D400 (Official Clan Ranking) - Oct 2021 ranking

[Swimmerdude99]

But, I maintain that no clan should lose points if they are perfect in a war. You will not change my mind on that.

Based on above explanations this means YOU should go write a new algorithm and collect all the data for it Otherwise you kinda just blowing smoke and restating something that obviously won't/can't be fixed without an overhaul. Someone really bothered by this would offer their time to take it upon themselves and run it

I think a lot of people are in agreement with you!

[JPlo64]

But, I maintain that no clan should lose points if they are perfect in a war. You will not change my mind on that.

Based on above explanations this means YOU should go write a new algorithm and collect all the data for it Otherwise you kinda just blowing smoke and restating something that obviously won't/can't be fixed without an overhaul. Someone really bothered by this would offer their time to take it upon themselves and run it

I think a lot of people are in agreement with you!

Well, to clarify what I said previously, a sure fire fix couldn;t happen.

HOWEVER,
I would be MORE than happy to produce rankings that limits this problem.
I have identified this as an issue long ago. While I do think the basic formula used for the rankings is good, I think that the original creators of the F400 made poor choices on some of their constants that are used.
Donelladan was kind enough to create his D400 program such that I can adjust different parameters/constants in the formula.

Below is a result of April rankings which I find better. With these constants, the MAX score = (Opponents Score)+1000

show: a100 b900

show: More explanation

[groovysmurf]

Awesome and great explanation! This can be the JP6400.

[Caymanmew]

I prefer the way you calculate these new rankings over the D/F400.

The margin of winning / losing should have much more significance than simply winning. Although this doesn't fix the "problem" it does reduce the frequency it seems.

[rockfist]

But, I maintain that no clan should lose points if they are perfect in a war. You will not change my mind on that.

Based on above explanations this means YOU should go write a new algorithm and collect all the data for it Otherwise you kinda just blowing smoke and restating something that obviously won't/can't be fixed without an overhaul. Someone really bothered by this would offer their time to take it upon themselves and run it

I think a lot of people are in agreement with you!

Well, to clarify what I said previously, a sure fire fix couldn;t happen.

HOWEVER,
I would be MORE than happy to produce rankings that limits this problem.
I have identified this as an issue long ago. While I do think the basic formula used for the rankings is good, I think that the original creators of the F400 made poor choices on some of their constants that are used.
Donelladan was kind enough to create his D400 program such that I can adjust different parameters/constants in the formula.

Below is a result of April rankings which I find better. With these constants, the MAX score = (Opponents Score)+1000

show: a100 b900

show: More explanation

Excellent!

[JPlo64]

THe problem with F400/D400 is that A is too large relative to B. TOo much emphasis on just winning/lowing the war as opposed to margin of win/loss.
This is why, in general, league hurts the ranking of top clans.

To further illustrate this problem,

Take a 41 game war vs Clan with score 1000. Below are what the relative war score's are for a few different results with the traditional F400/D400

Win 41-0: 1600
Win 21-20:1209.75
Lose 20-21: 790.24
Lose 0-41: 400

The marginal score difference between undefeated and winning by 1 is 390.25
The marginal Difference between 21-20 and 20-21 is 419.51

That is rather absurd.

[Keefie]

....

[josko.ri]

I disagree with this change. Winning or losing itself is more important than the margin, as it should be in every sport.

Even with this adjustment you still didnt adress issue of flawless winning because the lowest rankings are aroud 400 and the highest ranking are more than 1400 which is more than 1000 difference.

[josko.ri]

I also challenge the claim that flawless win shouldnot penalize winners. If TOFU wins, say, 1-0 vs BOB, what should then be TOFU's score? 1500? Of course not. So the same question comes for 2-0, 3-0 etc. And then we come to optimization in which number of games should the score be higher.

Flawless wins also have ots own limitations as described above and current limitations are good enough. Even in the new proposed system of A100 and B900 it can happen that point difference between top and low clan is more than 1000 so this solution does not address question of "flawless win" but still makes new mess in the ranking system because clans were making their gaming plans according to the old system where win itself is more important than margin.

[josko.ri]

In all competitions of CC wins count and margins are only used for tiebreaker. Wins should be the most important parameter, as it is right now.

