Write leagues system from scratch

[HHC]

?


Let's make a new rule. For Zippo we don't count season ranking, but reputation level 

[zippeurfou]

This seems quite a good idea. And I'm always with my hhc.
One question:
Did you take time to read what famous game use for ranking ?
look here for example:
http://www.wowwiki.com/Arena_PvP_system

[franz]

fada, that page is very outdated as noted in many spots on there.  some of the info might still be true/relevant, but they've constantly changed their calculations many times and now no one really knows the exact maths behind their current system.

[Rok]

In your opinion both players should gain/lose 0 points and stay at 1050, because this is the expected result.

Yes, IF the actual result of games is the same as expected result, then rating change is exactly 0. If a player played according to his skill/rating, then his actual skill didn't change. Sounds logical?

In my opinion both players' real skill level is 900. When they play a draw against each other they 'confirm' that they are equally good, namely '900'-good and not the 1050 that happens to be their season rating. Thus, it doesn't seem illogical to me, that both players drop a little in season rating (to resemble more closely their real skill-level of 900).

It is illogical. Season rating shows us how well a person did IN A GIVEN SEASON. Overall rating is supposed to show player's skill in the long run, over multiple seasons, over his entire league 'career'. These are two different things, don't you think? So why do you want player's seasonal rating to DROP DOWN when he is doing like expected?

I repeat again, if you do better than expected, your rating should rise. If you do worse, your rating should drop down. If you do as expected, it should stay the same. I hope you now understand why I have problems with your proposal.

I think all of your plans for changes come from the idea that seasonal rating should eventually approach overall rating. Did I get you right? But as I said, seasonal rating (why shouldn't even call it rating in the first place, but rather "performance", "score, or smth) should show us how players perform in a given season, not show the skill, that's what we have overall rating for.

The essence of zero-sum is that the sum of points gained by winner and loser is zero. This is not a necessity IMO.

Think again what is rating supposed to show you - the skill. So after a series of games, say 10, the result is as expected (according to player's ratings), for example 8-2. The result tells us that both players played ok, according to their skill, so the sum of rating changes is 0. If it wouldn't be, the ratings would change whereas we said players performed as expected. Don't you think this distorts the picture a bit?


Reading your work I didn't catch one detail (or maybe I missed it) - do you want this calculations to affect only seasonal scores or also overall ratings? If only seasonal score would be calculated this way, then what you're proposing is at least somehow acceptable, altough I still think it doesn't show seasonal performances correctly enough. There are still ways to tweak it somehow, I guess.

fada, that page is very outdated as noted in many spots on there.  some of the info might still be true/relevant, but they've constantly changed their calculations many times and now no one really knows the exact maths behind their current system.

WoW used ELo system too, now they switched to http://en.wikipedia.org/wiki/TrueSkill (according to wikipedia).

[franz]

Rok, so you seem more interested in just isolating every season separately.. basically just like resetting all points every new season (so overall rating really doesn't mean anything other than something nice to know or see).  That's also what SPW seemed to want as well, mentioned in his thread: https://www.tus-wa.com/forums/tus-discussion/tel-system-need-a-change/

It's not the worst alternative    I could live with it I guess, but you'll still have the issue with 'newbies' just giving points to a few lucky players early since their season rating always starts too high at the beginning of every season.  So anyone who starts late will be at a slight disadvantage.  I guess you could justify this by saying 'well its your own fault for starting late!'
edit: the other disadvantage is if you beat someone really good at the start of a new season.. it will feel pretty empty if you only win the standard small amount that everyone else is winning at the start.  that loss of 'history' can hurt in this way too

SPW has said before in his thread: "I want to have the best players in every season in it and not just those one's, which got a big advantage by starting at zero / low overall points or leaning back because of an huge overall-point score."  With HHC's idea, the starting at zero/low overall problem will be solved... but high overall scores will definitely have an advantage which SPW might not like.  but they still have to work for it by reaching that high season rating again through many games played, although true they will get some extra help from their high overall rating.  many could argue that this is worth it though, because it becomes more accurate to show the true skill level of all players..  as opposed to simply resetting every season and losing all that data to really show where everyone stands.

