April 2023 Movement Discussion Thread

[gbwead]

17 minutes ago, DoubleJ said:

Changing the cut-offs in an arbitrary manner doesn't seem like the best path forward tbh.

What is then? 

I could give a break down of every single thing you could do, but I have done that more than 10 times already in the past 5 years and I just wasted my time everytime. 

[drewq]

10 minutes ago, DoubleJ said:

That doesn't answer my question though.

I think top players should have more influence, sure, but at some point it's hard to quantify. 

 

Pure conjecture because we obviously can't see games played for those not on ladder, but we know there are many people who will spam ladder for the sake of rewards or seasonal hats. Again, I have no way to see if this is a significant portion of games played, but it would be possible, no? i.e. RandomPlayer32 really likes the porygon hat and will play 300 games with a 45% WR to get it. 

 

 

[Godhelll]

Anyway, the "yo-yoing" will never stop insofar as the tiers will not stop changing.

 

I think it's just enough to make Pokemons down when we can and to go up others when necessary, that's all. There are pokemon that are not in their tier and this should be fixed, quite simply, I don't think it requires a colossal amount of work, but I could be wrong

[gbwead]

I want to clarify some mechanics regarding cutoff points. Right now, at the end of each season, all mons over 4.36% move up and all mons under 4.36% move down. What happens if we choose another number for our end of season cutoff point?

 

If the new cutoff point is lower than 4.36%, this will mean two things: 

1. The tiers will be larger. There will be more pokemon in OU and there will be more pokemon in UU.
2. There will be more movements (more volatility). With larger tiers, more pokemon will be subjected to potential movements each month because more poemon will be in reach of the cutoff point. 

If the new cutoff point is higher than 4.36%, this will mean two things:

1. The tiers will be smaller. There will be less pokemon in OU and there will be less pokemon in UU.
2. There will be less movements (less volatility). With smaller tiers, less pokemon will be subjected to potential movements each month because fewer pokemon will be in reach of the cutoff point.

 

There are pros and cons to increasing/reducing the cutoff point:

• Reducing the cutoff point means larger tiers which imply more pokemon will be played throughout all tiers (+ versatility / + fun), but this comes at the cost of an increased volatility in movements (unstability).
• Increasing the cutoff point means smaller tiers which imply less pokemon will be played throughout all tiers (- vesatility / - fun) to be creative, but this comes with the benefit of a reduced volatility in movements (stability).

Whether we increase or lower the cutoff point, nothing will change regarding yoyoing. If we increase the cutoff point, all mons hovering around the new number will yoyo at the end of each season between two tiers and the same logic would apply to a lower cutoff point. There are mons that will hover the cutoff point whether the number is 2%, 5% or even 10%. Changing the cutoff point has no real impact on yoyoing. 

 

The solution for yoyoing is spliting the cutoff point into two cutoff points: one cutoff point for mons moving up and another cutoff point for mons moving down. By creating an cutoff point interval, yoyoing will get reduced automatically. Mons hovering around a certain number would no longer yoyo up and down. If the move down cutoff point is 3% and the move up cutoff point is 5%, mons that are hovering around 4% would not move up and down at the end of each season. Mons would need to seriously spike up or spike down in usage in order to move from a tier to another. It means that when a mon is able to reach the rise up cutoff point, in order for that pokemon to drop down it would need to reach the drop down cutoff point. By making the move up and down cutoff point different, it will be significantly harder for mons to yoyo up and down.

With that being said, my recommendation is this:

 

Change the current cutoff point of 4.36% to a move down cutoff point of 3.75% and a move up cutoff point of 4.75%. 

 

This change would accomplish 3 things:

1. Slightly larger tiers (more options for lower tiers) 
2. Significantly reduced amount of yoyoing
3. Slightly increased volatility in movements (but considering a big chunk of our movements are yoyoing related, overall the amount of movements would still decrease)

 

 

 

Important note: I am only talking here about the end of season cutoff point in order to make this post as simple to understand as possible, but I strongly believe we should still have movements within each season and that OU/UU movements should not be synchronized with UU/NU movements. 

 

 

 

 

[Ziiiiio]

The cutoff can be a good reference, but I still believe that the final step requires members to vote, which can come from within the TC.

