Push Winrates :: Insurgency General Discussion

Store Page

Insurgency

All Discussions Screenshots Artwork Broadcasts Videos Workshop News Guides Reviews

Insurgency > General Discussion > Topic Details

Sheppy

Mar 19, 2018 @ 7:55am

Push Winrates

Last edited by Sheppy; Mar 19, 2018 @ 7:56am

< >

Showing 1-9 of 9 comments

TheSting

Mar 19, 2018 @ 8:57am

it would be better for u to have the counts equal for each map. ignoring that, i find these stats quite interesting, esp sinjar

D3AD K1NG

Mar 19, 2018 @ 9:11am

Wow this is actually cool to see. Like TheSting, the one that suprised me was Sinjar.

Sheppy

Mar 19, 2018 @ 9:16am

Originally posted by TAW TheSting SaveAmbush.org:
it would be better for u to have the counts equal for each map. ignoring that, i find these stats quite interesting, esp sinjar

Thats the stats of the last 10 days, count means the number of rounds recorded on each map. (which essentially means there were only 3-4 best-of-5's on contact).

Sheppy

Mar 20, 2018 @ 6:30am

I tried putting the maps as fakeplayers into my rating system, this was the result. The rating of each map is represented as a gaussian with mean and varriation (mu/sig). I do not understand yet why map accquire such a high mean compared to normal played, maybe there is still a systematical bug in my code. This is now all data I have (about 1.5 Months).

sinjar_Insurgent mean: 4615 var: 349 WinRatio: 53% (449 Games)
sinjar_Security mean: 4064 var: 393 WinRatio: 46% (449 Games)

buhriz_Insurgent mean: 3873 var: 416 WinRatio: 54% (172 Games)
buhriz_Security mean: 3486 var: 470 WinRatio: 45% (172 Games)

market_Insurgent mean: 3727 var: 437 WinRatio: 57% (105 Games)
market_Security mean: 3130 var: 547 WinRatio: 42% (105 Games)

verticality_Insurgent mean: 3549 var: 488 WinRatio: 61% ( 70 Games)
verticality_Security mean: 2615 var: 575 WinRatio: 38% ( 70 Games)

station_Insurgent mean: 3464 var: 507 WinRatio: 54% ( 90 Games)
station_Security mean: 2974 var: 526 WinRatio: 45% ( 90 Games)

heights_Insurgent mean: 3451 var: 504 WinRatio: 52% (101 Games)
heights_Security mean: 3024 var: 516 WinRatio: 47% (101 Games)

tell_Insurgent mean: 3382 var: 487 WinRatio: 46% ( 93 Games)
tell_Security mean: 2956 var: 481 WinRatio: 53% ( 93 Games)

siege_Insurgent mean: 3141 var: 516 WinRatio: 65% ( 63 Games)
siege_Security mean: 2591 var: 599 WinRatio: 34% ( 63 Games)

district_Insurgent mean: 2975 var: 581 WinRatio: 52% ( 44 Games)
district_Security mean: 2368 var: 634 WinRatio: 47% ( 44 Games)

drycanal_Insurgent mean: 2953 var: 564 WinRatio: 65% ( 44 Games)
drycanal_Security mean: 2255 var: 676 WinRatio: 34% ( 44 Games)

embassy_Insurgent mean: 2999 var: 553 WinRatio: 47% ( 61 Games)
embassy_Security mean: 2619 var: 562 WinRatio: 52% ( 61 Games)

revolt_Insurgent mean: 2615 var: 611 WinRatio: 68% ( 19 Games)
revolt_Security mean: 1788 var: 744 WinRatio: 31% ( 19 Games)

kandagal_Insurgent mean: 2508 var: 604 WinRatio: 50% ( 28 Games)
kandagal_Security mean: 2006 var: 677 WinRatio: 50% ( 28 Games)

contact_Insurgent mean: 2469 var: 629 WinRatio: 56% ( 25 Games)
contact_Security mean: 2310 var: 638 WinRatio: 44% ( 25 Games)

If I didn't make any crucial mistake then this means verticallity is by FAR the most unbalanced map.

[Stargate] lazy_riveN has Insurgency

Mar 28, 2018 @ 2:18am

Originally posted by Sheppy:
Data from my 6v6 and 7v7 push Servers, not at all what I expected to be honest...

