Match Making Rating Tool - Page 94

Tistou

France14 Posts

October 01 2012 11:07 GMT

#1861

On September 29 2012 07:19 skeldark wrote:

0 from 68? thats a new record.
What was your old version? did you play with old the version of last season?
This did not work at all last weeks.

I'm not sure about which version I was on but I've been using the program for more than 2 months and always updated when required.

I could just do everything from scratch again I guess to have new data?

skeldark

Germany2223 Posts

October 01 2012 12:29 GMT

#1862

On October 01 2012 20:07 Tistou wrote:

Show nested quote +

No. You need a good game. So or so.
You use the memory scanner?
There are some MMR values where it is very hard to find a opponent who would give good data.
But over so many games you should aleast hit one.

68 is crazy even without memory scanner. Thats so unlucky that i think its not luck there must be a problem.
Can you post your main and game-logfile ( in spoilers) direct after you played a game?

Tistou

France14 Posts

October 01 2012 13:01 GMT

#1863

By memory scanner you mean memory reader? if so then yes I activated it, but just today.

Here are the logs
+ Show Spoiler +

14:32:40 Game Start detected
14:32:40 Reading Memory ...OK
14:32:40 OK Found Memory-Profile in Active-Profiles!
14:33:23 Webgrabber: Player true
14:33:26 Webgrabber: Opponent true
14:33:26 PRE-Trigger done
14:47:39 Webgrabber: Player true
14:47:42 Webgrabber: Opponent true
14:47:42 Gamecheck: true
14:47:42 POST-Trigger done
14:47:42 Player: Reading_OK: true Adjusted_OK: true Cap_OK: false Dev._OK: true
14:47:42 Opponent: Reading_OK: true Adjusted_OK: true Cap_OK: false Dev._OK: true
14:47:42 Player-data-Qualitiy: 75%
14:47:42 Opponent-data-Qualitiy: 75%
14:47:43 OK Uploader: game uploaded
14:47:43 Ladder Game Done!
14:47:43 --------------------------

btw I already met players with data quality : 100%...

14:32:40 Ramscanner: name1 : eLg%C7%82JotA
14:32:40 Ramscanner: id1 : 1-249072
14:32:40 Ramscanner: name2 : ZeonIX
14:32:40 Ramscanner: id2 : 2-23347
14:47:42 A 13 B 13

14:47:42
14:47:42 A -10 B -10

14:47:42
14:47:42 PREPLAYER 1:
Data: true
Reading: true
Adjusted: true
Dmmr: false
id: 1-249072
playerlink: http://eu.battle.net/sc2/en/profile/249072/1/eLgǂJotA/
server: EU
league: diamond
division_name: antioch zed
Main race: z
Game race: unkown
gamedate: 1/10/2012
mapname: Daybreak LE
totalbonuspool: 250
Wins: 33
Looses: -9999
points: 173
bonus: 134
change_points: -11
adjusted_points: 57

14:47:42 PREPLAYER 2:
Data: true
Reading: true
Adjusted: true
Dmmr: false
id: 2-23347
playerlink: http://eu.battle.net/sc2/en/profile/23347/2/ZeonIX/
server: EU
league: diamond
division_name: carrier november
Main race: z
Game race: unkown
gamedate: 1/10/2012
mapname: Daybreak LE
totalbonuspool: 250
Wins: 19
Looses: -9999
points: 73
bonus: 216
change_points: 0
adjusted_points: 39

14:47:42 POSTPLAYER 1:
Data: true
Reading: true
Adjusted: true
Dmmr: false
id: 1-249072
playerlink: http://eu.battle.net/sc2/en/profile/249072/1/eLgǂJotA/
server: EU
league: diamond
division_name: antioch zed
Main race: z
Game race: z
gamedate: 1/10/2012
mapname: Condemned Ridge
totalbonuspool: 251
Wins: 34
Looses: -9999
points: 199
bonus: 122
change_points: 13
adjusted_points: 70