 

It needs think both checks and usage once a Pokémon drop to lower tier, For example: Rise up Clef and drop Venomoth. Then we need Gigalith go together to lower tier. But how about giga do not meet the cutoff, do we wait for it become lower usage? So it would be a flexible situation when we need tier change. My opinion is TC vote is necessary. And cutoff could be a good reference to start voting.

[RysPicz]

Just get rid of usage tiering and bring back good old power-based tiering. No yoyo-ing, more power in TC/ player's hands and much more work for TC

[caioxlive13]

12 hours ago, DoubleJ said:

What is your concern, that the top players (or those with most games played) have too much influence or too little influence on stats?

Let's be honest, they always have much influence on stats. They are top ranks. If they use a specific team and it works well, you can expect that players from 550- will also want to use that team, because they will think that will work with they too. 

[BaliAds]

12 hours ago, DoubleJ said:

Changing the cut-offs in an arbitrary manner doesn't seem like the best path forward tbh.

I don't think GB is saying to just change the number but change the way it currently works 4.36% as a fixed percentage to decide whether something moves up or down seems counter productive to what we're trying to achieve. There's no room for leeway and a pokemon will continue to 'yoyo' around that fixed point especially when you're considering lower tiers have significantly lower amount of games played than OU. If we used GB's above suggestion and shifted to a dynamic threshold it makes it a lot less likely for a pokemon to bounce back and forth as it is currently. 

I think you're right in saying setting an arbitrary number isn't a suitable way to adjust this if we wanted to we could shift the percentages to better adjust for tiers by adding a weighted average on to OU, UU & NU if we went down this route we would get larger cut-off points that take into consideration the amount of games played in the tier and also gives us different thresholds for each tier based on the usage which could help a lot with movements in the lower tiers.

[smadagos]

9 hours ago, gbwead said:

With that being said, my recommendation is this:

 

Change the current cutoff point of 4.36% to a move down cutoff point of 3.75% and a move up cutoff point of 4.75%. 

This.
Moving up and down shouldn't have the same cutoff point. When a mon drops, it must be a solid reason to make it moving back up again.
Also not making tier changement at the same time between OU/UU and UU/NU will let the tiers adapt to new threats before making absurd changes based on usage of a different tier than it currently is. 

[TohnR]

19 hours ago, drewq said:

In NU at the moment there are several players at or near 1,000 battles played on ladder. 

 

The thing is they have 1000 battles over the whole 3 months season.
You can't use that number and compare it to June's amount of matches, you should compare it to whole season.
And yea it has always been the case with our usage system that the 50 most active players in UU/NU account for most of the usage, that's just the effect of having a small sample size.
Just so you know, when I joined Pokémmo in 2018 we had no ladder for these tiers and usage system was based purely off monthly tournaments, which made even less sense.
You could have a mon raise because it had 19 usages in the month :'D
It has been an issue for a long time, and we can't really improve that unless we stop usage tiering which imo would be much worse.

[gbwead]

Iirc, the leaderboard (top 100) of each tier represents constantly 20% of all usage data gathered. That's what I calculated a few months ago, but I'm not sure if that remains the case.

[caioxlive13]

3 hours ago, gbwead said:

Iirc, the leaderboard (top 100) of each tier represents constantly 20% of all usage data gathered. That's what I calculated a few months ago, but I'm not sure if that remains the case.

Directly is something around that. If you include indirectly, at least +60% of usage you can include. 

 

I've passed on almost all elo ranges on OU, except Top Rank. No matter the level, the amount of people copypasting teams is absurdely INSANE. Some people play well, other badly and other just robotic moves.

[Munya]

Quote

With that being said, my recommendation is this:

 

Change the current cutoff point of 4.36% to a move down cutoff point of 3.75% and a move up cutoff point of 4.75%. 

This is within a close enough range to what I was thinking, .25% off(was thinking 5 and 4) but this may actually be better as it keeps something I used as an example for something that would still yo-yo if we just had a straight 5% cutoff in OU(torkoal).  This month though we are probably retaining normal seasonal cutoff and going with it after asking the TC.

 

Something that was suggested internally within TC is this in the future:

1st month 1.7 to drop, no rises

2nd month 2.5 to drop no rises

3rd month 5 to drop or rise.

 

Which is where I brought up the yo-yoing thing and why a split would be better.  The above is apparently similar to how smogon handles it now?  I didn't look into the claim.