EPTskeldark: Can you release the season 7 data also as CSV files (You released the data in old XLS format. Also my Excel 2007 hits the 16384 limit with the files, even if v. 2007 supports more. Could it be XLS file format limitation).
Match Making Rating Tool - Page 48
| Forum Index > SC2 General |
|
korona
1098 Posts
skeldark: Can you release the season 7 data also as CSV files (You released the data in old XLS format. Also my Excel 2007 hits the 16384 limit with the files, even if v. 2007 supports more. Could it be XLS file format limitation). | ||
|
EPICWINZ
38 Posts
Are those the minimum mmr you need to be promoted into that league or are they the mmr at which you instantaneously get promoted into that league? Also, how is the moving average determined? Which games determine each point on the moving average line and how does it determine it? I'll assume it uses the same algorithim that blizzard uses to determine a users moving average but I'm quite curious to what that is. thanks :D | ||
|
Charon1979
Austria317 Posts
| ||
Excalibur_Z
United States12252 Posts
On June 20 2012 01:52 EPICWINZ wrote: so just a quick question on what the promotion lines mean. Are those the minimum mmr you need to be promoted into that league or are they the mmr at which you instantaneously get promoted into that league? Also, how is the moving average determined? Which games determine each point on the moving average line and how does it determine it? I'll assume it uses the same algorithim that blizzard uses to determine a users moving average but I'm quite curious to what that is. thanks :D I think the "moving average" in the tool is just SMA(5) but Blizzard actually uses an exponential. So, it won't be exact. The promotion lines represent the offsets that declare the range of each league. However, because of the moving average requirement, you don't get promoted immediately upon crossing one of the lines, you have to stabilize there first (or cross over the next league's line, whichever comes first). | ||
|
EPICWINZ
38 Posts
On June 20 2012 02:06 Excalibur_Z wrote: I think the "moving average" in the tool is just SMA(5) but Blizzard actually uses an exponential. So, it won't be exact. The promotion lines represent the offsets that declare the range of each league. However, because of the moving average requirement, you don't get promoted immediately upon crossing one of the lines, you have to stabilize there first (or cross over the next league's line, whichever comes first). What is SMA(5) ? And is there anyway to figure out how blizzard does the moving average? Also how off do you think this tool's moving average calculations is from blizzards? | ||
|
Excalibur_Z
United States12252 Posts
| ||
|
skeldark
Germany2223 Posts
On June 19 2012 18:39 korona wrote: skeldark: Can you release the season 7 data also as CSV files (You released the data in old XLS format. Also my Excel 2007 hits the 16384 limit with the files, even if v. 2007 supports more. Could it be XLS file format limitation). yep sorry. we lost data on the way but like 2k games. i see if i find the original files. On June 20 2012 01:52 EPICWINZ wrote: so just a quick question on what the promotion lines mean. Are those the minimum mmr you need to be promoted into that league or are they the mmr at which you instantaneously get promoted into that league? minimum mmr = Offset + 73 You have to be stable above it to get promoted, you have to be stable under it to get demoted. On June 20 2012 02:51 Excalibur_Z wrote: SMA(5) = simple moving average over 5 games. Obviously because it's a different calculation it's not going to match up with Blizzard's EMA, which has a whole bunch of hidden variables. As to how far it's off, we don't know exactly. It's just different. yes i use moving avg 5: average out of -2 -1 Game +1 +2. Blizzard obvious dont take the future into account ^^ I doubt its a complicated algorithm. When you look at the sigma, it just decrease with gamecount. I would make this way more complicated and take change of your skillchange into account. I have the feeling bliz average is more like: if (X-games between offset - nextoffset) then promotion. | ||
|
StarNaN
Sweden35 Posts
| ||
|
skeldark
Germany2223 Posts
On June 20 2012 13:09 StjarNaN wrote: I'm having trouble with the program now. Before it worked, then I reinstalled sc2 and sc2gears and the plugin. Now I have played 6 games since that, and it says my MMR is 0. What to do? I just cleaned by log(don't know why I would think that would help), so if you need it you will have to wait until I played a game. how many gamed did he say do you have ´? what say the log if you start the program? | ||
|
skeldark
Germany2223 Posts
Bnet webserver is down! I have to wait until they come up again and see what they changed! EDIT: And up again... with no change... 4.3.2 up. Only small changes | ||
|
StarNaN
Sweden35 Posts