14:47:42 POSTPLAYER 2:
Data: true
Reading: true
Adjusted: true
Dmmr: false
id: 2-23347
playerlink: http://eu.battle.net/sc2/en/profile/23347/2/ZeonIX/
server: EU
league: diamond
division_name: carrier november
Main race: z
Game race: z
gamedate: 1/10/2012
mapname: Condemned Ridge
totalbonuspool: 251
Wins: 19
Looses: -9999
points: 63
bonus: 217
change_points: -10
adjusted_points: 29

14:47:42 Dataset:
--- GAMESTART ---
TIME: 1349095617345
VALID: true
Player1_won true
--- GAMEEND ---

skeldark

Germany2223 Posts

October 01 2012 13:06 GMT

#1864

On October 01 2012 22:01 Tistou wrote:
By memory scanner you mean memory reader? if so then yes I activated it, but just today.

Here are the logs
+ Show Spoiler +

Yep. Normal caped game.

Play with Memory reader against one platinum guy and the game should be good.
Because you never had a good game against platinum, i just gamble and say you just dont play so many platinum and mainly diamond.
In this case you are between 0 and 73 points above the line.
If no other error happens 1 game against a platinum guy is good enough. (with memoryscanner / without you have to loose this game , looks like you refused to do that in the past)

barwick11

44 Posts

October 01 2012 13:24 GMT

#1865

On October 01 2012 13:12 skeldark wrote:

Show nested quote +

Edit3 : Checked it.

You data looks very solid. Different type of datapoints that all confirm the same position.

I looked closer at your opponent. Data looks fine. The diffrence between the skill is explained by the popularity.
There are just not enough sc2 players left to give you an equal opponent. Did you play at an unusual time?

To everyone that wrote me a pm and did not get an answer yet : please pm again.
I had a lot to do and im not sure if i answered everyone now.

Nope, think it was like 11 PM eastern, so plenty of rabble to play around with

Tistou

France14 Posts

October 01 2012 13:31 GMT

#1866

On October 01 2012 22:06 skeldark wrote:

Show nested quote +

On October 01 2012 22:01 Tistou wrote:
By memory scanner you mean memory reader? if so then yes I activated it, but just today.

Here are the logs
+ Show Spoiler +

mmmm ok.. Indeed I only play vs diamond players these days. However for me to play vs a plat player would mean that I lost quite a few games before so I don't know when that is going to happen (not soon I hope

) What if I play vs a master?

Anyways its alright, if the log is ok then i'll just be patient

thanks

skeldark

Germany2223 Posts

October 01 2012 15:34 GMT

#1867

On October 01 2012 22:31 Tistou wrote:

Show nested quote +

On October 01 2012 22:06 skeldark wrote:

On October 01 2012 22:01 Tistou wrote:
By memory scanner you mean memory reader? if so then yes I activated it, but just today.

Here are the logs
+ Show Spoiler +

) What if I play vs a master?

Anyways its alright, if the log is ok then i'll just be patient

thanks

Yu should play against platinum sometimes. Master would be good because a against master player disable all cap rules .

I dont know your exact MMR but the fact that you are diamond caped alone tells me:
your MMR cant be higher than 73 above the promotion line.
The question is how much lower is it.

Namelless

Germany38 Posts

October 01 2012 17:32 GMT

#1868

the graph of the player evorphi (a friend of me) does not make sense at all in my opinion, would like to hear your comment/opinion about that

we do not understand why this long skillpeak exists even though he has much good games
and why he is beside this skillpeak constantly under diamond
his promotion gave him a really short MASSIVE skillboost over a few games, so some data should be wrong?