Map: kandagal | SecurityWinrate: 0.70, InsurgentWinrate: 0.30 | Count: 20
Map: station | SecurityWinrate: 0.56, InsurgentWinrate: 0.44 | Count: 36
Map: embassy | SecurityWinrate: 0.54, InsurgentWinrate: 0.46 | Count: 24
Map: revolt | SecurityWinrate: 0.53, InsurgentWinrate: 0.47 | Count: 19
Map: tell | SecurityWinrate: 0.51, InsurgentWinrate: 0.49 | Count: 53
Map: buhriz | SecurityWinrate: 0.49, InsurgentWinrate: 0.51 | Count: 123
Map: sinjar | SecurityWinrate: 0.49, InsurgentWinrate: 0.51 | Count: 177
Map: district | SecurityWinrate: 0.45, InsurgentWinrate: 0.55 | Count: 33
Map: drycanal | SecurityWinrate: 0.43, InsurgentWinrate: 0.57 | Count: 21
Map: heights | SecurityWinrate: 0.42, InsurgentWinrate: 0.58 | Count: 38
Map: verticality | SecurityWinrate: 0.41, InsurgentWinrate: 0.59 | Count: 44
Map: siege | SecurityWinrate: 0.40, InsurgentWinrate: 0.60 | Count: 35
Map: market | SecurityWinrate: 0.30, InsurgentWinrate: 0.70 | Count: 40
Map: contact | SecurityWinrate: 0.17, InsurgentWinrate: 0.83 | Count: 12
(panj and peak are not in mapcycle)

Would love to see winrates from other Servers.

No offence, but it isnt a good statistics for couple of reasons:
1) Low quantity of matches - for adequate statistics you need a sample of thousands of matches but not dozens of.
2) number of slots - 6 vs 6 and 7 vs 7. Nobody playing on this servers. The standart for Ins servers - 32 slots (16 per team).

Sheppy

Mar 28, 2018 @ 6:32am

Originally posted by lazy_riveN:
1) Low quantity of matches - for adequate statistics you need a sample of thousands of matches but not dozens of.

Obviously, but thats all the data I have, it gets more over time. Thats why I wondered about other servers, here is a continously updated list though:
https://atlantishq.de/rating/maps (sampelsize now ~1.5k or something like that, I changed the rules about when matches are ignored to playtimedifference <2.0 thats why the gamecount shifted, it essentially means that alot of people ragequit on sinjar, thats why the "felt" winrate might be a lot higher)

I don't think these statistics will change after another 10k, but we will know in a few months :P.

Originally posted by lazy_riveN:
2) number of slots - 6 vs 6 and 7 vs 7. Nobody playing on this servers. The standard for Ins servers - 32 slots (16 per team).

Unfortunally my single core performance isn't good enough for 16v16 even if wanted to host it, plus i like servers with less people. However i sincerly doubt that the amount of players makes a big difference.

Last edited by Sheppy; Mar 28, 2018 @ 6:37am

Sheppy

Apr 14, 2018 @ 2:05pm

Update:

market still the worst, sinjar getting worse too, buhriz still balance, panj & peak still disabled

market_Insurgent mean: 1815 var: 192 WinRatio: 59% (175 Games)
market_Security mean: 1184 var: 192 WinRatio: 40% (175 Games)

contact_Insurgent mean: 1918 var: 245 WinRatio: 58% ( 90 Games)
contact_Security mean: 1081 var: 245 WinRatio: 41% ( 90 Games)

drycanal_Insurgent mean: 1626 var: 208 WinRatio: 57% (128 Games)
drycanal_Security mean: 1373 var: 208 WinRatio: 42% (128 Games)

sinjar_Insurgent mean: 1874 var: 140 WinRatio: 56% (665 Games)
sinjar_Security mean: 1125 var: 140 WinRatio: 43% (665 Games)

district_Insurgent mean: 1773 var: 254 WinRatio: 59% ( 83 Games)
district_Security mean: 1226 var: 254 WinRatio: 40% ( 83 Games)

revolt_Insurgent mean: 1771 var: 294 WinRatio: 53% ( 66 Games)
revolt_Security mean: 1228 var: 294 WinRatio: 46% ( 66 Games)

heights_Insurgent mean: 1636 var: 165 WinRatio: 53% (242 Games)
heights_Security mean: 1363 var: 165 WinRatio: 46% (242 Games)

siege_Insurgent mean: 1546 var: 179 WinRatio: 52% (170 Games)
siege_Security mean: 1453 var: 179 WinRatio: 47% (170 Games)

embassy_Insurgent mean: 1780 var: 209 WinRatio: 51% (109 Games)
embassy_Security mean: 1219 var: 209 WinRatio: 48% (109 Games)

station_Insurgent mean: 1629 var: 209 WinRatio: 51% (142 Games)
station_Security mean: 1370 var: 209 WinRatio: 48% (142 Games)

buhriz_Insurgent mean: 1451 var: 142 WinRatio: 50% (424 Games)
buhriz_Security mean: 1548 var: 142 WinRatio: 49% (424 Games)

verticality_Insurgent mean: 1593 var: 202 WinRatio: 50% (123 Games)
verticality_Security mean: 1406 var: 202 WinRatio: 49% (123 Games)

kandagal_Insurgent mean: 1775 var: 241 WinRatio: 49% ( 83 Games)
kandagal_Security mean: 1224 var: 241 WinRatio: 50% ( 83 Games)

tell_Insurgent mean: 1319 var: 169 WinRatio: 43% (166 Games)
tell_Security mean: 1680 var: 169 WinRatio: 56% (166 Games)

