Jump to content


Active Clan Ratings Proposal


  • Please log in to reply
88 replies to this topic

Corsair4790805 #1 Posted 15 December 2021 - 06:21 PM

    Second Lieutenant

  • Member
  • 3004 battles
  • 1,466
  • [A_C_E] A_C_E
  • Member since:
    01-10-2017

I've had some time now to dig deeper into the WoWP APIs to see what data I can pull and what might be useful for generating a new Active Clans Rating System.   The good news is, I've got a working POC that retrieves a list of all clans (NA only), and then gets the 7 day ratings for all players in every clan.   There are currently 18,329 clans in NA, the vast majority are not active.   I've got the execution time down to about 2 hours to retrieve all the relevant data (originally it was looking like it would be a 12+ hour execution time) - I can probably tweak it a bit more, but I believe Donald Knuth famously said premature optimization is the root of all evil.

 

I'm haven't gotten to the point where I'm actually storing and processing the data yet, but that will be done in-memory and should not add a significant amount of processing time (anyone with experience programming knows that I/O is the most costly operation in any program).

 

So the question now is, how to "calculate" the active clan rating.  I'm looking at leveraging the following pieces of data:

 

1. Active players vs total player count in clan.

2. Number of battles flown by active players in the clan.

3. Win rate of each clan (I'm looking at doing this a bit differently than what WG does - they do an average of all the players averages, I'm thinking of doing the actual W-L totals for all the players).

 

I'm currently not looking at # of Tier 10 aircraft, but if there's enough demand that it should be part of the equation, I can pull that data as well via the API (I might be able to do the ratings and aircraft calls in parallel using asynchronous functionality, but I'm concerned about running up against WG's API throttle limits and doing it sequentially might double the execution time).

 

Any suggestions from others on:

 

1.  How to balance it out (weight it) so values aren't skewed (say a 1 person clan who's active vs a 70 person clan with 30 active players)?

2.  How much should each of the three items above count?  Kind of like when we were in school - homework might 60%, quizzes might be another 10% and mid-terms/finals might be 30%.  I'm thinking activity level and w/r should be the two biggest.

3.  Any data points that might be relevant that are not captured in the 7-day ratings (other data available is total damage dealt to aircraft, total damage dealt to ground, enemy aircraft killed and ground targets killed, in addition to what I've already listed).

 

At some point (later next year, hopefully) I may be able to put a web front end on this and make it available to everyone, with scheduled updates (at a 2 hour run time I could conceivably do it every day via a Windows Service or a daemon), with historical data available.  But I have to get the MVP (Minimum Viable Product) done first.

 

The positive on this is that we'll be evaluating active clans on a weekly basis, so it should be a more accurate listing.   Plus any time I get to write code is a plus, but that's just me :)

 

 

 



Corsair4790805 #2 Posted 15 December 2021 - 07:19 PM

    Second Lieutenant

  • Member
  • 3004 battles
  • 1,466
  • [A_C_E] A_C_E
  • Member since:
    01-10-2017

Latest run shows 397 active clans in the last 7 days out of 18,329 - 2.17%.  This run only took 70 minutes, so even better on execution time.

 

Might be interesting to run this against EU and Russia (Asia is not in the APIs for WoWP).   I think there's something like 500,000 clans in Russia though.....


Edited by Corsair4790805, 15 December 2021 - 07:19 PM.


Captain_Underpants53 #3 Posted 15 December 2021 - 07:33 PM

    Colonel

  • Member
  • 48050 battles
  • 7,721
  • [94APS] 94APS
  • Member since:
    04-17-2017

I like it!  Especially the active players VS total player count!  That would eliminate the biggest fallacy currently present with WG's rating system.

 

As far as the skewing by very small active clans, why not set a floor for inclusion in the ratings like WG does.  15 or 20 members sounds good to me.  25 maybe?

 

:medal:

 

 


MSgt, USAF, (ret)

RoIand #4 Posted 15 December 2021 - 07:36 PM

    Command Chief Master Sergeant

  • Member
  • 6466 battles
  • 511
  • [USMIL] USMIL
  • Member since:
    07-05-2015

Hello Corsair!
I think your effort is really uniquely good.
I can't come close to your mathematical expertise, so I can only add suggestions/opinions.
To somewhat offset the shift from very low member clans, perhaps you could set a minimum number of clan members.
The weighting 60/30/10 or any other will lead to endless disscussions anyway.
I have no idea what would be best there.
Maybe a vote here in the forum will help when you are ready ? However, then only the forum readers are involved - but better than nothing...
In any case, I wish you good luck and only do it if you enjoy this programming hobby. You will get nothing for it except our thanks...



