|
On August 13 2010 14:13 paralleluniverse wrote:
Further, there is nothing to suggest the top 200 methodology uses MMR, or purely MMR.
Uh, yeah, there is a lot to suggest that.
In this case + 24 + 24 for winning 2 games means that player has 24 bonus pool.
No, it doesn't. It means he has 12 bonus pool per day and plays 1 game per day.
I never said the bonus pool creates no point inflation. It inflates everyone's points EQUALLY, as long as they use up their bonus pool, and so the bonus pool will not screw up rankings as long as bonus pools are used up.
As an aside, I find it questionable that you believe the bonus pool screws up rankings, yet you defend the bonus pool. If I believed the bonus pool screws up rankings, I would be ripping it to shreds, regardless of the hurt feelings of casuals.
It obviously doesn't inflate points equally because it doesn't apply to everyone equally.
I'm defending the bonus pool as a mechanism to draw in casuals, not as a mechanism to create accurate rankings. You're trying to argue that the bonus pool could do the former while still being accurate, and you're wrong.
The impact of the bonus pool does NOT depend on the number of games played, as long as enough games are played to use it all up.
So it does depend on it, except in some cases. Which is what I'm saying.
You can't abuse the bonus pool because everyone gets the same bonus pool.
Again, wrong.
|
On August 13 2010 15:42 kzn wrote:Show nested quote +On August 13 2010 14:13 paralleluniverse wrote:
Further, there is nothing to suggest the top 200 methodology uses MMR, or purely MMR. Uh, yeah, there is a lot to suggest that.
Like what?
"The top 200 players are determined across divisions by comparing their relative rankings and skill, while meeting certain requirements, such as ensuring that they’re active."
Top 200 rating = relative ranking + MMR + other factors
Show nested quote +In this case + 24 + 24 for winning 2 games means that player has 24 bonus pool. No, it doesn't. It means he has 12 bonus pool per day and plays 1 game per day.
So 24 bonus pool in 2 days. A player who plays 300 games after not playing for 2 days, and therefore getting the same 24 bonus pool will still end up with the same point on average as the player who played 2 games, as my example shows.
Show nested quote +I never said the bonus pool creates no point inflation. It inflates everyone's points EQUALLY, as long as they use up their bonus pool, and so the bonus pool will not screw up rankings as long as bonus pools are used up.
As an aside, I find it questionable that you believe the bonus pool screws up rankings, yet you defend the bonus pool. If I believed the bonus pool screws up rankings, I would be ripping it to shreds, regardless of the hurt feelings of casuals. It obviously doesn't inflate points equally because it doesn't apply to everyone equally.
Yes it does.
I'm defending the bonus pool as a mechanism to draw in casuals, not as a mechanism to create accurate rankings. You're trying to argue that the bonus pool could do the former while still being accurate, and you're wrong.
Of course it can do both.
http://img227.imageshack.us/img227/5416/graphs.jpg
And if it can't do both simultaneously, then the system is a failure.
Show nested quote +The impact of the bonus pool does NOT depend on the number of games played, as long as enough games are played to use it all up. So it does depend on it, except in some cases. Which is what I'm saying.
No that's not what your saying. You're saying people who play few games gets more out of bonus pool, while people who play more games gets less.
I'm saying as long as you play enough games to use up all your bonus pool (which is like 1 game a day), then you will get the same out of the bonus pool. Playing 2 games a day, and using up the bonus pool will on average be the same as playing 300 games a day and using up the bonus pool.
A player with x points and y bonus pool, can be compared to a person with z points and 0 bonus pool by simply adding the bonus pool to the points, i.e. comparing x + y to z.
No. It's not wrong, everyone gets the same bonus pool.
|
On August 13 2010 16:17 paralleluniverse wrote:
Like what?
"The top 200 players are determined across divisions by comparing their relative rankings and skill, while meeting certain requirements, such as ensuring that they’re active."
Top 200 rating = relative ranking + MMR + other factors
If you think Blizzard doesn't highly weight the MMR in that calculation, I don't know what to tell you.
So 24 bonus pool in 2 days. A player who plays 300 games after not playing for 2 days, and therefore getting the same 24 bonus pool will still end up with the same point on average as the player who played 2 games, as my example shows.
You can keep changing your example to make it work, or you can stick to the example you used in the first place and admit it doesn't.
I'm saying as long as you play enough games to use up all your bonus pool (which is like 1 game a day), then you will get the same out of the bonus pool.
So you're arguing numbers? The numbers don't matter. If you had to play 1 game a week or less to abuse bonus pool it would still result in inaccuracies in the ranking, and its rather closer to 1 per day.
No. It's not wrong, everyone gets the same bonus pool.