[josko.ri]

Also, look at paradox which new system brings. With the new system, first two ranked are FALL and TOFU who did not win any first class CC competition in last one year. With the current system, higher are ranked S&M, LHDD and A, who won CCup, RL and CL in last one year. Therefore, current system more evaluates perfection, which is actually winning CC events while playing versus the best of the bests. However, thenew system does notevaluate thatbutinstead evaluates high winning margins versus some of lower clans, which is irrelevant in terms of perfection.

[josko.ri]

THe problem with F400/D400 is that A is too large relative to B. TOo much emphasis on just winning/lowing the war as opposed to margin of win/loss.
This is why, in general, league hurts the ranking of top clans.

To further illustrate this problem,

Take a 41 game war vs Clan with score 1000. Below are what the relative war score's are for a few different results with the traditional F400/D400

Win 41-0: 1600
Win 21-20:1209.75
Lose 20-21: 790.24
Lose 0-41: 400

The marginal score difference between undefeated and winning by 1 is 390.25
The marginal Difference between 21-20 and 20-21 is 419.51

That is rather absurd.

This is how the real sports works. Wins count regardless of margin. Winners are always remembered in history (even if win is only by one point) but margins are only rarely mentioned.

[Extreme Ways]

I also challenge the claim that flawless win shouldnot penalize winners. If TOFU wins, say, 1-0 vs BOB, what should then be TOFU's score? 1500? Of course not. So the same question comes for 2-0, 3-0 etc. And then we come to optimization in which number of games should the score be higher.

Flawless wins also have ots own limitations as described above and current limitations are good enough. Even in the new proposed system of A100 and B900 it can happen that point difference between top and low clan is more than 1000 so this solution does not address question of "flawless win" but still makes new mess in the ranking system because clans were making their gaming plans according to the old system where win itself is more important than margin.

Not going to respond to JP and post relating to JP's because I haven't read the explanation, however:

a 1-0 win should be a positive result for the winning clan regardless of relative rank but should have negligible weight. The more games are played, the more weight is attached. We can also reason the other way around (and bear with me because I still dont know the maths behind it): if clan A wins 200-0 vs clan B, but clan B is the shittiest clan available, should clan A then still lose points?

I also dont think any clan with maybe the exception of TOP plans their wars according to D400/F400 ranking.

[Extreme Ways]

THe problem with F400/D400 is that A is too large relative to B. TOo much emphasis on just winning/lowing the war as opposed to margin of win/loss.
This is why, in general, league hurts the ranking of top clans.

To further illustrate this problem,

Take a 41 game war vs Clan with score 1000. Below are what the relative war score's are for a few different results with the traditional F400/D400

Win 41-0: 1600
Win 21-20:1209.75
Lose 20-21: 790.24
Lose 0-41: 400

The marginal score difference between undefeated and winning by 1 is 390.25
The marginal Difference between 21-20 and 20-21 is 419.51

That is rather absurd.

This is how the real sports works. Wins count regardless of margin. Winners are always remembered in history (even if win is only by one point) but margins are only rarely mentioned.

Say we take tennis as an example, should you then also not be able to lose rating points if you win 21-20 vs any other clan?

[Keefie]

I also dont think any clan with maybe the exception of TOP plans their wars according to D400/F400 ranking.

That's simply not true and very unfair on TOP..

I challenged ICON last year, when we were the 11th ranked clan on approx 1070 points. We were turned down because of our ranking. The Republic wanted Venom for their first war. We were planning a war with LoW so turned them down. I believe LoW also declined. I was left with the impression that they didn't want to drop too far down the rankings as close win could be detrimental to their ranking points.

[Extreme Ways]

I also dont think any clan with maybe the exception of TOP plans their wars according to D400/F400 ranking.

That's simply not true and very unfair on TOP..

I challenged ICON last year, when we were the 11th ranked clan on approx 1070 points. We were turned down because of our ranking. The Republic wanted Venom for their first war. We were planning a war with LoW so turned them down. I believe LoW also declined. I was left with the impression that they didn't want to drop too far down the rankings as close win could be detrimental to their ranking points.

It's unfair to current TOP, I agree, However, in the beginning of TOP I strongly feel like they dodged strong clans and abused the way F400 worked to get to #1 there.

And maybe I'm naive on other clans not caring that much about D400?

