Blizzard's top 200 show ladders are a charade. - Page 10

On August 13 2010 00:05 Voidious wrote:The top 200 may have taken minutes or hours to churn through all the battle data, which would be unreasonable for use on the ladder. It could include algorithms for filtering out disconnects, map hacks, or other noise. It could even use a Condorcet method, which definitely would not be reasonable for ladder use.

I hope they continue to improve the ladder system. A combined ladder using Elo or Glicko with fully transparent formulas and rankings/ratings would certainly get my vote. But tournaments are and always will be the true measure of a top player, so ladder inaccuracies are probably not worth getting worked up over.

On August 13 2010 00:39 Darkside- wrote:


I'm pretty sure that it's reasonably impossible to have a system that factors in disconnects, hacks and whathaveyou. Furthermore, I cannot imagine what Blizz could be doing that would require such an immense investment in computing power, unless they are literally reviewing everyone on a game-by-game basis and voting on the rankings, a la the BCS.

Like many in the thread have brought up, they need to change the ladder to reflect these supposedly more accurate rankings, or change the rankings to reflect accurate ladder standings. I don't understand the reason to have them in conflict.

On August 13 2010 01:17 Azile wrote:


Fixed for you there. Too many people assuming they know better than blizzard how to do things.

Quit trying to change the system because it's not going to happen, you're better off learning how it works and making it work for you.

The top 200 list is out, it's irrelevant whether you agree with it or not. Their system will decide the top players who play in their tournaments and ultimately play in a 'pro' league if they ever get around to making one. If you don't like that because the inner workings aren't laid out step by step for you or because they didn't rank you where you think you belong, feel free to quit playing ladder and wait for an iccup equivelant to come out.. if it ever does. Personally I feel it's not needed here, just because you don't understand the system doesn't mean it's wrong.

They have a LOT more information on hand to accurately judge players than you do or anyone else making a ranking site.

On August 12 2010 17:04 paralleluniverse wrote:
So what?

If the bonus pool is abusable go make a post about it.

This has nothing to do with whether the bonus pool is abusable. It has to do with 2 inconsistent methods attempting to do the same thing.

Your example doesn't show anything. Everyone gets the same bonus pool. Suppose that both players get 12 bonus pool.

Then the player who plays 300 games will see, on average, the following change in rating: +24 - 12 + 12 - 12 + 12 - 12 + .... - 12 = 12.

The player who plays 4 games will, on average see the following change in rating: + 24 - 12 + 12 - 12 = 12.

EDIT:
Completely off-topic: The bonus pool is NOT a psychological "bonus" that makes casuals feel better. It's because of the bonus pool that your rank DECREASES every time you log in.

Therefore, the bonus pool is as much a penalty as it is a reward.

On August 13 2010 04:44 AyJay wrote:
err think it this way

I'm at the bottom of the diamond. I fight with people around my skill level. I just got 500 points.

Now theres idra. He fights with pros everyday. He reached 500 points.

Does that mean Idra has equal skill to mine?

On August 13 2010 04:39 kzn wrote:
And its because of your bonus pool that you can expect to increase your rank every time you log in whether or not you increased your skill.

On August 13 2010 04:44 AyJay wrote:
err think it this way

I'm at the bottom of the diamond. I fight with people around my skill level. I just got 500 points.

Now theres idra. He fights with pros everyday. He reached 500 points.

Does that mean Idra has equal skill to mine?

On August 13 2010 01:39 Excalibur_Z wrote:


He does have a point, though. Just because we don't understand the system doesn't mean it's wrong, but it does mean it's confusing and nearly impossible for end-users to follow. We as players only have one trackable stat, and that's displayed rating. It stands to reason that a lot of people will get confused when they compare something that we always assumed to be true -- the SC2ranks linear ranking -- turns out to be different from Blizzard's internal tracking system.

In War3 and WoW, things were much more transparent. Seeing yourself ranked #701 on your in-game War3 profile meant that you could go to your web profile and see #701, and that you could go to the Ladder ranking page and see yourself at #701 based on the amount of XP you had accumulated. In WoW you could see a linear ranking of teams on your Battlegroup page based on the displayed rating you had earned. In SC2 everything we as players use to follow along is essentially smoke and mirrors, which understandably leads to frustration and in some cases outrage.

