Warrior General Games

Originally posted by InRomoWeTrust
That's EXACTLY what I was arguing for (adjusting the k-value takes care of that).

What I was arguing against is a system where Minnesota goes 5-0 and Queen City goes 5-0. Queen City had a tougher 5 games so Queen City is ranked above Minnesota. Completely ignoring that Minnesota has won it's last 30 games (or w/e you want for the hypothetical )

With the probable current elo gap between you and Minny, the K values would have to be ridiculously large for that to happen in just 5 games. So large I'm going to say it's basically not possible.

Originally posted by InRomoWeTrust
So in reality, if the k-value was appropriately adjusted and we simultaneously could see the ELO values, we'd have no problem with the system.

I understand what you're saying, but I wouldn't quite go that far because ...

Originally posted by Galithor
Elo works for mighty brawlers because of the never ending nature of the wins/losses. GLB2 has definitive start and end points every 30 games where the nature of teams get signficantly altered. ELO does not make sense for that structure.

ELO has a lag. Yes it self-corrects but it takes time to correct.

The problem is this:

In season, ELO works (slowly). Adjusting k values will make it move faster, but k values will never fix the below (In-between seasons).

In-between seasons, ELO is anywhere from correct (a team that has no roster changeover) to completely fucked (owner goes inactive and entire roster becomes CPU at 50 chem).

This is why Chemistry should be used to adjust ELO as it's the ONLY mechanic that deals with Roster turnover at the end/beginning of a Season.

A team of 43 players at 100 chem has a Chem Value of 4300. If 10 players swap out at 50 Chem, then the starting Chem Value of the next season is 3800. That's an 11.6% decline. The team's ELO could be adjusted be that much, or a completely different amount but using that 11.6% decline as a base.

If a team went full CPU, there would be 43 players at 50 chem equaling a Chem Value of 2150 for the Team. This -50% decline in Team Chem Value would lower ELO. Again the amount ELO is reduced needs to include this as a base but probably needs some type of adjustment factor.

So the problem isn't limited to intra-season ELO, it's also inter-season ELO.

If we fixed the latter, we may not need to fix the former. But everyone is focused on fixing the former first instead of the more obvious problem - the latter.

We all want Dead Teams to go down in Ladder ranking as fast as possible. We all have different views as to how fast competitive teams should rise/fall in the Ladder.

This is why in-between season ELO adjustments need to be made. If your team goes CPU, you get adjusted immediately and harshly. I don't see any disagreement in that.

Originally posted by Absolut Zero
With the probable current elo gap between you and Minny, the K values would have to be ridiculously large for that to happen in just 5 games. So large I'm going to say it's basically not possible.

I was referring to Galithor's post. Scenario had nothing to do with current rankings.

Originally posted by Xars
ELO has a lag. Yes it self-corrects but it takes time to correct.

The problem is this:

In season, ELO works (slowly). Adjusting k values will make it move faster, but k values will never fix the below (In-between seasons).

In-between seasons, ELO is anywhere from correct (a team that has no roster changeover) to completely fucked (owner goes inactive and entire roster becomes CPU at 50 chem).

This is why Chemistry should be used to adjust ELO as it's the ONLY mechanic that deals with Roster turnover at the end/beginning of a Season.

A team of 43 players at 100 chem has a Chem Value of 4300. If 10 players swap out at 50 Chem, then the starting Chem Value of the next season is 3800. That's an 11.6% decline. The team's ELO could be adjusted be that much, or a completely different amount but using that 11.6% decline as a base.

If a team went full CPU, there would be 43 players at 50 chem equaling a Chem Value of 2150 for the Team. This -50% decline in Team Chem Value would lower ELO. Again the amount ELO is reduced needs to include this as a base but probably needs some type of adjustment factor.

So the problem isn't limited to intra-season ELO, it's also inter-season ELO.

If we fixed the latter, we may not need to fix the former. But everyone is focused on fixing the former first instead of the more obvious problem - the latter.

We all want Dead Teams to go down in Ladder ranking as fast as possible. We all have different views as to how fast competitive teams should rise/fall in the Ladder.

