League General Discussion Thread

[MrRoboto]

@jkeller I’m sure your results in the past warrant you being placed in Tier M and it probably reflects your skill too.

The point still stands, though. Different players are handled differently because of the judgement of one single person. Now that moderator is luckily extremely fair, very benevolent and we can rely on him, but I think a system shouldn’t be dependend on having @gamerman01 around ;-)

But this raises an interesting issue, that’s absolutely up for debate: Should past results have any kind of influence on the current year? If yes, how big should that impact be?

Generally, there are two ways to think about it in an ELO system.

1. No influence at all. New year, Clean slate.
   This happens in most major sports: Basketball, Football, American Football, you name it. A new season starts and every past result is wiped clean, every team / player starts anew with the same clean sheet.

2. Retain some ELO rating from the past
   When the new season / year starts, the system could check the difference between the final ELO rating and the starting rating and then only move the new rating a certain percentage to the starting rating.
   For example: Starting Rating is 1500 for everyone. Player A finishes a year with 1200 and player B with 1900 Rating.
   We could say that instead of starting the new year at 1500 for everyone, we keep 40%.
   So A keeps 40% of -300 (1200-1500) and starts the year with 1380.
   So B keeps 40% of 400 (1900-1500) and starts the year with 1660.

Of course we could also use Option 2) but only for players above 1500, so that the lower skilled players have a new chance to start at the beginning, without baggage from last year.

I’m interested to hear the opinions of the community. Personally, I tend to lean towards option 2), because as @jkeller rightfully pointed out, we don’t have that many games here so this would balance this out a bit.

[MrRoboto]

By the way, this approach is used by many competitive online games.

Since climbing the ladder is time consuming and exhausting, players usually don’t fall back to square one when a new season starts, but fall back to certain safe thresholds for example. League of Legends or Hearthstone use similar systems for example.

[jkeller]

In chess, which probably is more similar to this as it’s an individual sport without much change from year to year( mental sport with no roster changes, coaching changes , injuries or much age based decline) you absolutely base this years initial rating on last years final rating. Why not? We need as much data as we can especially in this situation with extremely low sample size.

[MrRoboto]

Well if I’m not mistaken, chess sites like chess.com don’t even use seasons or yearly ratings at all. Isn’t it basically just a lifelong Rating, that you work on as long and as much as you are willing to?

Because in the end, for the reasons you stated, there would be no big reason to reset ratings on Jan 1st at all.
Do we want that? We could…

Actually, the more I think about it, the more inclined I am to agree with that sentiment.

[Martin]

Just brainstorming: could it make sense to have the curent rating always be based on the games finished during the last 365 days (rolling)? At least tennis players lose their tournament points after one year.

[MrRoboto]

Sure, everything is possible and I already know how to implement it.

Can you explain the upsides for that, @Martin ? I don’t see a big improvement over a rolling 365-rating instead of a calendar year rating.

I’d rather have some kind of decay for inactivity

[Martin]

@MrRoboto this was just an alternative to the “life long rating” which you mentioned.

[Stucifer]

I like the idea of a lifelong rating, as you mention about chess.com. I think we could use all the games we can get to influence ELO if league switches to that method.

[farmboy]

I’m not opposed to trying an ELO (or another) system either but would worry that the small sample size is going to mess it up more substantially. I do think past results should matter each new year (As far as I’m aware, in the current scoring, they just matter until someone has played 3 games and then their score is based on the outcome of the games played in the current year. So their effect is to increase the score of the opponent in the early games). And I also think we need a reset each year to ensure some variation in who gets to play in the league finals. A longer term ELO with the relatively small number of games played may make it harder for new (or much improved) players to break in.

My view is that if it isn’t (that) broke don’t fix it. I don’t doubt there are some issues with scoring. But we are a small amateur (and very niche) league with a small number of games played and with variation in the number of games played by player. Nothing is going to be perfect. And I don’t think changes in scoring will dramatically shift how the final standings look. Of course if we can use a method that is more streamlined, easier to use (not just for gamerman but for who comes after), I’m not opposed to that either. But the existing method of scoring doesn’t strike me as particularly onerous either.

