Boring Stuff

[bigWham]

BW, can you please posts your dataset for the last run. As a check on it we want to run it thru the F41.

Thanks

The dataset comes from the database, that is the clan war results you guys maintain. If you want to give a data structure/format I can possible export it in some way for you.

[Keefie]

Here's some comparative analysis I've done between the CC scoreboard and the F41. I've selected a few clans from the top middle and bottom end of the scale. Unfortunately there are gaps in the F41 information, this is caused by either the F41 not being run that month or a clans weight dropping below the minimum.

Conclusions

Decay - Looks like decay is working pretty ok. There is good correlation between the F41 and CC (look at The Pack)

There is a trend revealing itself here and that is the % difference between CC and F41 is much greater the lower down the scoreboard you go. Could the reason be that there's a problem with the way CC is calculating losses ? Clans like TOFU and ACE rarely lose, while clans like VVV and AOK lose more often, so any problem with calculating losses will be exaggerated the further down the leaderboard you go. The best example is HDN, they have never won a clan war and have the biggest difference of all.

Unexplainable point drops - Look at Knights Templar. They lose 205 points between Jan and Feb 2014, yet looking at the CC database there were no wars involving KNT that finished that month. As decay can't explain that large a drop there's something here that needs investigating further.

[Leehar]

BW, can you please posts your dataset for the last run. As a check on it we want to run it thru the F41.

Thanks

The dataset comes from the database, that is the clan war results you guys maintain. If you want to give a data structure/format I can possible export it in some way for you.

excel should work, or even text deliminated as shown earlier posting.php?mode=quote&f=336&p=4588197#pr4450140
We'll hopefully run the numbers on F41 with the same data set so as to try compare like with like in a sense

[bigWham]

Here's some comparative analysis I've done between the CC scoreboard and the F41. I've selected a few clans from the top middle and bottom end of the scale. Unfortunately there are gaps in the F41 information, this is caused by either the F41 not being run that month or a clans weight dropping below the minimum.

Conclusions

Decay - Looks like decay is working pretty ok. There is good correlation between the F41 and CC (look at The Pack)

There is a trend revealing itself here and that is the % difference between CC and F41 is much greater the lower down the scoreboard you go. Could the reason be that there's a problem with the way CC is calculating losses ? Clans like TOFU and ACE rarely lose, while clans like VVV and AOK lose more often, so any problem with calculating losses will be exaggerated the further down the leaderboard you go. The best example is HDN, they have never won a clan war and have the biggest difference of all.

Unexplainable point drops - Look at Knights Templar. They lose 205 points between Jan and Feb 2014, yet looking at the CC database there were no wars involving KNT that finished that month. As decay can't explain that large a drop there's something here that needs investigating further.

Specifically the regarding Knights Templar 'unexplained point drop', I see this war finishing in Jan 2014: http://www.conquerclub.com/public.php?m ... nWarId=633

Your F41 also drops in this month, but only by 89. The settings on my calculations include CLAN_MINSCOREGAIN=200 which I believe gives a minimum score gain of 200, meaning that I think there are some settings that are fundamentally different between the two calculations here, even if the algorithms are the same. Here are all the high level settings:

define("CLAN_STARTSCORE",1000);
define("CLAN_MINSCOREGAIN",200);
define("CLAN_MAXSCOREGAIN",600);
define("CLAN_SCORETIMEWINDOW",2*365.25*24*3600);

[Keefie]

BW,

In the Clan Tab a user can filter the wars by selecting 'Active' 'Complete' etc etc to select everything the user has to put their cursor on the blank space. Can this blank space be changed to either 'All' or 'All Wars'.

Thanks
K