Still wrong, because thats irrelevant.
|
On August 13 2010 16:24 kzn wrote:Show nested quote +On August 13 2010 16:17 paralleluniverse wrote:
Like what?
"The top 200 players are determined across divisions by comparing their relative rankings and skill, while meeting certain requirements, such as ensuring that they’re active."
Top 200 rating = relative ranking + MMR + other factors If you think Blizzard doesn't highly weight the MMR in that calculation, I don't know what to tell you.
You have no idea what Blizzard weights.
Show nested quote +So 24 bonus pool in 2 days. A player who plays 300 games after not playing for 2 days, and therefore getting the same 24 bonus pool will still end up with the same point on average as the player who played 2 games, as my example shows. You can keep changing your example to make it work, or you can stick to the example you used in the first place and admit it doesn't.
You're the one who changed the example.
I never said 24 bonus pool in 2 days. You did.
And I showed that even in such a case, the bonus pool is still fair for everyone.
It does not matter how many games you play, as long as you use up your bonus pool, your points will on average be the same.
Show nested quote +
I'm saying as long as you play enough games to use up all your bonus pool (which is like 1 game a day), then you will get the same out of the bonus pool.
So you're arguing numbers? The numbers don't matter. If you had to play 1 game a week or less to abuse bonus pool it would still result in inaccuracies in the ranking, and its rather closer to 1 per day.
You have still failed to find a situation where the bonus pool can be abused.
Still wrong, because thats irrelevant.
How is it irrelevant?
The fact that everyone gets the same bonus pool is the *very reason why it can't be abused*.
And even if people don't play, so don't use their bonus pool, you can still fairly compare their points by adding the unspent bonus pool to it.
|
On August 13 2010 12:57 paralleluniverse wrote:
Then make points converge to the rating used for the top 200 (or some monotonic transformation of it) so that points can both:
a) inflate over time with the bonus pool, so people feel progress
b) correctly rank players consistently with the correct methodology once sufficient games have been played.
b) can't happen because of a) and the fact it's basically ELO.
People will fluctuate wildly in points ranking based on if they've spent their daily bonus points or not and if they've been on a winning/losing streak lately. While they /could/ make it so that once your ranking is solid you only gain lose like 1-2 points per game, who would want that? The point of a ladder is first and foremost to be fun, accurate representation is secondary.
|
On August 13 2010 21:54 Zironic wrote:Show nested quote +On August 13 2010 12:57 paralleluniverse wrote:
Then make points converge to the rating used for the top 200 (or some monotonic transformation of it) so that points can both:
a) inflate over time with the bonus pool, so people feel progress
b) correctly rank players consistently with the correct methodology once sufficient games have been played. b) can't happen because of a) and the fact it's basically ELO. People will fluctuate wildly in points ranking based on if they've spent their daily bonus points or not and if they've been on a winning/losing streak lately. While they /could/ make it so that once your ranking is solid you only gain lose like 1-2 points per game, who would want that? The point of a ladder is first and foremost to be fun, accurate representation is secondary.
You can always consider a player who has 500 points and 50 bonus pool, as having 550 points for the purpose of comparing them with someone who has spent their bonus pool. This is justified because on average after spending their bonus pool they have points equal to their points before spending bonus pool, plus the bonus pool they spent.
http://img227.imageshack.us/img227/5416/graphs.jpg
The current system most likely already does what's in the graph just with "R" replaced by "MMR". I'm just suggesting that "MMR" be replaced with "R". It's nothing radical.
|
MMR, perhaps, shouldn't be the same as what you see on the ladder because accurate skill evaluation isn't exactly conducive to a functioning ladder. Here's why:
- It makes players afraid to play. They don't want to risk losing their precious points. This is exemplified by chess amateurs' unwillingness to play out their games to a win: better to take the safe "draw" than to lose points.
- Once you're at the top, you have no reason to continue proving your skill.
|
On August 13 2010 22:43 carwashguy wrote:MMR, perhaps, shouldn't be the same as what you see on the ladder because accurate skill evaluation isn't exactly conducive to a functioning ladder. Here's why: - It makes players afraid to play. They don't want to risk losing their precious points. This is exemplified by chess amateurs' unwillingness to play out their games to a win: better to take the safe "draw" than to lose points.
- Once you're at the top, you have no reason to continue proving your skill.
I think that's the reason why there is no MMR shown. And that it's much better to actually see some steady progress instead of you jumping from division to division up and down because MMR changes a lot faster than your real rating does.
The problem you are describing is from my point of view solved by the bonus pool, which guarantees that the ladder will grow in points all the time which makes it impossible for someone to stay on top without playing.