When I start sc2gears plugin log: 6:47:41 MMR-Stats 4.3 6:47:41 --------INIT START-------- 6:47:41 OK Encoding: UTF-8 is avalable on this System 6:47:41 OK Data-Settings loaded! 6:47:41 OK Gamedata: 6 Games loaded. 6:47:41 OK Database loaded! 6:47:41 DB Version:8.0 6:47:43 OK DB up to date! 6:47:43 OK Triggers ready! 6:47:45 OK Gameobserver: ready! 6:47:45 OK Timeclient: Systemtime offset: 0 sec 6:47:47 OK: Program is up to date 6:47:47 --------INIT END-------- 6:48:7 ERROR: Analyser F: Need more/better games! 6:49:50 -------------------------- After game: 6:49:50 Game Start detected 6:50:37 Webgrabber: Player true 6:50:37 No Profile, skip 6:50:37 Webgrabber: Opponent false 6:50:37 PRE-Trigger done 7:8:2 Webgrabber: Player true 7:8:6 Webgrabber: Opponent true 7:8:6 DATA Alg.B: failed, No Bonus Win 7:8:6 Adjused Calculate Player:true 7:8:6 Adjused Calculate Opponent: true 7:8:6 Gamecheck: true 7:8:6 POST-Trigger done 7:8:6 ERROR: Analyser F: Need more/better games! 7:8:6 Player-data-Qualitiy: 75% 7:8:6 Opponent-data-Qualitiy: 75% 7:8:7 OK Uploader: game uploaded 7:8:7 Ladder Game Done! 7:8:7 -------------------------- | ||
|
skeldark
Germany2223 Posts
On June 20 2012 14:09 StjarNaN wrote: Player-data-Qualitiy: 75% ERROR: Analyser F: Need more/better games! ... On May 03 2012 20:23 skeldark wrote: You need only one game with 100% player-data-quality to see the graph! just ladder, the games are caped. I need one not caped game In the second you have this, you see the data for all stored games. You play against people in higher leagues/tiers and your mmr is below their tierminimum. In this case blizzard cap the mmr so you dont get into 0 points. You have to play worse opponents or get promoted to see uncaped gamedata. Or play against a master guy, the cap rule is disabled for masters. | ||
|
StarNaN
Sweden35 Posts
| ||
|
skeldark
Germany2223 Posts
On June 20 2012 14:39 StjarNaN wrote: I am Grandmaster, think I played a master guy. But will play more and see if it works, thanks for help. grandmaster dont have cap rule too. But importand is your opponent. So if he is grandmaster or master there is no way your game is caped. 2 reason to dont give 100% is that the games have to high changes. That he win to much or loose to less. EDIT; Only have 2 games from you. In both the opponent lost only -2 points. Thats to less to make a good point. The deviation on -2 points is +--100 I need min -4 | ||
|
korona
1098 Posts
| ||
|
skeldark
Germany2223 Posts
On June 20 2012 18:28 korona wrote: GM league for season 8 has now been opened on all servers except Kr. Hopefully there is enough data collected through the tool to determine what were the lowest GM season entry MMRs for each server. if all 200 gm player use my tool... Even if, what to do with this information? I think its not very important whats the 200th highest mmr at week 1 at session 8. it change all the time. I just take from every server a gm player out of the database if i find on and take his mmr as promotion line. if hes very high i put it a little bit lower | ||
|
fapy
Australia52 Posts
eg 1886 1903 (win) 1900 (no game played) 1916 (win) | ||
|
skeldark
Germany2223 Posts
On June 20 2012 20:22 fapy wrote: After updating my MMR is randomly changing by 3 in between some games, was this intended? eg 1886 1903 (win) 1900 (no game played) 1916 (win) i dont understand. you did not play a game but he shows you a game? and this with 3 less mmr? can you give me a picture? | ||
|
korona
1098 Posts
On June 20 2012 19:48 skeldark wrote: if all 200 gm player use my tool... Even if, what to do with this information? I think its not very important whats the 200th highest mmr at week 1 at session 8. it change all the time. I just take from every server a gm player out of the database if i find on and take his mmr as promotion line. if hes very high i put it a little bit lower It is just interesting data as likely the easiest way to get in GM is to play & win just when the GM opens and have high enough MMR moving average. There are much more players near the promotion barrier when 200 spots are open than in midseason when only couple of spots are open and the barrier is much higher. This data would also be interesting to those high master players who are wondering if they are close to the promotion barrier in the beginning of a new season. Regarding seasons: It would be nice to have a line separating each season in the MMR graph as it is hard to remember the exact point when the season changed by just looking at the graph. | ||
|
skeldark
Germany2223 Posts
On June 20 2012 20:39 korona wrote: It is just interesting data as likely the easiest way to get in GM is to play & win just when the GM opens and have high enough MMR moving average. There are much more players near the promotion barrier when 200 spots are open than in midseason when only couple of spots are open and the barrier is much higher. This data would also be interesting to those high master players who are wondering if they are close to the promotion barrier in the beginning of a new season. Regarding seasons: It would be nice to have a line separating each season in the MMR graph as it is hard to remember the exact point when the season changed by just looking at the graph. actually its other way round. in midseasson there are several hundred master players who have higher mmr than half of gm. But its still easyer to get into gm in the beginning. This just show the problems with the gm system. about the line. Yes i need this data. sadly its not in the stored data. i think i have to update that | ||
| ||