[josko.ri]

Also, look at paradox which new system brings. With the new system, first two ranked are FALL and TOFU who did not win any first class CC competition in last one year. With the current system, higher are ranked S&M, LHDD and A, who won CCup, RL and CL in last one year. Therefore, current system more evaluates perfection, which is actually winning CC events while playing versus the best of the bests. However, thenew system does notevaluate thatbutinstead evaluates high winning margins versus some of lower clans, which is irrelevant in terms of perfection.

This is my the strongest argument against the new proposed system.

[JPlo64]

This is how the real sports works.

Any argument that is essentially, "This is better b/c that's how they do it" is usually a bad argument.
Particularly so when relating Conquer Club Clan Games and Wars to sports. The nature of a clan war is completely different than a game/match of the sports you are referring too.

Wins count regardless of margin. Winners are always remembered in history (even if win is only by one point) but margins are only rarely mentioned.

It's a matter of preference in what matters more, I guess. The biggest issue lies in small wars like league. There is so much luck that can make the difference between winning/losing.
Also, this allows clans to manipulate the rankings a bit by being very picky about their wars. There have been periods in the past where clans we're over ranked because of them choosing competitions simply to optimize rankings.
If you only looked at full scale wars, I could see good argument to have the W/L matter more proportionally.
Looking at different proportions can tell you slightly different things.
Maybe a=100, b=900 is to much of an extreme if there were 1 ranking. Could see merit of doing 3, one prioritizing W/L, one pure margin, one in between.

To further illustrate issues with the F400 formula (where [a,b] = [200,400]) here is example.

FALL vs A^, lets say FALL wins 21-20
LOW vs VNM, Let's say LOW wins 21-20

These are 2 wars with similar level opponents with realistic results.

D400[200,400] Individual War scores for each clan(estimated):
FALL: 1560
A^: 1155
LOW: 1244
VNM: 855

Can you objectively say that LOW winning 21-20 is significantly better than A^ falling 20-21 to FALL?

In the [100,900] here are estimates:
FALL: 1545
A^: 1344
LOW: 1155
VNM: 935

Still has a significant different in the importance of Winning the war (1545 vs 1344 AND 1155 vs 935)
And I would argue that objectively the score comparison of LOW and A^ result MUCH closer to how it should be.

[JPlo64]

Also, look at paradox which new system brings. With the new system, first two ranked are FALL and TOFU who did not win any first class CC competition in last one year. With the current system, higher are ranked S&M, LHDD and A, who won CCup, RL and CL in last one year. Therefore, current system more evaluates perfection, which is actually winning CC events while playing versus the best of the bests. However, thenew system does notevaluate thatbutinstead evaluates high winning margins versus some of lower clans, which is irrelevant in terms of perfection.

This is my the strongest argument against the new proposed system.

You get a trophy for winning a competition.
Rankings brings everything together.
It's purpose is to have a way to rank separate than just looking at who won the competitions.

[JPlo64]

To further illustrate issues with the F400 formula (where [a,b] = [200,400]) here is example.

FALL vs A^, lets say FALL wins 21-20
LOW vs VNM, Let's say LOW wins 21-20

These are 2 wars with similar level opponents with realistic results.

D400[200,400] Individual War scores for each clan(estimated):
FALL: 1560
A^: 1155
LOW: 1244
VNM: 855

Can you objectively say that LOW winning 21-20 is significantly better than A^ falling 20-21 to FALL?

In the [100,900] here are estimates:
FALL: 1545
A^: 1344
LOW: 1155
VNM: 935

Still has a significant different in the importance of Winning the war (1545 vs 1344 AND 1155 vs 935)
And I would argue that objectively the score comparison of LOW and A^ MUCH closer to how it should be.

I'll throw a 3rd in between match too:
ICON vs FOED, let's say ICON wins 21-20 (yeah yeah, I know they're going to win by bigger than that but this is an example)

[100,900]
ICON: 1367
FOED: 1136

This has ICON's result worth just slightly more than A^'s. Objectively I think most would agree that A^ performance would actually be better in this example. But like Josko argues, Winning the war is important and deserves a bonus. Thus the scores being close seems appropriate.

[IcePack]

IIRC the original was more closely tied to the overall goal of the clan dept at the time.

While only one war ever had this occur - they were trying to deincentivize the "farming" of lower clans by upper clans. One war even had the medals removed because it was deemed a "non competitive war". (Suffice to say, present day TOFU vs BOB is similar in that regard)

The big change happened very early on in CCup history, where because clans didn't get to choose their opponents, they would still be issued medals for THOSE wars, but for pick up wars this type of war would still be medal less due to the non competitive nature.