Perco_lator #5 Posted 15 December 2021 - 08:09 PM

    Command Chief Master Sergeant

  • Member
  • 11 battles
  • 960
  • Member since:
    08-15-2015

My suggestion would be have clan ratings based on actual clan activities like clan wars, tournaments etc. :honoring:


 



Corsair4790805 #6 Posted 15 December 2021 - 08:15 PM

    Second Lieutenant

  • Member
  • 3004 battles
  • 1,466
  • [A_C_E] A_C_E
  • Member since:
    01-10-2017

View PostCaptain_Underpants53, on 15 December 2021 - 11:33 AM, said:

I like it!  Especially the active players VS total player count!  That would eliminate the biggest fallacy currently present with WG's rating system.

 

As far as the skewing by very small active clans, why not set a floor for inclusion in the ratings like WG does.  15 or 20 members sounds good to me.  25 maybe?

 

:medal:

 

 

 

I'm hesitant to set a floor because there are a lot of small clans and I'd like them to have a home, so to speak, even if their size will limit their ability to climb the ladder.  WG has a requirement for a minimum of 15 members before a clan will show up in the ratings, IIRC, but once they do, they're in the ratings forever, even if their member count drops below the minimum.  

 

At least, that's what I've observed.

 

At the end of the day, whatever we come up with will be arbitrary to some degree  - I'm just trying to make it balanced, as much as possible.

 



Corsair4790805 #7 Posted 15 December 2021 - 08:17 PM

    Second Lieutenant

  • Member
  • 3004 battles
  • 1,466
  • [A_C_E] A_C_E
  • Member since:
    01-10-2017

View PostPerco_lator, on 15 December 2021 - 12:09 PM, said:

My suggestion would be have clan ratings based on actual clan activities like clan wars, tournaments etc. :honoring:


 

 

I think everyone would agree with this.  However, actual clan activities are not something that individual players or clans can really influence unless/until WG brings them into the game.



Corsair4790805 #8 Posted 15 December 2021 - 08:28 PM

    Second Lieutenant

  • Member
  • 3004 battles
  • 1,466
  • [A_C_E] A_C_E
  • Member since:
    01-10-2017

View PostRoIand, on 15 December 2021 - 11:36 AM, said:

Hello Corsair!
I think your effort is really uniquely good.
I can't come close to your mathematical expertise, so I can only add suggestions/opinions.
To somewhat offset the shift from very low member clans, perhaps you could set a minimum number of clan members.
The weighting 60/30/10 or any other will lead to endless disscussions anyway.
I have no idea what would be best there.
Maybe a vote here in the forum will help when you are ready ? However, then only the forum readers are involved - but better than nothing...
In any case, I wish you good luck and only do it if you enjoy this programming hobby. You will get nothing for it except our thanks...

 

I'm hardly a math expert :)  But programming is both a career and a hobby for me.

 

So here's a quick off the top of my head idea.  I may be (and probably am) overthinking this, but let's take my clan (A_C_E) and CaptainUnderpants clan (94APS) as an example that sort of anchors both ends of the spectrums (in terms of # of players).

 

For the score, 75% is based on player count/activity and 25% is based on the clan's overall win rate for the last 7 days.   The player count/activity is # battles * % active players * % of maximum clan members.  For the 2 clans in this example, it breaks down as follows:

 

94APS

157 battles * 1 (1/1 active players) * .01 (1 out of 100 maximum players) = 1.57 * .75 = 1.18

66.88% win rate * .25 = 16.72

Total Score:   17.9

 

A_C_E

2,313 battles * .5758 (57/99 active players) * .99 (99 out of 100 maximum players) = 1,318.51 * .75 = 988.88

61.54% win rate * .25 = 15.39

Total Score:  1,004.27

 

There's a big gap between A_C_E and 94APS, but there's also a big gap in number of active players (57 vs 1).  

 

I'll need to run this against a few more clans and see how the numbers shake out, and if they look relatively good I may go with this.