And if we want to avoid having too many players ranked E or M one way to address it is to just to shift the thresholds where one crosses over. E could be set at 5 and M at 6 or higher for example and that would spread things out more.

[AndrewAAGamer]

If I recall correctly, at Days of Infamy, it was a lifelong record.

[MrRoboto]

I genuinely appreciate your feedback, @farmboy !
Always good to hear some different opinions.

Your idea of a fix would help combat the points inflation, for sure.
But as I said, that is the smallest of the issues. It would not fix any of the 7 major problems the current system has.
Now how broken is the system? I guess that’s subjective. It IS working, for sure. It creates a somewhat realistic ranking and most of the times the lower PPG player actually loses against the higher PPG player.
So in general, the higher the PPG, the better the player.

But in my opinion, that is a very low bar. If it didn’t fulfill this basic requirement, it wouldn’t be a working system at all. Our demand on a system should be higher than that, even if we are only an amateur league. We can still strive to be as professional as possible.

@farmboy said in League General Discussion Thread:

And I don’t think changes in scoring will dramatically shift how the final standings look.

This is the current ranking in the official spreadsheet.

This is the ranking in my automated spreadsheet:

But this is another ranking with the exact same rules. No formulas changed, no other entries, everything is the same. Just the order of calculations shifted.

Notice how gamerman went from rank 4 to rank 13!

Or another ranking. Again: same rules, same results, same formulas.

Now I don’t know about you but if a system can produce multiple different rankings depending on HOW you apply the rules, I personally would consider that system broken. We don’t even know who #1 is right now…

You did raise 2 concerns with an ELO system however and I want to address those:

The small number of games is offset in my system: The first few games have a bigger impact on ELO change. This gradually diminishes with the number of games until it settles at 10 completed games (exact number up for debate).
It would be completely negated by the way, if we use a lifelong ELO rating.

The other concern is how difficult it is for new players to enter the top ranks.
This is actually done super fast. If you win a handful of games against current top players, you will climb the ELO extremely fast and can reach the top spots.
I just tested it with my ELO system. A new player could claim the #1 rank in my ELO system after going 4-0 with 2 wins against top player Adam and another 2 wins against GeneralDisarray and ArthurBomberHarris.

That being said: How we choose the participants for yearly playoffs is another matter. We could only count the results of the calendar year. We could (and should) require a certain number of games played this year.
Right now the participants are largely the same group of players too. This would actually even change with ELO, for the better!

[farmboy]

@MrRoboto thanks for the response and explanation.

One point to clarify. I’m not worried that an ELO will prevent people from entering the playoffs. But I am worried that trying to score over the longer term will. As long as the scoring for the playoffs is primarily determined by one’s play in a given year, than it should be fine with or without an ELO.

I certainly don’t have a great understanding of the math that goes into this, but if the issue is that the rules will produce different results depending on the order of the calculations than can one solution be simply being consistent in the order of the calculations?

I do think given the small and variable number of games most of us play (and that we don’t play everyone else), we are always going to find that the rankings won’t quite match up with reality, but it has always seemed to me that we are pretty close. And that is good enough for me. But I’m certainly open to trying alternatives.

[Stucifer]

A tangent here, I have been thinking how bids factor into games. Playing a game as a new player, against an M- or E-ranked player, there should be a way to make an allowance for a larger bid for the lower-ranked player palatable. It currently is not in those players’ interest from a PPG standpoint, as Mr Roboto has pointed out; bid has no impact on the points scored in the game, and winning against a Tier 3 opponent will bring down your PPG average if it is over 4.

Perhaps there is a solution, where we use the terms/bid and “current” player ranking at the start of the game to determine the points the game will award to each player upon completion?

For example, in PtV the average bid is ~Axis +9. If an E-ranked player instead grants the Tier 3 opponent an increased bid of +18, perhaps a victory over that opponent should then result in 5 points instead of 4.