SPW don't be shy, come speak for yourself ^^

[Rok]

Rok, so you seem more interested in just isolating every season separately.. basically just like resetting all points every new season (so overall rating really doesn't mean anything other than something nice to know or see).  That's also what SPW seemed to want as well, mentioned in his thread: https://www.tus-wa.com/forums/tus-discussion/tel-system-need-a-change/

It's not the worst alternative    I could live with it I guess, but you'll still have the issue with 'newbies' just giving points to a few lucky players early since their season rating always starts too high at the beginning of every season.  So anyone who starts late will be at a slight disadvantage.  I guess you could justify this by saying 'well its your own fault for starting late!'

We already have it like that! Seasonal points start fresh every new season. The overall ratings don't. That's why top players get low points for beating much lesser ratings, because it's calculated from overall standings. What SPW wanted is to reset everything each season. Now random00 gets 5 points for beating newbie00, if we reseted, he would get 40 until he we reach higher rating again. We don't need overall ratings anymore then. And we're back to the old problems: players who start playing late in the season gain much better points from those who earned points early in the season, people start to avoid good players and choose newbies instead. And we still get the game comments like "pfff -65" or "lol 5 points for win" again and we really haven't solved much, don't you think?

One other thing that really bothers people is that standings are displayed season-wise. Good players who do well are placed f.e. around 100th place in season, yet they still qualify for playoffs, on the other hand people who may be in 10th place in the season are actually nowhere near the playoffs. That's why I proposed standings table should be based on overall rating. Much less people will be bothered be severe rating changes if they still see themselves at the top of the list when they click "standings". There will also be less comments like "lol r3sp3ct in 1st place, who is that"? Yes, this way top players could sit back and get in playoffs easily, but that's why we have set minimun of games played and minimum win percentage.

SPW has said before in his thread: "I want to have the best players in every season in it and not just those one's, which got a big advantage by starting at zero / low overall points or leaning back because of an huge overall-point score."

I think SPW is making a bigger deal out of this than it really is. Just look at playoffs tables. Yes there are 1 or 2 newcomers who started low qualified in playoffs every season, but the rest are players who play a lot and have decent ratings and are really quite good. That's why I also said we should change overall/seasonal qualifying ratio to 6/2 rather than 4/4 as it is atm, to make qualifying for high rated players even easier.

[franz]

We already have it like that! Seasonal points start fresh every new season. The overall ratings don't. That's why top players get low points for beating much lesser ratings, because it's calculated from overall standings. What SPW wanted is to reset everything each season. Now random00 gets 5 points for beating newbie00, if we reseted, he would get 40 until he we reach higher rating again. We don't need overall ratings anymore then. And we're back to the old problems: players who start playing late in the season gain much better points from those who earned points early in the season, people start to avoid good players and choose newbies instead. And we still get the game comments like "pfff -65" or "lol 5 points for win" again and we really haven't solved much, don't you think?

You misunderstood me, but that's ok    I was basically describing what SPW wanted, which was fresh new seasonal points every new season (yes I know this already happens) but then point gains are only through that season rating too (overall would have absolutely no affect.. it's only a stat you would look up if it's even kept at all).

One other thing that really bothers people is that standings are displayed season-wise. Good players who do well are placed f.e. around 100th place in season, yet they still qualify for playoffs, on the other hand people who may be in 10th place in the season are actually nowhere near the playoffs. That's why I proposed standings table should be based on overall rating. Much less people will be bothered be severe rating changes if they still see themselves at the top of the list when they click "standings". There will also be less comments like "lol r3sp3ct in 1st place, who is that"? Yes, this way top players could sit back and get in playoffs easily, but that's why we have set minimun of games played and minimum win percentage.

I'm actually more inclined to support HHC's plan and even SPW's fresh start idea over trying to bandage the current system just by tweaking the standings to show overall.  All these 'seasonal/overall' seeding complications don't even come into play with HHC's and SPW's.

[Rok]

I guess I misunderstood you then. Hell, I'm inclined to the other way completely I might appear like religiously supporting rating system, but I really like the idea of it.