Raikoh42

Apr 14, 2018 @ 8:14pm

Now the real question is whether push is meant to be balanced for a 6v6 or 7v7 game? honestly i think not. i think of all the maps and some of them are terrible with low player counts. it is a game mode with 20-32 players in mind and it is meant to be balanced for large teams with members coordinating to take each position. the maps aren't meant to be played with 6 people trying to cover a possible 10 routes to the objective. It's meant for for 16 players maybe covering all 10 routes and putting force into a place to break into objective.

With this in mind as well, some objectives feel near impossible to play with low player counts. sometimes the defense just cannot win an objective because there are too many ways for the attackers to flank them or cut off reinforcements. Sometimes an objective just cannot be cracked by a small attack squad because there are very limited paths that are death traps without an obscene amount of firepower breaking through.

Also who is attacking and who is defending each map? what if one side is more suited to attacking and the other is better suited defending considering supply costs and weapon choice? how do we determine which group is the better attack/defender since they don't switch sides each map?

On a larger count server it isn't nearly as bad because so what if two of your buddies got slaughtered at some stairs lets say. you have two more ready to drop nades, a sniper and a support and another rifleman watching the street, and another three guys pushed up the next viable route. you just have to guess a corner and if you go down you have nades flying and another 3 guys will overwhelm whoever is left.

6v6. you lose one guy on the stairs in front of you, that means you better pray you figured out whoever killed him and get the trade, got forbid there are two angles on those stairs. Your sniper might have a wall of dozens of possible paths to you that he has to cover so you might get flanked at any time still. You and the other guy used up your nades so you could get in the room you are in now, and theres only one other guy pushed up the next closest lane. The other two might be at a whole other position,or one might be backing up the sniper, one might be laying down suppression for the guy moving up. if any of these guys or you die you bet someone is gonna push your attack and the flank will be devastating. if you get objectives that are hard to defend/attack, theres not a whole lot you can do cause that one spot is the focus.

The reason firefight and skirmish can work so well with low counts is if you lose someone across the map it won't affect you right away, If your two buddies who went A died but you took C, you have all the time in the world to set yourself up to stop a flank. And with so few people against so many routes its not so much attack and defense team so much as attack and counter attack for a partcular point. If a point is practically impossible to defend you just don't, you leave it and just return to attack it again if someone tries taking it.

Also to take into account what were the teams playing these 6v6 games? a bunch of randoms each time? Two teams that play against each other a lot? one team of randos and one team who plays with each other all the time? Depending on the team comps you can get some wildly one sided matches. Ever played Ambush on seige? playing insurgents with randoms makes VIP victory seem impossible on that map. Play it with people who play together and have strategies for that map and you find that security side cant do much but get a lucky kill on the VIP to win.

Sure you have some numbers to try and determine if these maps are balanced. But without the details behind each number, a larger test pool, and an understanding of how the mode is supposed to be balanced. They don't exactly mean anything.

Last edited by Raikoh42; Apr 14, 2018 @ 8:18pm

Sheppy

Apr 15, 2018 @ 3:14am

Originally posted by Raikoh42:
Also who is attacking and who is defending each map? what if one side is more suited to attacking and the other is better suited defending considering supply costs and weapon choice? how do we determine which group is the better attack/defender since they don't switch sides each map?

Attacker and Defender are always the same (i.e. Security attacks on Sinjar, Insurgents defend)

Originally posted by Raikoh42:
Also to take into account what were the teams playing these 6v6 games? a bunch of randoms each time?

If you take the probability distribution (mean/var) instead of the plain winrate, the system will account for something like this.

I am well aware of the *potential* systematical bias of this data. However just as I can't prove that the player count doesn't meaningfully impact the winrates/mapbalance, the opposite is no established as well. Like I said in the previous post, I suspect the difference between 6 adn 32 is neglectable. I would be very interested in data from other server (also Non-Push ones), as I stated in my first post, I unfortunally don't have unlimited hosting capabilities.

< >

Showing 1-9 of 9 comments

Per page: 1530 50

Insurgency > General Discussion > Topic Details

Date Posted: Mar 19, 2018 @ 7:55am

Posts: 9

Discussions Rules and Guidelines

Insurgency

Report this post