[gamerman01]

I am away from my computer and keyboard this weekend so won’t say much. But I do want to insist that the order in which scores are input is a minor one - it is really just a novelty observation. The primary goal has been that every player’s score can be calculated at any time and be verified, and this has always been the case. Any errors can always be found and corrected.
I don’t want to put my pride in the system above the enjoyment of the majority. I’ve been really busy moving to a new house and I’m trying to keep the championship game moving along. I will be evaluating what MrRoboto has very eagerly proposed and I envision a more automated and sleek system to start 2024 but I need to give guidance and direction before approving it and changing a lot of the league rules.
I am very confident that if I approve it, the great majority of you will be happy with it too.
My main goals are that another player besides myself is able to maintain it, that it’s fun, and that even with a limited number of games played, you have maximum insight into how strong your potential opponent is. -->

[gamerman01]

The objective has never been to be perfect or in line with other game ranking systems, certainly not chess or major League sports.
I understand human nature is to not be satisfied with the way things are, and that it is inevitable that some players will want to use their genius to try to make things bigger and better. Even though I think the tried and true system has been excellent and even though I do have long answers to each of the 7 statements nailed to my Wittenberg church door, I nevertheless am quite open to improvements if I don’t think things will overall be worse.
MrRoboto volunteered to learn to be my backup, and in no time at all was very eager to overhaul the system. Although this feels a bit like a coup (forgive me but I’m just being honest) I try not to be stubborn and I want to give the majority what they want. It is your league - I’ve just been the only player willing to be the moderator all these years.

[gamerman01]

No one has ever expressed any interest whatsoever in being a comoderator all these years even though it’s always been welcome. MrRoboto has also not yet expressed that kind of interest, to date.

Anyway, he is passionate about trying to overhaul the rznking system, and I’m open minded about that, but give me time. Please continue to post your thoughts, and I especially value those of longer term players, especially the more successful players, but all opinions are definitely considered!

[gamerman01]

Also I do not consider myself dictator for life! And if a solid moderator or 2 comes along, I would be happy to step down and pass the baton. If not, I am also very happy to continue being the last word on everything and organizing everything - it has always been well worth it!

[gamerman01]

But I will do what I can to never allow this league to fall into incompetent or nefarious hands - not in my lifetime.
These moderator discussion words are not about MrRoboto or the current discussions - at all - they just need to be posted once in awhile so that you can be assured about the league stability.

[MrRoboto]

Thanks for the words, gamerman.
Especially the last post: The fact that you take your position seriously and will do your very best to ensure the integrity of the league is invaluable. I think most if not all here are very grateful for what you do here.

I do understand that it feels like a coup to you, I can absolutely empasize with you! That’s actually the reason why I did not express interest in comoderator. In no way do I want you to feel threatened by me, that’s certainly not my intention!

There’s always tension when change is proposed. There is a reason why conservatives all around the world usually are the biggest group - as long as the current system is not totally broken that is.
Humans tend to be reluctant with change, when it’s working more or less.
And I do acknowledge that the current ranking system is doing fine. Otherwise it wouldn’t have sticked around all these years.

However, some of the concerns have been raised years ago. I can remember a discussion about winning against low Tier being bad for PPG a couple of years ago already. The idea of ELO has been swirling around the forums for years too.
While I did take part in that discussion and voiced my concern back then, I never fully went forward and took matters in my own hand, partly because I didn’t feel I’m in a position to go up against gamerman. It feels intimidating to challenge a dictators system, after all ;-) (using your word, not mine!)

So far, I have seen only people voting for an ELO system or people who are open for it, but none has spoken against it.
I am interested in more criticism, so that we arrive at the best possible outcome. Jkellers feedback for example convinced me that lifelong rating is more useful!

The most important thing at the end: I don’t want this to be a MrRoboto vs Gamerman thing. On the contrary, I want this system to be as simple and optimal as possible and with him continuing to lead all of us. I can gladly take care of managing / maintaining the rankings but I’m totally fine with just setting it up and gamerman then being in charge.

[MrRoboto]

@mr_stucifer
That’s an interesting concept.

Upside:
Results would reflect the feat of overcoming an unusually high bid and still win or losing despite having big bids.

Contra:
It complicates the system.
It introduces math and more tactics into the bidding process and might lead to games with fewer bids, despite higher bids being necessary (giving the statistics).