Amidon37 wrote:

smoke wrote:

Take option I, which has a lot to be said for it. Right now, if I wanted more CP, I'd have to go play some different boards, and do well. (What MB does.) Under Option I, I'd just need to keep playing boards I sorta dominate. (There'd be a point of diminishing returns, but 2000 would be very reasonable to reach on most boards)

 Board Ranking seems to have a natural ceiling to the scores - so I don't think you will be able to build a high "I" by playing a few boards like you can with the current GR.  

Good argument @smoke, but I think the way A37 does on this.  Which way is it easier for you to achieve with ICPs?  ..the first 200 on a new board, or the last 200 that would move you from 1800 to 2000 on a board you dominate? (assume 2000 is your perceived equilibrium).  I'm willing to concede a draw on this one, but I think it's pretty clear there's more potential with the new board, and unless you have a good feel for where equilibrium is (most of us don't), your best bet is to play more boards.

berickf wrote:

which of these CP models would most adequately group players by skill and thus less arbitrarily by mixing those with skill with players with a persistent mediocrity?  OR, is that not possible at all? 

I think all of these will, since all of them move CP closer to GR. I'm not convinced that is a good thing. Currently CP is the only thing rewarding playing a diverse group of boards. It also greatly rewards playing new boards.

I mostly like option A and option I since both of them are relatively easy to understand (although the earlier misunderstanding about option A makes me wonder). I think both of them have the problem of moving CP too far towards rewarding play on fewer boards. I think a little would be helpful, but not that much.

Since Option I seems to be the most popular so far, I'll just look at that one.

I don't like increasing CP points into the hundreds and thousands. It is way more 'championship' feeling when you are fighting over 1 point at a time than a hundred poitns at a time. To solve this we could in addition to subtracting 1000 from the GR also divide by 100 and round down. This also automatically puts a floor at 1100. So a board GR of 1100 to 1199 would get you 1 CP, 1200 to 1299 would get you 2 CPs etc. For someone with a really high score (like say IRsmart's 3004 on wgwf) would be around 20 CPs.  For comparison IRsmart currently has 20 CPs for wgwf. But a lot more people would 'be in the points' with this new system.

The other problem is that I think it rewards playing the popular boards too much. Many of the boards I enjoy playing have a top score below 2000. Below 1500 even. With this system I would lose championship points on every single board that I currently have championship points on. And only gain championship points on wgwf (for which I am currently ranked #710 hardly a champion). It means being ranked #1 on Battle of Bladensburg would be only worth 9 points instead of 20. Spy vs Spy would only be worth 7 points instead of 20.

To fix this problem I think the top 10 (or whatever number) players should have extra championship points. Either just add together the current championship points and option I or give 10 extra points to #1, 9 points to #2 etc.

To bring out just my suggestion for modifying Option I and not the rambling explanation:

Take the GR for a given board. Subtract, 1000, divide by 100 round down. This would give between 1 and 20 CPs for ranks between 1100 and 3000.

• 1100 to 1199: 1 CP
• 1200 to 1299: 2 CP
• 1300 to 1399: 3 CP
• ...
• 2000 to 2099: 10 CP

Since these are Championship Points, we should add more points for actually being a champion on a specific board regardless of the state of the GR on that board. Here is one example:

• #1 on the board gets an additonal 10 CP (Gold)
• #2 on the board gets 5 CP (Silver)
• #3 on the board gets 3 CP (Bronze)
• #4-#10 get 1 CP (Honorable Mention)

Extra CPs would require a minimum of 1100 on the board.

Korrun wrote:

berickf wrote:

which of these CP models would most adequately group players by skill and thus less arbitrarily by mixing those with skill with players with a persistent mediocrity?  OR, is that not possible at all? 

I think all of these will, since all of them move CP closer to GR. I'm not convinced that is a good thing. Currently CP is the only thing rewarding playing a diverse group of boards. It also greatly rewards playing new boards.

I mostly like option A and option I since both of them are relatively easy to understand (although the earlier misunderstanding about option A makes me wonder). I think both of them have the problem of moving CP too far towards rewarding play on fewer boards. I think a little would be helpful, but not that much.

Since Option I seems to be the most popular so far, I'll just look at that one.

I don't like increasing CP points into the hundreds and thousands. It is way more 'championship' feeling when you are fighting over 1 point at a time than a hundred poitns at a time. To solve this we could in addition to subtracting 1000 from the GR also divide by 100 and round down. This also automatically puts a floor at 1100. So a board GR of 1100 to 1199 would get you 1 CP, 1200 to 1299 would get you 2 CPs etc. For someone with a really high score (like say IRsmart's 3004 on wgwf) would be around 20 CPs.  For comparison IRsmart currently has 20 CPs for wgwf. But a lot more people would 'be in the points' with this new system.