Ok let's say, we choose to reset everything every season. What about people who like to compete yet don't have time to play enough worms to be able to fight for playoffs? I'll give myself for example - I don't think I'll ever have enough time again to play ~100 leagues games in a season nor I wish to play that much. But I'd still like to have some comparison to other players. A rating table does that for me just fine, while fresh seasonal standings every two months simply don't.

I could even live with overall ratings and seasonal scores completely separated, as long as I have something to fight for. That's my biased view but I'm sure there are others around wormnet who feel the same way.

[Mablak]

In your opinion both players should gain/lose 0 points and stay at 1050, because this is the expected result.

Yes, IF the actual result of games is the same as expected result, then rating change is exactly 0. If a player played according to his skill/rating, then his actual skill didn't change. Sounds logical?

In my opinion both players' real skill level is 900. When they play a draw against each other they 'confirm' that they are equally good, namely '900'-good and not the 1050 that happens to be their season rating. Thus, it doesn't seem illogical to me, that both players drop a little in season rating (to resemble more closely their real skill-level of 900).

It is illogical. Season rating shows us how well a person did IN A GIVEN SEASON. Overall rating is supposed to show player's skill in the long run, over multiple seasons, over his entire league 'career'. These are two different things, don't you think? So why do you want player's seasonal rating to DROP DOWN when he is doing like expected?

I repeat again, if you do better than expected, your rating should rise. If you do worse, your rating should drop down. If you do as expected, it should stay the same. I hope you now understand why I have problems with your proposal.

I think all of your plans for changes come from the idea that seasonal rating should eventually approach overall rating. Did I get you right? But as I said, seasonal rating (why shouldn't even call it rating in the first place, but rather "performance", "score, or smth) should show us how players perform in a given season, not show the skill, that's what we have overall rating for.

I agree that seasonal ratings should be kept completely, 100% separate from overall ratings. If overall rating is used at all in the calculation of seasonal rating, then seasonal rating loses its importance, and becomes something more like quarterly performance. And this would again impact activity, which is about the highest concern we should have for a league.

If, as franz implies, assigning more importance to seasonal ratings diminishes the importance of overall ratings, then so be it; it's still a boon for activity and competition. Overall ratings will still be useful to look at, for those who want to look at them.

The essence of zero-sum is that the sum of points gained by winner and loser is zero. This is not a necessity IMO.

Think again what is rating supposed to show you - the skill. So after a series of games, say 10, the result is as expected (according to player's ratings), for example 8-2. The result tells us that both players played ok, according to their skill, so the sum of rating changes is 0. If it wouldn't be, the ratings would change whereas we said players performed as expected. Don't you think this distorts the picture a bit?

I think a zero-sum system isn't ideal, for worms at least. It seems like your problem is that non-zero-sum systems wouldn't be quite as objective, which is slightly true, right now you can have a good feel for a player's skill if you know they're at 1800, or 1600, or 1400. However, as a counter example, table tennis ratings are wonderfully accurate and employ a non-zero-sum system, as shown here: http://www.ratingscentral.com/HowItWorks.php. And I always understand almost exactly how good a table tennis player is just by their rating there. This system allows young unranked prodigies to gain say, 400 points for beating a high-ranked player, without that player going down too dramatically.

And over at NNNL, although everyone's points inflate as the season goes on, top players end up with roughly the same upper limit on their seasonal rating. You can definitely create non-zero-sum systems that yield ratings that correspond to intuition.

[HHC]

Ok let's say, we choose to reset everything every season. What about people who like to compete yet don't have time to play enough worms to be able to fight for playoffs? I'll give myself for example - I don't think I'll ever have enough time again to play ~100 leagues games in a season nor I wish to play that much. But I'd still like to have some comparison to other players. A rating table does that for me just fine, while fresh seasonal standings every two months simply don't.

I could even live with overall ratings and seasonal scores completely separated, as long as I have something to fight for. That's my biased view but I'm sure there are others around wormnet who feel the same way.

This is what my system does, it seperates season-scores from the overall-rating.
When you play a match your overall rating might go up 8 points (and opponent losing 8, according to old ELO), while your season rating may go up to say 45 points.

The overall ranks have no significance anymore for P/O's, instead they are used as a tool to see how skilled one's opponent is (and based on that, determine how valuable your win is).
However, since this eliminates the importance of the overall ranking I proposed the idea of the global TUS ranking in which the overall ratings of the different leagues are calculated together.. as to give people something to work on in the long run.