Corsair4790805 #9 Posted 15 December 2021 - 08:46 PM

    Second Lieutenant

  • Member
  • 3004 battles
  • 1,466
  • [A_C_E] A_C_E
  • Member since:
    01-10-2017

Ok, so I threw in DRACS and OWSS as well, and here's what I have with the above formula:

 

DRACS

131 battles *  0.04 (4/76 active players) * .94 (84 out of 100 maximum players) = 4.93 * .75 = 3.70

80.57% win rate * .25 = 20.14

Total Score:  23.84

 

OWSS

2,593 battles * .6262 (62/99 active players) * .99 (99 out of 100 maximum players) = 1,607.51 * .75 = 1,205.64

68.98% win rate * .25 = 17.25

Total Score:  1,222.89

 

So for the 4 clans sampled, the ranking would be:

 

OWSS  1,222.89

A_C_E  1,004.27

DRACS  23.88

94APS    17.9

 

This seems to be more reflective of activity over a 7 day period, and 94APS could potentially overtake DRACS :)

 

Again, it's arbitrary, but hopefully it's balanced.


Edited by Corsair4790805, 15 December 2021 - 08:52 PM.


Captain_Underpants53 #10 Posted 15 December 2021 - 09:26 PM

    Colonel

  • Member
  • 48050 battles
  • 7,721
  • [94APS] 94APS
  • Member since:
    04-17-2017
:medal:
MSgt, USAF, (ret)

RoIand #11 Posted 15 December 2021 - 09:51 PM

    Command Chief Master Sergeant

  • Member
  • 6466 battles
  • 511
  • [USMIL] USMIL
  • Member since:
    07-05-2015

View PostCorsair4790805, on 15 December 2021 - 09:46 PM, said:

...

 

Again, it's arbitrary, but hopefully it's balanced.

I think this is a very good approach ! :great:
I'm curious how the other feedback / assessment of the forum readers is.:B



CorvusCorvax #12 Posted 15 December 2021 - 11:21 PM

    Colonel

  • Member
  • 7305 battles
  • 8,694
  • [OWSS] OWSS
  • Member since:
    01-26-2012

View PostCorsair4790805, on 15 December 2021 - 08:15 PM, said:

 

I'm hesitant to set a floor because there are a lot of small clans and I'd like them to have a home, so to speak, even if their size will limit their ability to climb the ladder.  WG has a requirement for a minimum of 15 members before a clan will show up in the ratings, IIRC, but once they do, they're in the ratings forever, even if their member count drops below the minimum.  

 

At least, that's what I've observed.

 

At the end of the day, whatever we come up with will be arbitrary to some degree  - I'm just trying to make it balanced, as much as possible.

 

Maybe the floor should be set based on battles/active member.

If you have a very small, active clan, you remain on the roster.  if you have a large, mostly non-player membership, your clan becomes inactive.  Inactive clans aren't tracked.

This would be sort of an on/off switch.  If you meet the battles/active member threshold, your clan is part of the clan tracking.  If you do not, then your clan's stats are not tracked or displayed.

This would give clan leaders incentive to prune the dead wood, and give incentive to those leaders to encourage their members to play more/recruit.

 


I might not be King of the Hill, but I am the Prince of Potatoes.

SkyWolf__WM #13 Posted 15 December 2021 - 11:52 PM

    Major

  • Closed Beta Tester
  • 20873 battles
  • 5,836
  • [T_E_S] T_E_S
  • Member since:
    07-13-2012

View PostCorsair4790805, on 15 December 2021 - 03:46 PM, said:

Ok, so I threw in DRACS and OWSS as well, and here's what I have with the above formula:

 

DRACS

131 battles *  0.04 (4/76 active players) * .94 (84 out of 100 maximum players) = 4.93 * .75 = 3.70

80.57% win rate * .25 = 20.14

Total Score:  23.84

 

OWSS

2,593 battles * .6262 (62/99 active players) * .99 (99 out of 100 maximum players) = 1,607.51 * .75 = 1,205.64

68.98% win rate * .25 = 17.25

Total Score:  1,222.89

 

So for the 4 clans sampled, the ranking would be:

 

OWSS  1,222.89

A_C_E  1,004.27

DRACS  23.88

94APS    17.9

 

This seems to be more reflective of activity over a 7 day period, and 94APS could potentially overtake DRACS :)

 

Again, it's arbitrary, but hopefully it's balanced.

I worry about 7 days. I'm in the middle of a move and out 10 days so far. Everybody else is flying though (almost). 


***************Fail to Suck****************

I shall use my aircraft and my skill to slightly inconvenience mine enemies.


