QLD Smash has recently implemented an ELO ranking system which utilises the tournament data submitted to the site. Results are not considered, only the outcome of individual matches and players gain or lose points depending on the outcome of each match.
Of note:
* Players who defeat somebody with a high ELO rating will gain more points than if they defeat somebody with a lower ELO rating
* Players who lose to somebody with a lower ELO rating will lose more points than if they lose to somebody with a higher ELO rating
* Players who compete in regions where there are consistently high placing players will gain points more easily than players in a region with a mostly transient player base
* Players who do not have much data in the system or have not been active recently may not appear as high in an ELO rating as they should
This is a system which while not perfect, can give some insight into who the motivated Super Smash Bros. players are in Australia. It rewards people who work to never drop a set and punishes players who do things like sandbag in pools matches. It is not intended to be an absolute form of ranking players in Australia. It is intended to promote discussion, recognise top players in an objective manner, and to encourage players to perform their best all the time in a competitive setting.
03/22/2015 06:52:41
NINTENDO 64!!!!!
Thanks for taking the time to implement this into the site. The points breakdown is great.
Is there any sort of scaling to favour more recent tournaments? Because in the case of newcomers like me, I was not very good at my first tournament, but have improved a lot. And some people that may have done well in past tournaments may not be able to do quite so well now (eg. Jaice). I'm not sure if this would actually benefit more accurate results, but I just thought I'd bring it up. I'm really interested in the concept of perfecting the accuracy of ranking algorithms, but I don't have too much experience as of yet.
03/23/2015 06:26:03
Thanks for taking the time to implement this into the site. The points breakdown is great.
Is there any sort of scaling to favour more recent tournaments? Because in the case of newcomers like me, I was not very good at my first tournament, but have improved a lot. And some people that may have done well in past tournaments may not be able to do quite so well now (eg. Jaice). I'm not sure if this would actually benefit more accurate results, but I just thought I'd bring it up. I'm really interested in the concept of perfecting the accuracy of ranking algorithms, but I don't have too much experience as of yet.
03/23/2015 06:26:03
You get more points for beating players with more points, and lose more for losing to players with less points.
The Elo system I've used before had the ability to decay points but that only really helps with inactivity
03/23/2015 07:42:45
The Elo system I've used before had the ability to decay points but that only really helps with inactivity
03/23/2015 07:42:45
There isn't a decay factor set in but due to the nature of the algorithm and the fact that the matches are sorted chronologically, earlier matches have less of an impact on the final score than recent matches. There's been an update to this so that it shows the breakdown of characters used to get their score.
For example, Jezmo hardly used Marth really.
04/02/2015 16:00:46
For example, Jezmo hardly used Marth really.
04/02/2015 16:00:46
Why do I have so much Mario? I've only ever used him in one match...
04/02/2015 17:19:58
04/02/2015 17:19:58
So he used Marth to beat Billy (who had never played smash until the night before). You should replace that with Pit who he actually used to beat you in Grand Finals last UQ IMO :S
04/03/2015 00:25:34
04/03/2015 00:25:34
This should all be updated.
Elo data can be updated so if you feel it's inaccurate just post to clarify.
04/03/2015 06:07:42
Elo data can be updated so if you feel it's inaccurate just post to clarify.
04/03/2015 06:07:42
DAMNIT WHEN DID SHAYA OVERTAKE ME
04/03/2015 06:14:20
04/03/2015 06:14:20
WHEN I TAGGED ALL THE NSW SM4SH PLAYERS
SA is underrepresented atm.
04/03/2015 06:28:03
SA is underrepresented atm.
04/03/2015 06:28:03
If I'm going to be specific, I only played Falcon for one match against KK, and Ike for one match against Cetro.
04/03/2015 08:33:48
04/03/2015 08:33:48
This character data for Jett has been updated.
04/04/2015 01:22:08
04/04/2015 01:22:08
I've played 5 games as Yoshi and the rest were Pac.
04/04/2015 01:37:46
04/04/2015 01:37:46
I don't really have a good mechanism for showing change on ELO because of how frequently the data can updated, so I thought I'd post some changes I noticed after uploaded the Smashbox I results.
