Author Topic: Chaos March - Feedback Thread  (Read 8214 times)

0 Members and 1 Guest are viewing this topic.

Offline Starch1ld

  • Lance Captain
  • ***
  • Posts: 688
  • l33tp0intz: +47/-25
  • hi
Re: Chaos March - Feedback Thread
« Reply #15 on: January 20, 2015, 08:14:17 PM »
I see.
Vivicector had a good idea that could actually help. Round the asset prices to the nearest multiple of 5000
example, Thanatos Prime and Warhammer A would cost 95K, while Uller B and Owens E would cost 40K
Maybe rounding them to the nearest multiple of 10000 could speed up the generator even more.

This could let us determine an actual amount of money for a dropdeck, and could make the things easier

Basically we pre-generate some hundreds of dropdecks (with 4 or 5 assets)
Depending on how "random" the lists are supposed to be, generating a couple hundred dropdecks could easily take a hundred years. We don't have enough assets in MWLL. My program is far from optimized and thus takes up to a couple minutes to generate a list with many restrictions - if you drop the limitations, you end up with decks that nobody wants to use. It's not that easy.
Well the lists are supposed to be basically fully randomized (except some veery little limitations, like "no more than 1 flying asset for each dropdeck". All 4-asset decks should have the same Cbill, weight and BV limit. Limits that will be different, of course, for the 5-asset decks.


if you drop the limitations, you end up with decks that nobody wants to use. It's not that easy.
Maybe, but you would still have lots of dropdecks available for a match, so there are surely going to be some good ones. I hope so at least.
Of course there would still be some sub-optimal assets in the drop decks. But this is the actual point of the Chaos March, after all.

Offline thEClaw

  • Star Captain
  • ***
  • Posts: 976
  • l33tp0intz: +75/-0
Re: Chaos March - Feedback Thread
« Reply #16 on: January 20, 2015, 08:50:23 PM »
Modern PCs won't benefit from rounding numbers. Adding up integers/doubles is the least amount of work the generator has to do (and no matter the size of the number, it's still only one arithmetical operation - calculating 169043*6403 takes just as long as 1*1) - it is much more about the statistical complexity of the problem, there are more than 10^16 possible drop-lists, and that's for 8 players only. Adding a dozen different criteria restricts you to maybe 1/1000000 of those, so searching nice droplists can take a little bit of time. (When not restricting the generator too much, lists are generated in less than a second. My examples are somewhat extreme.)

With so much customization to the generator, however, I would argue that every generated list is nicely playable. You could even make that part of the game: Either you receive 10 drop-lists to choose from, or you are allowed to impose additional criteria (1 criterion = 3 droplists less) during the generation of your droplist. Gamble or fine-tune, and still have nicely balanced games.

Also, the lists should not be too random. Some people like the concept of totally randomizing their game(s) (like me), some can, under no circumstances, live with a sub-optimal asset. And I think the lists I posted are quite nice while giving the lance leader the opportunity to get lists that agree with their strategy for a specific map. There are many more possibilities to fine-tune the generation, after all.

Offline BuRNeR

  • Recruit
  • *
  • Posts: 166
  • l33tp0intz: +11/-0
Re: Chaos March - Feedback Thread
« Reply #17 on: January 20, 2015, 09:38:57 PM »
I donīt think the randomdrops will work, because it is really close to our old regiment system.
Without any elaboration, I would dismiss that statement as "pessimistic". No idea what a "regiment system" is supposed to be.




The "regiment system" was the previous way of the chaos march. During the time we made many changes there and were now with the pool/droplist system. It is not pessimistic, because it is my opinion which is based on my prior experience as a fraction leader in the chaos march.

edit: this is how we started in this event: http://forum.mechlivinglegends.net/index.php/topic,20413.0.html

« Last Edit: January 20, 2015, 09:49:14 PM by BuRNeR »


"If you want to face us, fine. But we will not hold back. Prepare to reap the whirlwind." Sarah McEvedy 8th October 2823

Offline xInVicTuSx

  • Apprentice Dev
  • Living Legend
  • *
  • Posts: 3683
  • l33tp0intz: +193/-5
  • Knight of the Inner Sphere
Re: Chaos March - Feedback Thread
« Reply #18 on: January 21, 2015, 04:54:33 AM »
This is sounding more and more like the regiment system except randomized rather than hand made.

I would love to see those lists you made fight and see how it goes.
-Invictus ne Vindicetur-

Offline ELH_Vivicector

  • Star Colonel
  • ****
  • Posts: 1285
  • l33tp0intz: +49/-4
Re: Chaos March - Feedback Thread
« Reply #19 on: January 21, 2015, 06:15:37 AM »
As far as I have heard, the regiment system used nearly no balancing between the regiments and that was the problem! They were just handmade 4 mech picks, far from being perfect. Here, we have dual-stat balancing of assets and ability to provide more specialized (even if still randomized) droplists when needed (no Arrow4s on Thunderrift, thank you!).