I have no doubt that, given full match histories for each of these Top 200 players, it may be possible to determine and replicate the methods used by Blizzard's internal stat tracking system and eventually produce Top 200 lists that are consistent with theirs in the future. Because there is so much missing information and too many unknown variables though, we're left in the dark which is an annoying place to be for an end-user.

On August 13 2010 04:39 kzn wrote:


So they aren't actually converging.

>Implying the bonus pool isn't the majority of the reason for displayed rating not matching MMR

And the player who plays 1 game will, on average, win 50% of his games. Every day that he loses, he will lose 12 (according to this ludicrously simple example). But every day he loses, his bonus pool will not decrease. Thus, the next time he manages a streak of 2 consecutive winning days, he will see +48 total over those two days.

I'm not entirely certain, but I'm pretty sure this averages out to a +24 - 6 = 18 average rating change.

Hence, I'm still right.

And its because of your bonus pool that you can expect to increase your rank every time you log in whether or not you increased your skill.

Fixed for you there. Too many people assuming they know better than blizzard how to do things.

Quit trying to change the system because it's not going to happen, you're better off learning how it works and making it work for you.

The top 200 list is out, it's irrelevant whether you agree with it or not. Their system will decide the top players who play in their tournaments and ultimately play in a 'pro' league if they ever get around to making one. If you don't like that because the inner workings aren't laid out step by step for you or because they didn't rank you where you think you belong, feel free to quit playing ladder and wait for an iccup equivelant to come out.. if it ever does. Personally I feel it's not needed here, just because you don't understand the system doesn't mean it's wrong.

They have a LOT more information on hand to accurately judge players than you do or anyone else making a ranking site.

On August 13 2010 05:11 Zironic wrote:
Stop the bloody whining. The ingame ranking system is for fun, it's so you can feel that you're actively competing against everyone else in your division and it'll move around a lot (A losing streak can easily mean you lose 100 points)

The top 200 system is based on your actual matchmaking rating. They will not show your actual matchmaking rating in-game because it's boring, in the beginning it'll jump around wildly as it tries to figure out how good you are and as time goes by it'll become solid as a rock as it's confidence increases, no one wants to sit around being 2200 for 3 months in a row. (Except in WoW, arena points wooo)

On August 13 2010 12:52 paralleluniverse wrote:
How do you know the points have not matched MMR?

No, you are not right. On average the player will win 50%, so if he wins 2 games in a row, on average he will lose the next 2 games. Any deviation from this is only temporary, and will evaporate when he plays more games.

It is not possible for the bonus pool to confer an advantage to anyone in the long term, because everyone gets the same bonus pool.

No. What the bonus pool does is this: You log off at rank 50, when you log in the next day you drop to rank 60, and play games with the bonus pool, you move back up to rank 50 again. It's not a reward.

On August 13 2010 13:44 kzn wrote:


Because the displayed rating rankings don't match the Blizzard rankings.

Except for the fact that wins are, relatively, rewarded more the less you play.

If he wins 2 games in a row, after a loss, he gets +24+24. When he loses the corresponding games, he takes -12-12. Thats a net gain of 24, thus inflation that is not present for players who play more per day.

Again, I am right.

I really find it hard to believe you're trying to argue that a bonus pool doesn't create rating inflation, and that such inflation isn't liable to create ladder inaccuracies. It is patently obvious to anyone with basic arithmetic skills that unless a 1-1 record against an equally skilled opponent generates zero deviation in relative rating, the system will be inaccurate, and it is patently obvious to anyone with basic arithmetic skills that the bonus pool implementations that act as bait for casuals apply their benefits in a way that creates just such a non-zero-sum system.



No, they really don't. It is inaccurate to look at bonus pools as an absolute value, because the impact of a bonus pool obviously depends on the number of games played.

No. What the bonus pool does is this: You log off at rank 50, when you log in the next day you drop to rank 60, and play games with the bonus pool, you move back up to rank 50 again. It's not a reward.

You're arguing semantics. That is a reward, it just comes with a loss (and it doesn't, not in the long term).

If your only goal is to have the highest displayed rating of any player, without concern for MMR, you can do that almost regardless of skill purely by abusing the bonus pool, unless the system resets (which it does).

The bonus pool provides a reason for players to play. That is its purpose, and that purpose cannot be fulfilled via the same methodology without rendering displayed rating inaccurate.