The other problem is that I think it rewards playing the popular boards too much. Many of the boards I enjoy playing have a top score below 2000. Below 1500 even. With this system I would lose championship points on every single board that I currently have championship points on. And only gain championship points on wgwf (for which I am currently ranked #710 hardly a champion). It means being ranked #1 on Battle of Bladensburg would be only worth 9 points instead of 20. Spy vs Spy would only be worth 7 points instead of 20.

To fix this problem I think the top 10 (or whatever number) players should have extra championship points. Either just add together the current championship points and option I or give 10 extra points to #1, 9 points to #2 etc.

Or, if you want the cp's to reflect something close to what they are now for the top players but still reward more players, you could divide by 25 instead of 100.  Then 1500 board GR would translate to 20 cp as it does now...  Of course one could still push their CP up from there under option I divided by 25 though...  Would that correct and maintain your CP on all your boards other then WGWF?

Korrun wrote:

To bring out just my suggestion for modifying Option I and not the rambling explanation:

Take the GR for a given board. Subtract, 1000, divide by 100 round down. This would give between 1 and 20 CPs for ranks between 1100 and 3000.

• 1100 to 1199: 1 CP
• 1200 to 1299: 2 CP
• 1300 to 1399: 3 CP
• ...
• 2000 to 2099: 10 CP

Since these are Championship Points, we should add more points for actually being a champion on a specific board regardless of the state of the GR on that board. Here is one example:

• #1 on the board gets an additonal 10 CP (Gold)
• #2 on the board gets 5 CP (Silver)
• #3 on the board gets 3 CP (Bronze)
• #4-#10 get 1 CP (Honorable Mention)

I'm mixed about the extra points for #1, 2, etc. only because they are arbitrary numbers.. why not 5 4 3 or 25 20 15?   I'm going to start another Option I thread and mention your idea in it, but I think there needs to be a less arbitrary method for 'bonus' points.

For example on WGWF:

• #1 would have 30 CP
• #2: 21 CP
• #3: 19 CP
• #4: 16 CP
• #5: 14 CP
• ...
• #11: 12 CP (currently has 0)
• ...
• #15-#19: 10 CP
• ...
• #86-#130: 5 CP
• ...
• #488-#796: 1 CP

M57 wrote:

Korrun wrote:

To bring out just my suggestion for modifying Option I and not the rambling explanation:

Take the GR for a given board. Subtract, 1000, divide by 100 round down. This would give between 1 and 20 CPs for ranks between 1100 and 3000.

• 1100 to 1199: 1 CP
• 1200 to 1299: 2 CP
• 1300 to 1399: 3 CP
• ...
• 2000 to 2099: 10 CP

Since these are Championship Points, we should add more points for actually being a champion on a specific board regardless of the state of the GR on that board. Here is one example:

• #1 on the board gets an additonal 10 CP (Gold)
• #2 on the board gets 5 CP (Silver)
• #3 on the board gets 3 CP (Bronze)
• #4-#10 get 1 CP (Honorable Mention)

I'm mixed about the extra points for #1, 2, etc. only because they are arbitrary numbers.. why not 5 4 3 or 25 20 15?   I'm going to start another Option I thread an mention your idea in it.

I had also suggested a 10 9 8 7 6 5 4 3 2 1, but would rather see it increase faster for the top 3 spots. 10 6 3 1 1 1 1 1, would be a more even pattern. Then it would be +1, +2, +3, +4 as you go towards the #1 spot. If it is a 'Championship' Point I really feel like the top player(s) on a board should get some recognition regardless of the popularity of the board. Plus it would make something to fight over which would be fun. If the only difference between #1 and #6 is 6 CP then it doesn't seem like as big of a deal to fight over the top spots.

Korrun wrote:

To bring out just my suggestion for modifying Option I and not the rambling explanation:

Take the GR for a given board. Subtract, 1000, divide by 100 round down. This would give between 1 and 20 CPs for ranks between 1100 and 3000.

• 1100 to 1199: 1 CP
• 1200 to 1299: 2 CP
• 1300 to 1399: 3 CP
• ...
• 2000 to 2099: 10 CP

Since these are Championship Points, we should add more points for actually being a champion on a specific board regardless of the state of the GR on that board. Here is one example:

• #1 on the board gets an additonal 10 CP (Gold)
• #2 on the board gets 5 CP (Silver)
• #3 on the board gets 3 CP (Bronze)
• #4-#10 get 1 CP (Honorable Mention)

Ok... So the end result of the way you have formulated it to be here... which makes option I somewhat comparable to the current CP system, is that it would mostly reduce the top cp scores gained on each board, except on the most popular boards or boards one runs there score up on, but increase the points available to all... I think?  It will still give emphasis to those playing many boards and less to those who are skilled but stick to few boards... So, tries to be more egalitarian by nature with regard to each board, but would still prioritizing those who want to specialize and niche play many boards irregardless of player skill so long as one can run the score up on a few boards by winning more then others even if they lose a lot along the way...  Of course skill is the best way to run up a board's score, but, determination and volume can work as well.  Not complaining, just understanding and saying my observations.

berickf wrote:

Or, if you want the cp's to reflect something close to what they are now for the top players but still reward more players, you could divide by 25 instead of 100.  Then 1500 board GR would translate to 20 cp as it does now...  Of course one could still push their CP up from there under option I divided by 25 though...  Would that correct and maintain your CP on all your boards other then WGWF?