* Jaice +245 points with straight Rosaluma, now #2 in QLD (I like Jaice atm in spite of this)
* Fool +178 points, now #5 in QLD
* Luminescent debuts with 1644 point at #6 with Wii Fit Trainer
* Jyoshi +149
* Aero +123
* GIRL +115
* Cats +109
05/10/2015 09:23:16
* Jaice +245 points with straight Rosaluma, now #2 in QLD (I like Jaice atm in spite of this)
* Fool +178 points, now #5 in QLD
* Luminescent debuts with 1644 point at #6 with Wii Fit Trainer
* Jyoshi +149
* Aero +123
* GIRL +115
* Cats +109
05/10/2015 09:23:16
Not sure if this makes much of a difference, but the pools matches at Smashbox were calculated after the bracket matches
05/10/2015 15:07:29
05/10/2015 15:07:29
This is a good point ill look into this for sure.
05/10/2015 22:35:27
05/10/2015 22:35:27
Yes, order in which you win or lose your matches does affect ELO Ranking (assuming this is like Chess). ELO is supposed to be a predictive score, meaning it should figure out who you should win and lose against. When it doesn't match predictions, points are added and/or removed. So, a match won against a harder opponent will predict a victory over an easier one, resulting in fewer points in the second match.
However, the variance is only relevant for people who play less frequently. Those with larger sampling sizes (aka matches) will experience less variance of their ELO score. Of course, we're just talking about if player skill is static.
05/18/2015 11:33:57
However, the variance is only relevant for people who play less frequently. Those with larger sampling sizes (aka matches) will experience less variance of their ELO score. Of course, we're just talking about if player skill is static.
05/18/2015 11:33:57
The ELO rankings should now calculate pools matches before double elim matches in each tournament.
05/19/2015 13:33:18
05/19/2015 13:33:18
It feels a bit unrealistic that my ELO skyrocketted as much as it did. Is it because i beat 2 players who had placed much higher than me and lost to elton, which means my points didnt suffer as much?
05/19/2015 15:19:07
05/19/2015 15:19:07
#abusethesystem
05/20/2015 13:22:22
05/20/2015 13:22:22
JonCheee said:
It feels a bit unrealistic that my ELO skyrocketted as much as it did. Is it because i beat 2 players who had placed much higher than me and lost to elton, which means my points didnt suffer as much?
Joncheee take it xD top 20 in Australia is good
05/21/2015 09:36:16
JonCheee said:
It feels a bit unrealistic that my ELO skyrocketted as much as it did. Is it because i beat 2 players who had placed much higher than me and lost to elton, which means my points didnt suffer as much?
Play me so I can fix your score! ;-)
05/21/2015 11:14:13
Wats the difference between elo and power rankings?
05/23/2015 00:52:51
A good thing about the ELO system is once there's a good number of results it can readjust its rankings quite quickly. This also happened with Jaice in QLD - he got good but not great results with his string of women and man-child characters but then he switched to solo Rosaluma, took a game off Jezmo and shot straight to #2 in QLD after one event. All you have to do is beat players who are as good as or better than you.
JonCheee @ Myriade May Smash 4 Singles 2
> Started at 1378 which is pretty decent already
> First loss in tourney was early in the bracket which works in your favour by making you earn more points for beating people with a slightly lower score, also was to someone ranked above you which means it doesn't hurt as much
> Beat Davin, Talcn, Zivhayr and Exor, all of who were pretty hard opponents and gave you a good share of points
> Lost to Elton which caused negligible point loss
Overall you only lost to people better than you, you beat two people better than you and two people within 100 points of you. I'd say getting two upsets, no bad losses and two wins against people on your level without any other losses in a tourney makes sense as to how you got your score.
05/21/2015 12:41:36
JonCheee @ Myriade May Smash 4 Singles 2
> Started at 1378 which is pretty decent already
> First loss in tourney was early in the bracket which works in your favour by making you earn more points for beating people with a slightly lower score, also was to someone ranked above you which means it doesn't hurt as much
> Beat Davin, Talcn, Zivhayr and Exor, all of who were pretty hard opponents and gave you a good share of points
> Lost to Elton which caused negligible point loss
Overall you only lost to people better than you, you beat two people better than you and two people within 100 points of you. I'd say getting two upsets, no bad losses and two wins against people on your level without any other losses in a tourney makes sense as to how you got your score.
05/21/2015 12:41:36
Will you be adding pools from the last UQ event?
05/22/2015 14:11:13
05/22/2015 14:11:13
It was on my pile lol. This should now be uploaded, along with some ACT pools data Atyeo kindly sent me a while ago.
Am I correct in my understanding of the BAM SM4SH pools that they are each a double elim bracket where the grand final match isn't played?
05/23/2015 11:48:16
Am I correct in my understanding of the BAM SM4SH pools that they are each a double elim bracket where the grand final match isn't played?
05/23/2015 11:48:16