Aligulac.com changelog and feedback thread - Page 25

On July 16 2015 04:59 PickyProtoss wrote:
Hey man, I was just wondering how Byun got number 10 in Aligulac? I thought that there would be a system which would take into consideration the opponents skill level when determining the increase to the ELO. At least the tier in which the players are playing, i.e., low tier = .3; mid tier =.6; top tier = 1. This could work as a function of WCS points administered. It would be helpful as given the recent region blocks, or at least attempts at region blocking, it would not be so skewed towards Koreans.

I've noticed your post

Do you weigh games differently?
No, I don't. Korean tournaments and players receive no special treatment. The GSL is difficult because good players play there; the players aren't good because they play in the GSL. (Huh?)...

When fitting the model (deciding which parameters to use) I also don't weigh games differently, but this is something we're looking into. Since the database contains an overwhelming majority of non-Korean games, the system will tend to adapt to the non-Korean scene. We've noticed, however, that most people using the service are interested in the Koreans. We are still working on a useful solution to this problem.

Couldn't you control for non-Korean versus Korean in your statistical model? Just create binary variable and use in ANOVA or Regression. Not weighting the system is far from ideal, nevertheless, super job! You guys are heros!

On July 16 2015 15:10 207aicila wrote:
It's not sudden though... look how little he's lost this year...

You know it's funny, I used to lament the death of ESV, EWM and KSL, but it seems that Korean pros simply don't give a shit anymore about online tournaments. You can see that in the turnout for Olimoleague... Can't really fault ByuN for his choices, he sees the easy money. Beat, at most, one B-tier has-been Zerg like BboongBboong for $75? Sounds like a good deal.

@PickyProtoss:

As the quote you've pasted says, we're still looking into it and trying to come up with a good solution. It's been brought up many times and ByuN is only the latest in a surprisingly long list of players who got into the top 10 by farming relatively low level opponents.

But rest assured we are still discussing it, and it's not a lack of ideas that's the problem.

As a tangent, and this is only my personal opinion, not that of the team, WCS points are fucking worthless for measuring skill. We already have our own rating, we'd probably use that for any kind of "tier" system or indeed any other means of preventing this kind of farming.

On July 16 2015 16:24 Grovbolle wrote:

You make it sound so simple, yet it isn't.

Keep in mind the system does consider the difference in skill level when distributing points. It is a tradeoff between having the ranking adapt faster to monster performances and not allowing people to farm too many lower tier players.

On November 19 2014 00:30 Prillan wrote:

This is indeed a problem with rating inflation and it's kind of sad. It is caused by a combination of things like good players playing more games than bad players and so on. The fix we are working (slowly) on will hopefully fix some of this but I don't expect us to be able to solve this problem completely.

We do have a kind of decay, deviation decay. Each period we increase the uncertainty of the rating.

On July 19 2015 01:38 freeamount wrote:

The inflation is still going on. I am wondering if there is any approach to fix this, like any normalize solution.

On August 09 2015 01:07 CtrlAltDefeat wrote:
Why does the site have so many issues with downtime?

On November 18 2014 10:04 edwahn wrote:
Hi guys, long time fan of this site.

I was just wondering if you had noticed that literally every single "top 5 highest ratings ever" was achieved since April 2014 under the all race section. Do you think it's really likely that the most dominating players have been seen since April 2014 this year, or is this just an uncontrolled inflation of ratings? I know that there's no point decay system here so does this contribute to this problem?

Personally I would have loved to see the most statistically dominating players across time, but the highest ratings of all time basically turns out to be who's the hottest player in recent memory. I know the HoF basically performs this function, but it's less easy to compare between players/specific periods in time this way!

Things I expected to see in highest ratings of all time
1. MVP's dominating TvT in the early days
2. MVP in general
3. Stephano's ZvP at his peak
4. MMA's TvZ during Slayers era

etc

Thanks guys and keep up the good work!