Not sure what you mean by 1500 translates to 20 CP as it does now? Whatever number you pick to divide by just scales the scoring differently. That doesn't affect anything other than the minimum score to get a point (1025 instead of 1100).

Current CP Top Score / my score:

• WGWF: 20 CP / 0 CP
• Invention: 20 CP / 0 CP
• Battle of Waterloo: 8 CP / 6 CP

Dividing by 25 Top Score / my score:

• WGWF: 80 CP / 5 CP
• Invention: 20 CP / 4 CP
• Battle of Waterloo:12 CP / 11 CP

My suggestion top score / my score:

• WGWF: 30 CP / 1 CP
• Invention: 21 CP / 1 CP
• Battle of Waterloo: 13 CP / 7 CP

berickf wrote:

Ok... So the end result of the way you have formulated it to be here... which makes option I somewhat comparable to the current CP system, is that it would mostly reduce the top cp scores gained on each board, except on the most popular boards or boards one runs there score up on, but increase the points available to all... I think?  It will still give emphasis to those playing many boards and less to those who are skilled but stick to few boards... So, tries to be more egalitarian by nature with regard to each board, but would still prioritizing those who want to specialize and niche play many boards irregardless of player skill so long as one can run the score up on a few boards by winning more then others even if they lose a lot along the way...  Of course skill is the best way to run up a board's score, but, determination and volume can work as well.  Not complaining, just understanding and saying my observations.

It would give small amounts of CP to lots more players. It would slightly raise the number of CP for being top ranked on a popular board. It would slightly raise the number of CP given to top players on unpopular boards (like Battle of Waterloo). It would slight lower the CP given to top players on moderately popular boards (Battle of Bladensburg, Spy vs Spy).

For example for berickf:

• Iwo Jima and BoB both go from 20 CP to 19 CP.
• CC goes from 0 to 5 CP
• Invention 0 to 6
• Antastic 0 to 1
• WGWF 0 to 5
• Hordes of Africa 12 to 7
• Civil War 12 to 15 (both HoA and CW he is in 3rd place, but Civil War is more popular so more CP in the new system instead of them both being the same)
• Micro Mission 0 to 2
• Pangea 12 to 6 (also 3rd place but even less popular than HoA)
• World of War 0 to 1
• Battle of USA 0 to 1
• French Resistance 0 to 1
• Civil War 1860 0 to 1

berickf wrote:

smoke wrote:

(Sorry tumor ...) berickf, your GR vs CP peer comparison is nonsense--I'm surprised that's not obvious to you. It takes a lot longer to reach your "true" peer level in CP than in GR, and you have to be attempting to do it. And that's why it's worth doing. 

As stated previously, the extension of the discussion that you see as nonsense came out of an attack that I was responding to.  If I am put to task, or asked a question I do have a right to respond, do I not? Anyway, I was originally simply looking at the factual comparison between GR with h-rank and CP with h-rank.  The fact that h-rank is all over the place when looking at any 'snap shot' at a chunk of CP shows that their truly is no near same skill level peer group that is ever reached with regard to the CP rank as it stands now.  I was simply asking if any of the models corrected this.  That is the only point I was trying to make which then brings us back to the subject at hand, and what I was originally asking, which of these CP models would most adequately group players by skill and thus less arbitrarily by mixing those with skill with players with a persistent mediocrity?  OR, is that not possible at all?  Sorry if I was not clear enough and allowed my thoughts to ramble, but, I have been writing with the topic of this thread in mind.  I'm also sorry if my responses to other ramblings have led me off course. I have heard option I put forth, so lets see it!

Um ... No. I'm referring to your comparison in post #21 -- that's your first post in this thread. It wasn't in response to a question or assertion directed at you. MB didn't start drunk-posting until after that.

Back to your points. If I, for instance, look at my peers in CP, I see all tough players. Most of them have H-ratings close to mine. My GR peer group, less so in both regards. These things are all flawed. CP depends on effort. GR is volatile and reflects game choice. H-ratings aren't adjusted for opponent "skill". I think the intention is that GR should be the primary measure of skill, which is why the TrueSkill alternative is attractive. 

smoke wrote:

I think the intention is that GR should be the primary measure of skill, which is why the TrueSkill alternative is attractive. 