If Blizzard adds some statistics to each account like approximate overall rank, winpercentage per race and stuff like that I think the matchmaking and ladder system is pretty damn good 
|
On August 13 2010 23:54 Bommes wrote:Show nested quote +On August 13 2010 22:43 carwashguy wrote:MMR, perhaps, shouldn't be the same as what you see on the ladder because accurate skill evaluation isn't exactly conducive to a functioning ladder. Here's why: - It makes players afraid to play. They don't want to risk losing their precious points. This is exemplified by chess amateurs' unwillingness to play out their games to a win: better to take the safe "draw" than to lose points.
- Once you're at the top, you have no reason to continue proving your skill.
The problem you are describing is from my point of view solved by the bonus pool, which guarantees that the ladder will grow in points all the time which makes it impossible for someone to stay on top without playing.
I thought that, too. However, this guy says otherwise. He claims Starcraft 2's MMR/Ladder system comes from World of Warcraft, where inflation does not occur (even with th Bonus Pool).
|
I think it would be nice if blizzard kept the current system but added some kind of permanent ELO rating.
|
On August 14 2010 00:17 carwashguy wrote:Show nested quote +On August 13 2010 23:54 Bommes wrote:On August 13 2010 22:43 carwashguy wrote:MMR, perhaps, shouldn't be the same as what you see on the ladder because accurate skill evaluation isn't exactly conducive to a functioning ladder. Here's why: - It makes players afraid to play. They don't want to risk losing their precious points. This is exemplified by chess amateurs' unwillingness to play out their games to a win: better to take the safe "draw" than to lose points.
- Once you're at the top, you have no reason to continue proving your skill.
The problem you are describing is from my point of view solved by the bonus pool, which guarantees that the ladder will grow in points all the time which makes it impossible for someone to stay on top without playing. I thought that, too. However, this guy says otherwise. He claims Starcraft 2's MMR/Ladder system comes from World of Warcraft, where inflation does not occur (even with th Bonus Pool).
He doesn't make that claim, and your claim is obviously wrong. The bonus pool ensures inflation, and in this case that's a good thing because it ensures people don't stay at the top without playing, and yet if they return they can quickly get back up to the top if they deserve it, thanks to the bonus pool system.
|
Dionyseus, I loled when I read this! Are you trolling me?
On August 14 2010 00:56 Dionyseus wrote:
He [ZapRoffo] doesn't make that claim
"I see a lot of people spreading the idea that bonus pool inflates the ladder, when it doesn't." -ZapRoffo
On August 14 2010 00:56 Dionyseus wrote:
and your [carwashguy's] claim is obviously wrong.
My original claim and yours are the same.
"Bonus pool points actually do cause inflation (how could they not?). The point is that, despite them, the leaderboard will still accurately rank players relative to each other--so long as they keep playing games." -me
On August 14 2010 00:56 Dionyseus wrote:
The bonus pool ensures inflation.
"I thought that, too." However, this guy [ZapRoffo] says otherwise." -me
"[...] will make this effect rather small and equal for everyone, and non-inflationary (it will not grow over time)." -ZapRoffo
Try reading before posting next time, so I don't have to waste my time spelling it out to you!
|
First, hello everybody , this is my first post on TL forums.
Second, I strongly agree that there's something terribly wrong with the current ranking system, and am trying to figure out how it works myself.
I was placed into gold after going 4 - 1 in placement (losing to cannon cheese), since then I have moved up to #1 in my division in Gold, and since reaching #1, have gone roughly 10 - 2 against platinum players, but still haven't moved up.
I suspect I'm either a high plat or low diamond player, but I'm stuck in gold because of this ranking system.
Maybe it uses in game statistics like scores at the end to determine these rankings?
|
United States12224 Posts
On August 14 2010 02:35 grahamcrackuh wrote:First, hello everybody , this is my first post on TL forums. Second, I strongly agree that there's something terribly wrong with the current ranking system, and am trying to figure out how it works myself. I was placed into gold after going 4 - 1 in placement (losing to cannon cheese), since then I have moved up to #1 in my division in Gold, and since reaching #1, have gone roughly 10 - 2 against platinum players, but still haven't moved up. I suspect I'm either a high plat or low diamond player, but I'm stuck in gold because of this ranking system. Maybe it uses in game statistics like scores at the end to determine these rankings?
Welcome to TL. Please search first, though.
http://www.teamliquid.net/forum/viewmessage.php?topic_id=118212
http://www.teamliquid.net/forum/viewmessage.php?topic_id=142211
|
On August 13 2010 15:42 kzn wrote:
It obviously doesn't inflate points equally because it doesn't apply to everyone equally.
Within a particular division, over the long run, the bonus pool should apply to everyone equally if they meet a minimum activity threshold. Everyone in the division accrues bonus points at the same rate, and it doesn't take that many wins (one or two a day) to stay ahead of the bonus pool capping out.
