Arena rating decay for inactivity

[OEJack]

I'll repeat what i said above - why should good performance earlier in the season be penalised compared to the same performance later in the season?

Why does me parking stop you climbing the ladder after me? I had to beat people to get to the top - so will you if you want to.

If you introduce rating decay all I'll do is not bother with arena at all until the last few weeks, so you still won't get to play me any more often. I'll just play the last 3 weeks instead of the first 3.

Those are exactly the right questions and get to why there should be decay. A good performance earlier in the season should be penalized compared to later performances because the field is not the same on day 1 versus day 90. There is an inherent advantage you receive by being at/near the top when it comes to having to start a new season, i.e. you have to win fewer games to reach top. Due to the nature of the early season rankings, I would also argue that there is more rating to be had playing those games. This is supported by looking at the number of 1800+ ratings after 2 weeks of the season compared to now (if I recall correctly it was 6 or 7 versus 0). Where did those ratings go? They went in to the great sea of averaging thanks to (and I am sure in some cases regretfully) the willingness of the top rated players to risk losing. There is also the factor of player rating averaging out over time, and while I would not argue that it is true in every case because players are free to join the season at any point and play their first games, I would say that player ratings more accurately reflect their 'true' average later in the season and therefor penalties and rewards for games against them more accurately impact overall ratings later in the season

The rating award system stops me from climbing the ladder after you. Don't get me wrong, even with a decay system I think the rating penalty/reward ratio needs to be tweaked. If the top X number of players do not provide the opportunity to beat them then the problem skews even worse. Let's say you gain 4 points for beating a higher rated player and lose 1 for losing and conversely lose 4 for losing to a lower rated player and gain 1 for winning. How are you ever going to move up if the players ranked above you don't play? (the numbers are actually worse but I wanted to simplify). Do you have to win 80+% of your games just to move up?

If like you posit it does not matter when you play, then why does the overall rating of the top 20 shrink noticeably after 2 weeks? If you truly believe it doesn't matter when you play, presuming you cannot be convinced to just play regularly throughout the season, I welcome you to only play the last 3 weeks instead of the first 3. I think you will be surprised with the difference. If nothing else, it will help address the issue that some people have raised about matches being harder to find later in the season.

[dschingis27]

There is an inherent advantage you receive by being at/near the top when it comes to having to start a new season, i.e. you have to win fewer games to reach top. Due to the nature of the early season rankings, I would also argue that there is more rating to be had playing those games.

I absolutely don't get your point. I don't think it's true that it is in principal easier at season start to get a high rating. It might be easier to win at season start because many more average players actually play in Arena, while over the course of the season, the field of regular players can shrink down to the "pro players".

Do you have to win 80+% of your games just to move up?

Yes, usually you have to do exactly that in order to get to the very top. It's a myth that most top-ranked players get to the top by winning against next top-rated players. Usually, the top-ranked players have had an exceptional streak of luck and high-level play, but essentially against players from many different ranks. It's not like the best players match that often in Arena.

If like you posit it does not matter when you play, then why does the overall rating of the top 20 shrink noticeably after 2 weeks?

Could you give examples of games where that happened? I never noticed it and it seems odd to me.

[OEJack]

There is an inherent advantage you receive by being at/near the top when it comes to having to start a new season, i.e. you have to win fewer games to reach top. Due to the nature of the early season rankings, I would also argue that there is more rating to be had playing those games.

I absolutely don't get your point. I don't think it's true that it is in principal easier at season start to get a high rating. It might be easier to win at season start because many more average players actually play in Arena, while over the course of the season, the field of regular players can shrink down to the "pro players".

When you finish the season ranked higher you start the new season ranked higher. Higher ranked players don't start at the bottom (zero stars).

Do you have to win 80+% of your games just to move up?

Yes, usually you have to do exactly that in order to get to the very top. It's a myth that most top-ranked players get to the top by winning against next top-rated players. Usually, the top-ranked players have had an exceptional streak of luck and high-level play, but essentially against players from many different ranks. It's not like the best players match that often in Arena.

The exact numbers weren't the point here. I have been inside of the top 15 (I think top 10 but don't know where to confirm) the past two seasons (the only ones I have played) so I have first-hand knowledge of what it takes to get there. The point is that winning more than you lose should move you up and winning more than you lose against higher rated players should move you up even higher. This was more of a commentary on the impact of the reward/punishment ratio which if tweaked could help persuade higher ranked players inclined to 'park' their ratings to get back to playing.

If like you posit it does not matter when you play, then why does the overall rating of the top 20 shrink noticeably after 2 weeks?

Could you give examples of games where that happened? I never noticed it and it seems odd to me.

This does not reference any particular game it refers to the ratings at a particular point of the season. It has happened in both of the past two seasons.