@Franz: can you explain to me why high overall rankings have a benefit in the season rankings in my system? I don't think that's the case to be honest. Random receives the same amount of points for beating Noob than say HHC (I have an overall rank of 1030 or something). Likewise, I lose as much points as Random does if I lose.
Like I said, the overall rating is merely used to determine your opponents strength. It has no influence on one's own stats.


Back to Rok

I think all of your plans for changes come from the idea that seasonal rating should eventually approach overall rating. Did I get you right? But as I said, seasonal rating (why shouldn't even call it rating in the first place, but rather "performance", "score, or smth) should show us how players perform in a given season, not show the skill, that's what we have overall rating for.

Season ranking should resemble the skill of a player in a given season.
Right now, the season points are way off because they are affected by the overall rankings. They won't be anymore in my system.

When I say that someone's season rating will approach his overall rating I'm merely saying that, based on his historical results, a player is expected to gain a certain amount of points. For example, right now, Uber's TEL overall ranking is 1700. In my system he will start the season with 1000 points as season rank. I expect him to win many games again, up to the point where he approaches his 'old' rank of 1700. At this point he will win/lose the same amount of points as he does now (as his new season ranking of 1700, which is used to calculate the points one gets, equals the 1700 overall ranking that was used to calculate his points in the old system).
This is what Mablak reffers to with his 'upper limit' on Season ranking. Uber will go up in points as the season progress, until he reaches his 'natural limit'.

Sooo, Uber is expected to finish a season with a rating of 1700 if he performs as expected. If he does better this season, he will end up with say 1800... if he does worse, 1600.
In the old system Uber would have 1000 season ranking if he performs as usual, 1100 if he does better, 900 if he does worse.

As you can see, in the old system the season rating is waaay off the overall rating, while in my system the season rating is expected (based on historical results) to be somewhere near his (usual) overall score. This is what I mean when I say that season is likely to approach overall.

The big advantage here is that we now can see who's really doing the best this season. Uber is no longer ranked 156th with 1000 and a brilliant record, but instead ranked where he should be: with 1700 points (prolly in the top10).

[franz]

@Franz: can you explain to me why high overall rankings have a benefit in the season rankings in my system? I don't think that's the case to be honest. Random receives the same amount of points for beating Noob than say HHC (I have an overall rank of 1030 or something). Likewise, I lose as much points as Random does if I lose.
Like I said, the overall rating is merely used to determine your opponents strength. It has no influence on one's own stats.

Hmm, yeah you are right.. I had it mixed up in my head just a bit.  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.
ok yea, I think I'm liking HHC's plan more now ;p

though Mablak explained to me the benefits of NNN's system, and it's very nice also. he strictly believes that 'the order in which games are played in a season' should not matter at all. this way, any good players joining a season late would not have to be avoided by higher ranked players ->  because as this good player climbs the ranks, all his early games that hurt a high ranked player wouldn't hurt as much anymore. I don't know how the system could be coded though or how easily it could be implemented. someone else should voice all this if they can.  if it's not a realistic solution though, I would feel just fine with the HHC way.

[Mablak]

Yeah, to elaborate, I think one of the most important changes that needs to be made to TUS is treating all matches between two opponents with equal importance; your total record against someone for the season is all that should be taken into account.

Let me attempt to list all the advantages of such a system:

1. Fairness. As it is, you could end up beating a great player 10 times and losing twice, but still losing points on the whole, if all your wins occurred when you were high-ranked and they were low. But simply put, you should always benefit when you've beaten your opponent more than they've beaten you in a season.

2. Starting late in the season doesn't give you an advantage. Right now, towards the end of the season there are players with tons of points, so if you wanted to climb the ranks with minimal effort, you could just start later in the season, getting a high rank with much less effort than many of the good players who were playing from the beginning. Simply because of the order of reporting, we've rendered games early in the season fairly useless, it's just more strategically sound to start playing later. But not everyone can do this, so the functionality of the league depends on the work of active players who decide to start early, which is quite unfair to them.