This is why in-between season ELO adjustments need to be made. If your team goes CPU, you get adjusted immediately and harshly. I don't see any disagreement in that.



As if chemistry wasn't already a big enough headache, you want to further punish teams for making offseason moves? Of course you would love the rankings to be based on chemistry because you will never have to deal with it. Meanwhile, a team like Harrisonburg faces constant roster turnover but continues to churn out victories. They are already punished enough by having their players not playing at full potential until later in the season. No reason to further punish them and other competitive teams by dropping their ranking based on roster changes.

Originally posted by Xars
The problem in GLB2 is that the "INPUT" isn't the same is it is in Chess/whatever. Teams change season to season. Players leave, there are Chemistry hits, Owners go inactive, etc.

So when the old INPUT isn't similar to the new INPUT, then the ELO score is false. Could be a little false or a lot false.

Yup, definitely. There's also the problem of brand new Vet teams that just came up from Pro. They're at a massive ELO disadvantage because of at least one (probably more) full season(s) with fewer games played, so their total ELO score will always be lower. For a team like Minny or HBurg who have been up here awhile, it'd take a long time for new Vet teams to reach their massive ELO score.

I think scaling every teams ELO, down or up, towards a universal average is the next best step for fixing the current ELO problem. Not sure if it should be a universal average or tier-wide average. Universal average keeps the existing ladder ranks the same. Tier-wide average would condense each tier's ELO, some ladder rank movement between tiers, but would also prevent the new Seasoned teams from playing really good Pro teams. It gives every team at a fairer shot.

Example:
Let's say Stunners (a team who has had many seasons to stockpile insurmountable levels of ELO) has an ELO of 2200, and MBR (brand new team to Vet) has an ELO of 1700. Assuming they're about equal with on-field play, it'd take a really long time for Team Y to catch up to Team X, even if the K value is increased dramatically. And since they're about equal with on-field play, MBR could never catch up to the Stunners in one season.

Let's say the average ELO is 1200, we'd need to find an off-season scaling that lets MBR have a shot at the Stunner spot. The first time scaling happens, it would have to be a big percentage, but in the ensuing off-seasons, the percentage could be a modest amount.

So first off-season of scaling has to be large to get rid of the insurmountable ELO stockpiles, let's say we do 50%. Stunners ELO then drops down to 1700, and MBR's drops down to 1450. It's still a large difference, but one that gives MBR a chance to catch up to. The second (and every off-season afterwards), the scaling could be lowered to 25% and it'd give fairer ladder movement.

Originally posted by Xars
It's why I recommended changing the ELO score of a team when their was high roster turnover. (Thread I got torched on.) But perhaps people are now realizing I was right.

...

I still think my way of using Chemistry values to adjust ELO makes more sense than penalizing player performance on the field, but in order for that idea to get adopted all of you need to catch up to me.

If the ELO adjustment you suggest can be positive or negative, then it seems ridiculously complicated to do correctly, and for just Corndog to figure it out, basically impossible. It would require every player to have some sort of ELO value, which is difficult because there are 10 other teammates on the field indirectly influencing each other.

If the ELO adjusment is only negative, a team could be hurt from a reduction in ELO, despite an improvement in their roster from the prior season. That doesn't seem fair.

Originally posted by Rob.
As if chemistry wasn't already a big enough headache, you want to further punish teams for making offseason moves? Of course you would love the rankings to be based on chemistry because you will never have to deal with it. Meanwhile, a team like Harrisonburg faces constant roster turnover but continues to churn out victories. They are already punished enough by having their players not playing at full potential until later in the season. No reason to further punish them and other competitive teams by dropping their ranking based on roster changes.



You haven't been reading my posts on Chemistry. I'll re-iterate.

I don't want any player to have ANY negative effect on the field due to Chemistry.

Chemistry is supposed to prevent dropping players mid-season by the Ownership/GM/staff. Instead, if a player gets dropped, the Team Chem Value would go down. This is would lower the Ladder ranking of a team. I'm not sure how else to penalize someone for being a dick.

But the Agent of the Player that is cut shouldn't be the one who gets screwed. And he IS the one who gets screwed with Chemistry, because once he does sign on to a new team, his player will play at a reduced level.