Captain_Underpants53 #14 Posted 15 December 2021 - 11:56 PM

    Colonel

  • Member
  • 48050 battles
  • 7,721
  • [94APS] 94APS
  • Member since:
    04-17-2017
28 days would work.
MSgt, USAF, (ret)

tdave1953 #15 Posted 16 December 2021 - 12:22 AM

    Senior Master Sergeant

  • Member
  • 11147 battles
  • 198
  • [R-A-W] R-A-W
  • Member since:
    03-16-2021

This is just great, Corsair. Thank you for your creativity and diligence.

Oh, and thank you for what I can say to my wife the next time I screw something up.

"Sorry hon, it's just my asynchronous functionality."



WhoaBlackBetty #16 Posted 16 December 2021 - 12:40 AM

    Captain

  • Member
  • 7257 battles
  • 3,992
  • [A_C_E] A_C_E
  • Member since:
    07-27-2014
7 days. Please.  Or no more than 10 to 14. Hooked on seeing the rankings!

De-arse your bunks! Man the planes! The Fokkers are in sight! BAM ba LAMMM!
J311yfish: " 'Seal clubbing' and 'head hunting' are an inseparable part of the competitive nature of the game, and both provide a valuable service."

  "Fair" is a 4 letter "F" word.                                                                                                                   

       


Thorondar #17 Posted 16 December 2021 - 04:59 AM

    Senior Master Sergeant

  • Member
  • 7749 battles
  • 257
  • [DIXX] DIXX
  • Member since:
    01-28-2014
Corsair so far that formula looks good.  I vote NOT to include the T10 stats as I think that would encourage people to want to fast track to T10 with little hours played and even little experience in high Tiers and can take down a clans ratings.  I am all for a formula that includes 1 man clans such as Captain Undies (I know there are more out there) and would suggest NOT having a minimum amount of players in a clan (like the 15 required now to get on WeeGees ratings).  I think you get a fair score out of going by active players/win rates/battles.  Weekly results posted would be great but it depends how much of your time it takes up, if it winds up being too much then I would suggest bi-weekly rating updates.  Keep up the good work.

Corsair4790805 #18 Posted 20 December 2021 - 11:05 PM

    Second Lieutenant

  • Member
  • 3004 battles
  • 1,466
  • [A_C_E] A_C_E
  • Member since:
    01-10-2017

Thanks for the feedback everyone :)

 

As of this morning, there were 430 active clans out of 18,331.   Of those 430 clans, 256 had an average of 1 or more battles a day.  This is based on the previous 7 days.

 

Here's a look at the top 50, using the following formula:  (Total Battles * (Active Members / Total Members) * (Total Members / 100) * 0.75) + (Win Rate * 0.25).   I'm still not entirely happy with the formula, but I haven't come up with anything better.  The scores are rounded to the nearest whole number.  Tie breakers are determined in the following order:  Average battles per member per day, then win rate.   Compared with the other list, the currently present clans rank on this list as follows (only those not listed in the table below).   Let me know if this is going in the direction everyone is more or less looking for.

 

[A-S-S] #51

[AIRVI] #68

[FLD] #82

[KRTS] #95

[FPOA] #129

[_TAW_] #157

[BANDI], [-GCR-], [F_I_C], [335TH], [707TH] and [NO-10] are either completely dead or did not meet the 1.0 game/day average threshold.

 

Position Clan Active Members Total Members Average Battles/Day Win Rate Score
1 [R-A-W] The Order of Radiant Air Warriors 76 90 9.28 53.79% 3,345
2 [OWSS] Order of the White Silk Scarf 67 99 8.05 68.63% 2,819
3 [A_C_E] International ACES Alliance 62 100 7.30 60.02% 2,392
4 [USMIL] The Salty Veterans 65 92 7.29 52.88% 2,303
5 [THC-] Team Hard Core 60 100 6.98 55.29% 2,213
6 [FK] Flying Knights 56 84 8.09 53.88% 2,012
7 [GAVCA] 1 Grupo de Aviacao de Gaca - Senta a Pua - Brasil 43 99 8.32 61.96% 1,875
8 [-REX-] Rebeldes Exilliados "La vieja Guardia" 55 98 6.29 51.96% 1,794
9 [FALCO] Fuerza Armada Latina de Combate Organizada 33 98 7.82 53.65% 1,341
10 [T_E_S] The Eagle Squadron 43 60 9.62 63.33% 1,319
11 [ALAS] The Wingman Bar - Open 24/7 37 96 5.50 51.58% 1,039
12 [BSSF] Black Sheep Squadron Fighters 38 99 5.09 55.91% 1,019
13 [-DOW-] DogsofWar 35 100 5.47 54.55% 1,018
14 [GW__S] Gray Wolf Squadron 22 74 10.19 73.17% 889
15