3. Avoiding is reduced. If you're not keen on playing a decent player because you know they're much better than their rating shows, you no longer have to be afraid of losing too many points. Standings are adjusted for you as new games are reported, giving you more points if someone you previously beat starts beating tons of players.

4. Less possible noob bashing. Franz pointed out that by simply playing a noob earlier in the season before other people do (before their points go down), you squeeze more points out of them than you deserve. Once again, standings would be adjusted as time goes on, and an early victory against a noob wouldn't give you many points in the end.

One objection to this system would be my claim 'you should always benefit when you've beaten your opponent more than they've beaten you in a season'. This isn't true if players are actually gaining or losing skill throughout a season, but it's fairly impossible to measure changes in skill over an extremely short time period, and I don't think any system is capable of that. The assumption we have to make here is that a player's skill level stays roughly the same over the course of a season. Though this isn't always true, it works very well for seasons of 2 or 3 months, and the benefits I've listed are well worth it.

[franz]

3. Avoiding is reduced. If you're not keen on playing a decent player because you know they're much better than their rating shows, you no longer have to be afraid of losing too many points. Standings are adjusted for you as new games are reported, giving you more points if someone you previously beat starts beating tons of players.

4. Less possible noob bashing. Franz pointed out that by simply playing a noob earlier in the season before other people do (before their points go down), you squeeze more points out of them than you deserve. Once again, standings would be adjusted as time goes on, and an early victory against a noob wouldn't give you many points in the end.

after thinking about this aspect more... all of it sounds well and good.. but this assumes that the 'good late joiners' and 'noobs' will actually play enough games to accurately reflect their skill level -> thus accurately affecting your own ranking. the problem is that it's very common for players to play just a few games in a season.. so without that historical element (overall rating) to give you proper benefit for beating a 'good late joiner' or 'noob' you still have imbalance in the system.

If you're suggesting a new plan make sure to think about what would you do about the current players stats.

The best idea or combination of the ideas will start from new season.
This is HHC's plan, please read it (long text, but won't bore you)

regarding what to do about the current players stats -> I see HHC's plan working/transitioning just fine with the same overall ratings that everyone has worked hard to achieve. but if a complete reset is something many people want, I could understand that too.  sounds like a good question that might want to be asked/polled depending which system is eventually chosen.

[Mablak]

3. Avoiding is reduced. If you're not keen on playing a decent player because you know they're much better than their rating shows, you no longer have to be afraid of losing too many points. Standings are adjusted for you as new games are reported, giving you more points if someone you previously beat starts beating tons of players.

4. Less possible noob bashing. Franz pointed out that by simply playing a noob earlier in the season before other people do (before their points go down), you squeeze more points out of them than you deserve. Once again, standings would be adjusted as time goes on, and an early victory against a noob wouldn't give you many points in the end.

after thinking about this aspect more... all of it sounds well and good.. but this assumes that the 'good late joiners' and 'noobs' will actually play enough games to accurately reflect their skill level -> thus accurately affecting your own ranking. the problem is that it's very common for players to play just a few games in a season.. so without that historical element (overall rating) to give you proper benefit for beating a 'good late joiner' or 'noob' you still have imbalance in the system.

But it's fairly redundant to take steps to make the seasonal ratings approximate the overall ratings, at a certain point they just become the overall ratings again. If you're looking for long-term accuracy, you might as well just look at the overall ratings. And I think incorporating any historical element would be harmful to activity, because certain people will start with advantages and disadvantages, which makes the prospect of a new season seem unappealing, since the whole concept is about getting a fresh start.

And your concern about people just playing a few games is valid, but because it's so few games, I don't think it can compromise the system greatly. The system would still be highly balanced if the majority of active pros play at least say, 15 or so games. In any case, I don't feel like the best players typically go a season with only a few games; when they know they can at least do the bare minimum to make it into playoffs, they often will, at least that's been the case in NNNL.

[zippeurfou]

Or you can do as it is in wow.
Everyone start at 0.
If you win a game you win point. If you loose a game you loose point.
If you're at 0 you do not loose point.
If you win and is at 0 you win a lot of point whereas your opponent does not loose too many point.
In wow it's like +96 and -15.After some time you get to find people your level and the amount of win/lose point is fair.