[dschingis27]

When you finish the season ranked higher you start the new season ranked higher. Higher ranked players don't start at the bottom (zero stars).

Everyone starts with an EAS of 1500 at the start of a new season. There is absolutely no advantage here in starting in a higher league.

The point is that winning more than you lose should move you up

That is not at all how leagues work. You don't automatically move up when you win more than 50%. You have to win more than the opponents you want to overtake. And the opponents in the Top10 have won a lot in a row. It's not possible that everybody moves up. I mean, we can have infinitely increasing scores, but then you give the medals automatically to the ones who play the most.

This does not reference any particular game it refers to the ratings at a particular point of the season. It has happened in both of the past two seasons.

You could still simply tell me which game(s) (Azul or Catan or what?) you have played in Arena. Maybe this was a misunderstanding.

[OEJack]

You could still simply tell me which game(s) (Azul or Catan or what?) you have played in Arena. Maybe this was a misunderstanding.

My mistake. I linked here through the Can't Stop guru page and did not realize this was not a Can't Stop specific topic. Please filter all of my remarks to apply to Can't Stop and the Can't Stop arena season.

I'm not sure what the technical name is for the system but I am referring to the rankings Bronze>Silver>Gold>...>Elite. I started the season not having to work my way from the bottom (Bronze 0 stars) as I did initially. That meant I had to play/win fewer games to reach Elite status. Owing to the nature of the severe penalty for losses fewer games played (aka less exposure to the chance of losing) means an advantage for those players who play fewer games to reach higher status.

I mean, we can have infinitely increasing scores, but then you give the medals automatically to the ones who play the most.

I disagree with your reasoning. Playing does not equate to winning. Your math does not add up. In the current reward/penalty system, playing aligns more with losing which is why a penalty for not playing is being considered. Players sitting at the top (speaking loosely here) choose to park and not play rated games rather than expose their ranking.

I understand the temptation, especially if you are sitting at number 1. I considered it myself after hitting 12th out of 6,000+. I just enjoy the arena format too much to 'take a knee' for months. If I was sitting at 1 I might have felt differently. If I knew that there was a decay system in place pushing me to play or lose rating and watch as I slipped out of the top spot, I assure you I would keep playing. It checks both boxes, enjoying the game and having a chance to stay atop the leaderboard. As it is, I'm 2 games under .500 for my last 18 games and have slipped to somewhere around 600. I think that is a bit excessive and should be adjusted but I get it and accept it and it is only tangential to this topic. There are however players out there who would rather take that knee rather than risk being in my position.

Rewarding players for not playing just doesn't seem on brand for a gaming service. Even if you disagree with the premise of that statement, there is no denying that the decay system would encourage play. I think that should be the goal not accommodating the non-playing strategy adopted by a few players. Will a few players be upset that their (non)playstyle gets called out? I have no doubt. Will they get over it? I'm guessing so, and we will all be better off in the form of more games played.

[estroncio]

(...)

And another aspect. Stopping play in Arena after a lucky streak is just the correct competitive idea for heavily luck-based games. Even a (slight) point decay will not change this tactics at all.

In a strategic game the best players have a strong incentive to play on, even if they reach a top spot. Because they can expect to win by their superiority. In luck based games this is not the case - at least not in this extend.

Thus, one would need to discuss the rating decay separately for different games.

You hit the nail on the head there. Decaying doesn't look like a good solution to me though.

---

What we need is to dampen the score you get after a game based on the luck factor of the game.

Let's say you play coin toss game (luck factor = 1), why would you get -37 if you loss against a "weak" opponent and only get +1 if you win. It is a totally random game so you should get the same reward/punishment.

Now if you play chess or hive (luck factor = 0), it is reasonable you get -37 for losing against a weak opponent and +1 if you win. Or the other way around.

Code: Select all

Reward after losing/wining a game against weakest possible opponent
(somehow made up numbers... just an example)
+-------------+------+------+-----+
| Game        | Luck | Lose | Win |
|-------------|------|------|-----|
| Chess       |  0.0 |  -39 |  +1 |
| Hive        |  0.0 |  -39 |  +1 |
| Patchwork   |  0.3 |  -34 |  +6 |
| Carcassonne |  0.4 |  -32 |  +8 |
| Catan       |  0.5 |  -30 | +10 |
| Coin Toss   |  1.0 |  -20 | +20 |
+-------------+------+------+-----+

BGA people should have enough data to calculate a sensible luck factor for each game.

That dampen score would produce a ranking with less oscillations.
You'd be always encouraged to play because scores are fair, you wouldn't sink just because you are unlucky in a few games. Likewise, you wouldn't reach the top just by being lucky against a stronger opponent.

(Better late than never )