it's always about the entire collection of games you've played through the season. you can't ignore all your first games and just look at the last 10 games.. because all those first games put you into that position to play those last 10 games.

as for your example (I'm sure Random loves the attention) -> beating Random in elite at the start of the season -> you say the effect of beating him is entirely lost after about 10-20 games... why? you'd get the appropriate reward of points due to wherever his overall rating is (most likely high) and if he were to go on a big losing streak, then appropriately fewer points will be rewarded to anyone who beats him later. the system accounts for your overall skill fluctuating -> and rewards people correctly due to their performance, however good or however bad.

[DarkOne]

Well, then why does it matter in which order games are reported?
The answer is that if I play 2 games against someone and I score 1-1, then my rating will be higher if I win the second game.

I don't know what exactly is the rating system, so I can't give any objective numbers, but just pick your last 10 elite results, but then throw in a win against Random at the start and calculate your rating. Then have the same results and put the win against Random at the end instead of the start.
Compare ratings. You will see a big difference.