if tl;dr

player: evorphi
league: diamond (he promoted with the first game with this skill peak over a few games, and did not get demoted up to now)
opponents: feeled 60% platinum
graphscreenshot: http://s1.directupload.net/file/d/3030/oijbikg8_png.htm

skeldark

Germany2223 Posts

October 01 2012 18:49 GMT

#1869

On October 02 2012 02:32 Namelless wrote:
the graph of the player evorphi (a friend of me) does not make sense at all in my opinion, would like to hear your comment/opinion about that

we do not understand why this long skillpeak exists even though he has much good games
and why he is beside this skillpeak constantly under diamond
his promotion gave him a really short MASSIVE skillboost over a few games, so some data should be wrong?

if tl;dr

player: evorphi
league: diamond (he promoted with the first game with this skill peak over a few games, and did not get demoted up to now)
opponents: feeled 60% platinum
graphscreenshot: http://s1.directupload.net/file/d/3030/oijbikg8_png.htm

Pm, one bad calculated game.

People check the Anlayse-log.
If you have SINGLE spikes in the data just mark the game as bad in the dataeditor.
The analyser search for such games but he will not mark them automatic because my algorithem could have wrong positives.

SirPsychoMantis

United States180 Posts

October 02 2012 01:35 GMT

#1870

On October 02 2012 03:49 skeldark wrote:

Show nested quote +

Is there any way to mark games as bad from the graph?

Fuzzwah

Australia29 Posts

October 02 2012 08:07 GMT

#1871

For a while I was getting what seemed like decent results using this, but have had some wackiness!

http://i.imgur.com/Ih1Lc.jpg

skeldark

Germany2223 Posts

October 02 2012 17:38 GMT

#1872

On October 02 2012 10:35 SirPsychoMantis wrote:

Show nested quote +

Is there any way to mark games as bad from the graph?

On October 02 2012 17:07 Fuzzwah wrote:
For a while I was getting what seemed like decent results using this, but have had some wackiness!

http://i.imgur.com/Ih1Lc.jpg

People check the Anlayse-log.
If you have SINGLE spikes in the data just mark the game as bad under Settings-Data.
Click "Refresh" , Select the game that you think is wrong, and click on "Mark as bad"
The analyser search for such games but he will not mark them automatic because my algorithem could have wrong positives.

SirPsychoMantis

United States180 Posts

October 02 2012 20:52 GMT

#1873

On October 03 2012 02:38 skeldark wrote:

Show nested quote +

I just mean it would be a lot easier if you could right click a node on the graph -> mark as bad, rather than having to dig through logs or count to get the correct game number.

Hairy

United Kingdom1169 Posts

October 03 2012 09:55 GMT

#1874

Howdy! I was setting up dropbox to automatically sync all my replays etc, and would also like to sync all my MMR data too - what files/folders for the plugin are the minimum required to have my data on all computers?

Right now I have the whole folder synced, which is fine, but it means that I can only auto-load SC2Gears on startup on ONE computer, as otherwise SC2Gears might be in use while dropbox is trying to sync the MMR tool's software itself (eg after an MMR tool update).

Cheers

InfCereal

Canada1759 Posts

October 03 2012 10:41 GMT

#1875

On October 03 2012 18:55 Hairy wrote:
Howdy! I was setting up dropbox to automatically sync all my replays etc, and would also like to sync all my MMR data too - what files/folders for the plugin are the minimum required to have my data on all computers?

Right now I have the whole folder synced, which is fine, but it means that I can only auto-load SC2Gears on startup on ONE computer, as otherwise SC2Gears might be in use while dropbox is trying to sync the MMR tool's software itself (eg after an MMR tool update).

Cheers

Should be the games.dat file, no?

BBLN

16 Posts

October 03 2012 12:51 GMT

#1876

I am in bronze league, won 6 games in a row but my MMR decline instead of rising,
First it was stuck on ~315 then jumped to 360, next game 362 and now 356.
Is it a bug or MMR could really get down after win?

skeldark

Germany2223 Posts

October 03 2012 16:45 GMT

#1877

On October 03 2012 05:52 SirPsychoMantis wrote:

I just mean it would be a lot easier if you could right click a node on the graph -> mark as bad, rather than having to dig through logs or count to get the correct game number.

Sure it would be a lot easier.