Finally, I've stated that single-agent teams like mine won't ever have to deal with Chemistry. I've also stated it's bad for the game. I don't want to have the play style I do. I have the play style I do because of Chemistry.

Originally posted by Rob.
Meanwhile, a team like Harrisonburg faces constant roster turnover but continues to churn out victories. They are already punished enough by having their players not playing at full potential until later in the season. No reason to further punish them and other competitive teams by dropping their ranking based on roster changes.

And as to HBurg and the like. Under my concept, they would suffer NO ON FIELD DEGRADATION OF PERFORMANCE DUE TO ROSTER CHANGEOVER.

They would have an ELO adjustment downward (but a lot of Vet teams would because at least 1/4th of Vet goes Legend every year) due to inter-season ELO adjustments.

But then during the course of 30 games WITH NO ON FIELD DEGRADATION OF PERFORMANCE they should easily be able to claim the #1 Ladder spot. That's intra-season ELO which is the k value everyone talks about. And that needs to be high enough so a team that re-stocks it's roster for a good reason (re-loading) instead of a bad reason (CPU) doesn't suffer any negative effects by season end.

As far as "new to vet" teams go. As Pro's in season 7 - the stunners went 31-1. Only loss was to #1 vet Hawaii Bulls. We beat at least a few other #1s that season and several top 10's.

Then in season 8 (our first at Vet) - we went 28-3 with wins over many top 10's and #1's -- and we finished at #3 overall.

It would be pretty difficult imo for a 1st year vet to hit #1 under old ELO rules...
But as the number of Vet teams seems to continue to decline - it could be possible

Originally posted by Absolut Zero
If the ELO adjusment is only negative, a team could be hurt from a reduction in ELO, despite an improvement in their roster from the prior season. That doesn't seem fair.

Remember, there's inter-season ELO effects and intra-season ELO effects.

Basically, you have this choice:

ELO can be correct on Game 1 of a season.

ELO can be correct on Game 30 of a season.

Which do you choose? (You can't have both.)

Originally posted by Xars


You haven't been reading my posts on Chemistry. I'll re-iterate.

I don't want any player to have ANY negative effect on the field due to Chemistry.

Chemistry is supposed to prevent dropping players mid-season by the Ownership/GM/staff. Instead, if a player gets dropped, the Team Chem Value would go down. This is would lower the Ladder ranking of a team. I'm not sure how else to penalize someone for being a dick.

But the Agent of the Player that is cut shouldn't be the one who gets screwed. And he IS the one who gets screwed with Chemistry, because once he does sign on to a new team, his player will play at a reduced level.

Finally, I've stated that single-agent teams like mine won't ever have to deal with Chemistry. I've also stated it's bad for the game. I don't want to have the play style I do. I have the play style I do because of Chemistry.



Sorry, I'm at work and all your posts are novels

So if there is no on-field penalty I can still see people being dicks and cutting people midseason or even late in the season. A lot of people don't care about ladder rank and just care about league trophies.

I don't mind elo for managing the rankings during a season. the problem is the Season-2-Season transfer of ranking value.

Even if your roster hasn't changed, you're still susceptible to buffs/nerfs from game changes each season.

Originally posted by Galithor
I don't mind elo for managing the rankings during a season. the problem is the Season-2-Season transfer of ranking value.

Even if your roster hasn't changed, you're still susceptible to buffs/nerfs from game changes each season.

Which I think can be solved by readjusting the elo scores each offseason. Remove the large gaps that occur when a bunch of teams fold. Give teams a generic starting point.

For example:

#1 - 4000
#2 - 3900
#3 - 3850
#4 - 3800
#5 - 3750
etc.

Also - a major penalty for teams based on the % of their roster that is human (to knock CPU teams way down...still not perfect but weeds out some)

Originally posted by TxSteve
Also - a major penalty for teams based on the % of their roster that is human (to knock CPU teams way down...still not perfect but weeds out some)

That's a good idea.

Originally posted by TxSteve
Also - a major penalty for teams based on the % of their roster that is human (to knock CPU teams way down...still not perfect but weeds out some)

yes, great idea