In this case, its another war that ultimately was discouraged to even occur during an official event.

While changing the formula adjusts for these types of 'forced' occurrences, conceivably the draw back would be while possibly lightening up and encouraging more pick up wars between top and lower ranked clans would it possibly open itself up for more "farming" type rank wars?

While I dont see the majority of clans focusing in on this and caring much, even in this thread people are commenting how TOP maximized their rank performance previously and what would happen if someone did that, but to lower clans first?

Is the current clan philosophy similar to previous where they would withhold the medal of pick up wars that are deemed non competitive? Will any other preventative measures be undertaken to ensure there is no "farming" of points? In the past, this discussion lead to "why not have some wars unranked" which was deemed to be a central pillar of the ranking system, ALL wars were to be ranked regardless as long as they met certain criteria (which also neatly fit into the minimum criteria for medalling) so again, is the central pillar of whatever ranking system is being considered to ensure that ALL wars be ranked, or are you going to do non ranked vs ranked wars? etc.

I haven't looked at the proposed changes myself in depth enough to see where my opinion lands, just providing some original context / where this discussion has gone previously.

[IcePack]

But, I maintain that no clan should lose points if they are perfect in a war. You will not change my mind on that.

As my previous post indicates, I would fully agree that should be the case on clans that are in the same overall stratosphere. However when the ranking points and skill levels are so extremely different, the ranking system and clan dept at the time sought to discourage this war from happening at all, as it was deemed non competitive farming.

Once official events occurred and you didn't get to pick your opponents, the competition and the goal of the dept and ranking system started becoming at odds with one another, mostly in the RL and Cup scenarios. But ultimately, the discouragement of farming type practices held up over the far and away "occasional" type nature of lopsided wars / match ups that it wasn't changed because as you said, the ranking will work itself out eventually.

[JPlo64]

Just to throw another option out there which reading IP's post made me think of again.

In regards to pick-up wars vs tournament wars
Intuitively, I think most clans would agree that the tournament wars Mean more and clans put WAY more effort into CCup and the like.

Seems that it would be interesting to give extra weight to Tournament competition vs Pick-up competition.
One could introduce a multiplier coefficient to be added to increase weight of wars.
CCup and CL could be x2, maybe RL and other minor tournaments x1.5 and pick-up wars just x1 ?
Clans could potentially decide if they want their war to be ranked or not before the start making their coefficient 0?
This would in exchange make a fun/casual pick-up war would be less significant on your ranking and then potentially more attractive to some?

This however is not as simple to do, b/c it would require additional info to be added to the data dating back to the beginning of clans (first result is from Jan 2 2007)

[JPlo64]

IIRC the original was more closely tied to the overall goal of the clan dept at the time.

While only one war ever had this occur - they were trying to deincentivize the "farming" of lower clans by upper clans. One war even had the medals removed because it was deemed a "non competitive war". (Suffice to say, present day TOFU vs BOB is similar in that regard)

The big change happened very early on in CCup history, where because clans didn't get to choose their opponents, they would still be issued medals for THOSE wars, but for pick up wars this type of war would still be medal less due to the non competitive nature.

In this case, its another war that ultimately was discouraged to even occur during an official event.

While changing the formula adjusts for these types of 'forced' occurrences, conceivably the draw back would be while possibly lightening up and encouraging more pick up wars between top and lower ranked clans would it possibly open itself up for more "farming" type rank wars?

While I dont see the majority of clans focusing in on this and caring much, even in this thread people are commenting how TOP maximized their rank performance previously and what would happen if someone did that, but to lower clans first?

Is the current clan philosophy similar to previous where they would withhold the medal of pick up wars that are deemed non competitive? Will any other preventative measures be undertaken to ensure there is no "farming" of points? In the past, this discussion lead to "why not have some wars unranked" which was deemed to be a central pillar of the ranking system, ALL wars were to be ranked regardless as long as they met certain criteria (which also neatly fit into the minimum criteria for medalling) so again, is the central pillar of whatever ranking system is being considered to ensure that ALL wars be ranked, or are you going to do non ranked vs ranked wars? etc.

I haven't looked at the proposed changes myself in depth enough to see where my opinion lands, just providing some original context / where this discussion has gone previously.

Interesting. Thanks for the knowledge and info.