I thought about doing it BUT:
Its not doable with reason amount of time. Errors are not normal so i did not want to program so much code for a situation that does not happen often.

The datafolder is :
Sc2gears\User Content\Plugin file cache\mmr.plugin.MMRPlugin\
The games are in the games_new.dat
Dont run 2 instances at the same time! They overwrite the change of each other.

On October 03 2012 21:51 BBLN wrote:
I am in bronze league, won 6 games in a row but my MMR decline instead of rising,
First it was stuck on ~315 then jumped to 360, next game 362 and now 356.
Is it a bug or MMR could really get down after win?

Need bnet username and server to have a look

Dont know about the jump but the 360 366 356 ( all wins ) can come with the error margin.
360 was not corrected because its a jump. Errors are not corrected because i can not know what site is the error ( old value new value)
360 is most likely to high. its on max value allread.
Next game is corrected automatic to be higher than first one ( because last one was a win)
But first one was allready to high.. Now game nr 3 can not copete with such a high value and is lower than last even if last was a win.

TLDR: will have a look in the jump, the rest is follow up mistake from the jump and will correct over time.

Namelless

Germany38 Posts

October 05 2012 14:14 GMT

#1878

On October 02 2012 03:49 skeldark wrote:

Show nested quote +

a) the skillpeak is a few games long with a few good games in it (before and after the peak too) your pm advice did not correct the skillpeak
b) even if the skillpeak would dissappear it would not correct the problem
that he promoted to diamond without being over the dia line for a few games
and that (except) the skillpeak he is constantly really far below the diamond-line (without being demoted yet)
he also is not jumping around under the dia-line a lot, its more nearly a flat line (below the dia-line)

skeldark

Germany2223 Posts

October 05 2012 17:11 GMT

#1879

On October 05 2012 23:14 Namelless wrote:

Show nested quote +

a) the skillpeak is a few games long with a few good games in it (before and after the peak too) your pm advice did not correct the skillpeak

The picture and my data show clearly only one good game in the peak.
86 is a game thats calculated one changepoints to less i think

b) even if the skillpeak would dissappear it would not correct the problem
that he promoted to diamond without being over the dia line for a few games
and that (except) the skillpeak he is constantly really far below the diamond-line (without being demoted yet)
he also is not jumping around under the dia-line a lot, its more nearly a flat line (below the dia-line)

Lets see the rest of the analyer.

119 false: DMMR 77 opponen league diamond errormargin 28 opp. dmmr: 99
120 true: DMMR 227 opponen league platinum errormargin 28 opp. dmmr: 81

-So you are caped in diamond. = You must be under 73 points of diamond!
-We have a good platinum game at 227 points above the paltinum 0 line
-The promotion was around game 70 ( this should be diamond)

Now take into account that when i calculated platinum to diamond i was not really sure.
I come up with something between 240-340, The data let me think its high.
Looks like the offset is not 300 its lower.
Lower offset = lower dia line.

I have to claculate the data with all the new games again.
Problem: i have a presentation Monday and im far away from done

TLDR:
-No its just one wrong game.
-Yes there is another problem. The platinum offset is most likley lower than i thought.
-Will fix it but not this weekend.
PS: Demotion is no indicator. You can drop down to bronze level without being demoted!

EDIT:
i let him create new dataset and did a quick look.
New data give even higher result: 333,191489. Dont know whats wrong will have a closer look next week.

Can it be that promotion line are no longer synced with the offset? that would be crazy.
Or they lowered the promotion even UNDER the offset. would be also crazy.
Would make no sense to do so,
On the other hand blizzard is crazy.

genius_man16

United States749 Posts

October 05 2012 18:01 GMT

#1880

Do we know how accurate the promotion lines are? As in if someone were to stay above X-line they should be promoted soon? Or is it mostly just a guess?

Prev 1 92 93 94 95 96 207 Next

Please or register to reply.

Match Making Rating Tool - Page 94

Completed

Ongoing

Upcoming