Among sufficiently active players (meaning players who played enough to stay ahead of the cap), you can eliminate differences introduced by the bonus pool simply by adding on their unearned bonus pool points. It's possible that (within a single division) the ranking you'd get when you did that would be the same as the ranking you'd get by ordering the players according to their hidden skill ratings. However, I'm not sure it's possible to tell that with the information we have.
Across divisions, those point scores are probably not comparable even accounting for bonus points, because there seems to be a degree of inflation over time over and above bonus point accrual. This is just a gut feeling based on what I've seen, I don't have data for it.
|
This issue is about processed data vs raw data. Its a constant battle between people who have the information and don't want to release it, and people who want the information to collect statistical insight from the raw data. Raw data is very valuable and many times, very expensive to collect.
|
On August 13 2010 16:37 paralleluniverse wrote:
You have no idea what Blizzard weights.
Yes. Yes I do.
You're the one who changed the example.
Nope. You just dont understand how bonus pool works.
I never said 24 bonus pool in 2 days. You did.
Yes, you did. 12 bonus pool per day, with bonus pool only applying to wins, equates to 24 bonus pool for every loss/win/win cycle that occurs.
And I showed that even in such a case, the bonus pool is still fair for everyone.
No, you didn't. You showed that in a different case, with a different bonus pool, that it would work the way you wish it worked in reality.
The fact that everyone gets the same bonus pool is the *very reason why it can't be abused*.
Except that fact doesn't establish that conclusion at all.
And even if people don't play, so don't use their bonus pool, you can still fairly compare their points by adding the unspent bonus pool to it.
It would be more accurate to subtract the total acquired points from bonus pools, which is actually (for the first time in this thread) a decent suggestion to keep the casual baiting and allow accurate rankings to be displayed.
[edit]
Within a particular division, over the long run, the bonus pool should apply to everyone equally if they meet a minimum activity threshold. Everyone in the division accrues bonus points at the same rate, and it doesn't take that many wins (one or two a day) to stay ahead of the bonus pool capping out.
IF.
How people can make an absolute argument that is prefaced by an if is beyond me.
At a certain point in the future, barring system resets, the displayed ratings of players who meet this minimum activity threshold will begin to converge on each other, inflating at a rate equal to the rate at which bonus pool accrues.
At said point, if I wanted to climb to the top of the displayed ladder, all I would need to do is play slightly under that minimum activity value. Over time, assuming nobody else is doing the same thing and nobody else gets radically better, I will climb the ladder without actually improving.
|
Massing games isn't a good determination of skill. It just shows you have more determination. The ladder is a handicap elo system in a nutshell so a website like sc2rankings using a win/loss elo ranking will place someone with a better record overall as higher rated especially given the closeness in rating (diamond). So with all that said it's pretty obvious why idra is ranked higher than dayvie when you toss out the pooled points system.
|
On August 13 2010 22:43 carwashguy wrote:MMR, perhaps, shouldn't be the same as what you see on the ladder because accurate skill evaluation isn't exactly conducive to a functioning ladder. Here's why: - It makes players afraid to play. They don't want to risk losing their precious points. This is exemplified by chess amateurs' unwillingness to play out their games to a win: better to take the safe "draw" than to lose points.
- Once you're at the top, you have no reason to continue proving your skill.
It's far more common for strong chess players to take safe draws than amateurs. I don't know anyone under 1800 who ever takes draws except when they're clearly worse.
Also, why do I care whether or not people who are afraid to play because they might lose points play? They're probably not going to get very strong with that attitude, so they're not going to provide competition/exciting games/etc.
|
On August 14 2010 15:46 PJA wrote:Show nested quote +On August 13 2010 22:43 carwashguy wrote:MMR, perhaps, shouldn't be the same as what you see on the ladder because accurate skill evaluation isn't exactly conducive to a functioning ladder. Here's why: - It makes players afraid to play. They don't want to risk losing their precious points. This is exemplified by chess amateurs' unwillingness to play out their games to a win: better to take the safe "draw" than to lose points.
- Once you're at the top, you have no reason to continue proving your skill.
It's far more common for strong chess players to take safe draws than amateurs. I don't know anyone under 1800 who ever takes draws except when they're clearly worse. Also, why do I care whether or not people who are afraid to play because they might lose points play? They're probably not going to get very strong with that attitude, so they're not going to provide competition/exciting games/etc.
Ah, my mistake. That should've been "amateurs are less likely to play out of fear of losing points." Apparently I'm drawing from something I read in the past, and I "misremembered" it.
Anyway, you may not care if people are afraid of losing points or not, but Blizzard should (and I do, too). We'd prefer a system that, while still accurate, fosters the likelihood of a new player playing more games rather than less.
|
|
|
|
|
|
EPT