[FTR] Fear The Reapers

33 97 4.85 54.95% 829
16 [LNSTR] LANNISTERZ 18 92 7.10 56.82% 631
17 [E-S-S] Elite Sky Soldiers 20 72 5.85 55.92% 456
18 [NAPG] North American Pilots Guild 14 45 11.48 73.87% 398
19 [X3M] X3M Fighters 30 97 2.20 54.21% 350
20 [-BFS-] Black Flight Squadron 12 76 6.88 57.79% 344
21 [QUID] Cool Hands 20 32 7.91 53.61% 279
22 [CWM] Clan Widow Makers 23 30 6.88 56.10% 263
23 [DRACS] Knights of Dracarys 6 94 8.19 60.76% 258
24 [SS-G1] Sigma Squadron 9 100 4.24 57.30% 215
25 [B_C_S] Black Crow Squadron 10 58 6.04 58.63% 199
26 [AKSS] Ace.Killing.Sky.Soldiers 13 15 16.69 56.55% 185
27 [HVAR] High Velocity Ace Rocketeers 6 29 16.52 87.32% 173
28 [VULCN] Vulcan 5 81 6.51 46.49% 150
29 [-ES-] Eagle Squadron 12 78 2.40 60.89% 133
30 [PZMOE] Panzermoe 1 96 15.14 85.85% 98
31 [SF] Special Forces 5 41 7.43 57.69% 94
32 [IMG] KAIHATSUBU 1 95 13.43 68.08% 84
33 [524] 524 Sturgeon Squadron 5 33 7.23 52.17% 76
34 [SWAK] Sealed with a Kiss 2 82 7.57 40.57% 75
35 [QC-ML] QUEBEC_MONTREAL 1 80 14.00 48.98% 71
36 [Y-I-Y] Czechoslovak Air Squadron. 9 36 3.27 56.80% 70
37 [ICB] The Iron Cross Brigade 1 89 10.86 59.21% 66
38 [ORFC] Old Rejects Flying Club 10 10 9.54 51.20% 63
39 [PNUTZ] ACE Snoopy League 11 34 2.47 60.00% 63
40 [7ARBN] 7th Armored Battalion (Tiger Battalion) 1 79 11.86 51.81% 62
41 [-D_R-] Death Rattler Squadron 5 10

17.31

53.14% 59
42 [-HBR-] Heart Break Ridge 1 78 11.57 46.91% 59
43 [SS_G2] _Sigma Squadron_G2_ 1 54 16.71 40.18% 57
44 [RO-G] ROLLOUT GAMING 1 38 21.14 54.73% 56
45 [WARMG] Warmongers 1 43 19.14 50.75% 56
46 [OLDGS] Old Guys 1 27 28.14 59.39% 55
47 [KMOFO] Kill me or fly off 8 8 11.33 62.36% 54
48 [CGN] CanadianGunNutz 2 82 3.57 52.00% 44
49 [SSGS] Super Secret Gentlemen's Squadron 2 66 4.36 44.26% 41
50 [13-AD] 13th Armored Division / Black Cats 3 17 10.38 50.00% 40

Edited by Corsair4790805, 20 December 2021 - 11:06 PM.


WhoaBlackBetty #19 Posted 21 December 2021 - 12:01 AM

    Captain

  • Member
  • 7257 battles
  • 3,992
  • [A_C_E] A_C_E
  • Member since:
    07-27-2014
Yes, I like this a LOT lot.

De-arse your bunks! Man the planes! The Fokkers are in sight! BAM ba LAMMM!
J311yfish: " 'Seal clubbing' and 'head hunting' are an inseparable part of the competitive nature of the game, and both provide a valuable service."

  "Fair" is a 4 letter "F" word.                                                                                                                   

       


ItsSubmersible #20 Posted 21 December 2021 - 12:41 AM

    Command Chief Master Sergeant

  • Member
  • 7310 battles
  • 660
  • [USMIL] USMIL
  • Member since:
    08-19-2018

View PostWhoaBlackBetty, on 20 December 2021 - 07:01 PM, said:

Yes, I like this a LOT lot.


:medal:






1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users