Agree.  Unfortunately TS is not looking likely unless someone (or some group of people step to the plate and help tom code it). Regardless, CPs sit on top of GR, which are, and if the site ever gets TS, will continue to be the primary measure of skill.   It's the CP system that's up for debate here. 

M57 wrote:

smoke wrote:

I think the intention is that GR should be the primary measure of skill, which is why the TrueSkill alternative is attractive. 

Agree.  Unfortunately TS is not looking likely unless someone (or some group of people step to the plate and help tom code it). Regardless, CPs sit on top of GR, which are, and if the site ever gets TS, will continue to be the primary measure of skill.   It's the CP system that's up for debate here. 

This may help if we can implement something like this:

http://www.wargear.net/forum/showthread/4152/How_about_bounties_for_new_features

smoke wrote:

berickf wrote:

smoke wrote:

(Sorry tumor ...) berickf, your GR vs CP peer comparison is nonsense--I'm surprised that's not obvious to you. It takes a lot longer to reach your "true" peer level in CP than in GR, and you have to be attempting to do it. And that's why it's worth doing. 

As stated previously, the extension of the discussion that you see as nonsense came out of an attack that I was responding to.  If I am put to task, or asked a question I do have a right to respond, do I not? Anyway, I was originally simply looking at the factual comparison between GR with h-rank and CP with h-rank.  The fact that h-rank is all over the place when looking at any 'snap shot' at a chunk of CP shows that their truly is no near same skill level peer group that is ever reached with regard to the CP rank as it stands now.  I was simply asking if any of the models corrected this.  That is the only point I was trying to make which then brings us back to the subject at hand, and what I was originally asking, which of these CP models would most adequately group players by skill and thus less arbitrarily by mixing those with skill with players with a persistent mediocrity?  OR, is that not possible at all?  Sorry if I was not clear enough and allowed my thoughts to ramble, but, I have been writing with the topic of this thread in mind.  I'm also sorry if my responses to other ramblings have led me off course. I have heard option I put forth, so lets see it!

Um ... No. I'm referring to your comparison in post #21 -- that's your first post in this thread. It wasn't in response to a question or assertion directed at you. MB didn't start drunk-posting until after that.

Back to your points. If I, for instance, look at my peers in CP, I see all tough players. Most of them have H-ratings close to mine. My GR peer group, less so in both regards. These things are all flawed. CP depends on effort. GR is volatile and reflects game choice. H-ratings aren't adjusted for opponent "skill". I think the intention is that GR should be the primary measure of skill, which is why the TrueSkill alternative is attractive. 

I don't know the point of continuing on in this conversation with you smoke when you just want to be blind to the truth... Anyway, sticking to the facts of it... add up the 5 players above you and below you in GR for their h-ranks (which includes the one anomaly on the first page even), divide by ten, and you get 74.5%... Compare that to your h-rank of 76%.  The average is very close to your own and represents a peer group of mostly similar h-rank success.  Do the same for CP and you get an average of 66% which is far and below your 76% and not at all a similar group to you.  So, your GR group is closer on this front where your CP group is not.  You can say it isn't all you want, but the numbers don't lie and you are a perfect example of exactly what my original point was saying.  You'll get this same result for most players save for a few anomalies.

And, I don't mean to detract from your op itsnot, just answering smoke here.

That said, my original post was simply referring to h-rank being the purest measure of skill that the site has and was querying if CP could somehow reward skill more... That's it.  The GR talk was just a comparison and not a side track.  So, smoke, if it's ok with you I'd rather just let it die here and let the cp discussion get back on track.

Thanks.

H-Rating, as wonderful and powerful a analytical tool it is, has its flaws like every other metric in the house.  We may all consider it differently when we use it to evaluate the competition, but one thing is for sure - as an indicator of achievement across boards, it's poor, and definitely doesn't have value in this discussion topic.

Dividing by 25 is pretty slick at equating the 1500 Board Rating to 20 I points.  

Amidon37 wrote:

Dividing by 25 is pretty slick at equating the 1500 Board Rating to 20 I points.  

Right, but 1500 is no longer the gold standard, right?

Amidon37 wrote:

Dividing by 25 is pretty slick at equating the 1500 Board Rating to 20 I points.  

20 is the divisor for calculating GR, so dividing gross ICPs by 20 has an H-Rating feel to it by awarding 1 point for every net win against an equal opponent.

That sounds good too!

M57 wrote:

Amidon37 wrote:

Dividing by 25 is pretty slick at equating the 1500 Board Rating to 20 I points.  

20 is the divisor for calculating GR, so dividing gross ICPs by 20 has an H-Rating feel to it by awarding 1 point for every net win against an equal opponent.

 

Has some sense to it to make it feel intuitive to the other rank dynamics.