I don't like the 4 and 5 men decks since it limits us to a certain amount of players, and others will have to wait. No fun for them. Full pre-generated drop lists can be done when the amount of people playing is set, since it would not take much time.

Offline thEClaw

  • Star Captain
  • ***
  • Posts: 976
  • l33tp0intz: +75/-0
Re: Chaos March - Feedback Thread
« Reply #20 on: January 21, 2015, 06:37:53 AM »
Here a look at my first halfway functioning interface for the generator:


The general settings are almost all visible, the criterion-settings are not (I have no idea yet how to properly visualize them without overwhelming the user). UI is a bit of work since I wanted to use this project to get into QML and my prior experience with that is almost zero. So I won't promise anything pretty, but functionality should be given. ;)

PS: Settings for the two lists were quite simple, only a slight duplicate-protection, somewhat tight mass- and price-ranges and a mass-distribution around 75t (i.e. lighter and heavier assets get more unlikely the bigger the difference of their weight to 75t).

Offline Starch1ld

  • Lance Captain
  • ***
  • Posts: 688
  • l33tp0intz: +47/-25
  • hi
Re: Chaos March - Feedback Thread
« Reply #21 on: January 21, 2015, 12:48:08 PM »
I don't like the 4 and 5 men decks since it limits us to a certain amount of players, and others will have to wait.

Nope, this system is actually made up to give way more flexibility. And I completely forgot about BAs while talking about it, lol. Let me give examples

------------ Case 1 - 12 players in TS

---Both teams (6 players)
4 players -> 4-asset dropdeck
1 player  -> APC w / 5-spawn Battle Armor
1 player -> Single-Spawn Battle Armor

------------ Case 2 - 19 players in TS

---Team A (9 players)
4 players -> 4-asset dropdeck
4 players -> 4-asset dropdeck
1 player  -> APC w / 5-spawn Battle Armor
---Team B (10 players)  (uneven players, remember that decks should be properly balanced so a 5-asset deck would not stomp a 4-asset one!)
4 players -> 4-asset dropdeck
5 players -> 5-asset dropdeck
1 player  -> APC w / 5-spawn Battle Armor

------------ Case 3 - 13 players in TS

Exactly like Case 1, but with only 1 spectator

------

I can write all the cases we can have if you want. But the point is that we basically would have 1 spectator in only few cases, instead of having it everytime we got an odd number of players :)

@thEClaw, I'm curious to see how the generator will behave with Battle Values...btw, they are here https://docs.google.com/spreadsheets/d/1zHWBdgEQ5PciP8pABVOeeBv_nzy0efZUujLiMRI9Mgs/edit

FYI: Battle Armor Prime is supposed to be a no-respawn BA, while the APC is supposed to bring also a 4-REspawn BA with it. And it can refill ammo to every ally asset
« Last Edit: January 21, 2015, 01:00:31 PM by Starch1ld »

Offline thEClaw

  • Star Captain
  • ***
  • Posts: 976
  • l33tp0intz: +75/-0
Re: Chaos March - Feedback Thread
« Reply #22 on: January 21, 2015, 02:07:27 PM »
I would assume that BV is not a good measure to generate droplists by - they probably stem from the boardgame, right? Out of tonnage, price and battle value I think the price is the one that gives the best chance of producing a balanced list of assets.
However, I can add the BVs to my data and see what happens. Will be back in...however long that takes.

EDIT:
Battle values are in place.


I don't really know how to use the BV, though. I just assumed "8 players -> 24 total BV", even though some other approach might be better. Here the total settings (mass and price still included):
Code: [Select]
[duplicates] #slight protection against duplicate assets
criterion_type=RepetitionBlock
softness=0.6

[mass] #average mass of 75 tons, total range from 0.2*75=15 to 1.8*75=135
budget=75
criterion=mass
criterion_type=SingleBudget
inverse=false
isLimit=false
softness=0.8

[massTotal] #total mass of 550 tons
budget=550
criterion=mass
criterion_type=TotalBudget
maxPercentage=0.97
minPercentage=0.92
softness=0.01

[priceTotal] #total budget of 650000
budget=650000
criterion=price
criterion_type=TotalBudget
maxPercentage=0.97
minPercentage=0.92
softness=0.02

[bvTotal] #total battle value of 24
budget=24
criterion=battlevalue
criterion_type=TotalBudget
maxPercentage=1.0
minPercentage=0.85
softness=0

And here some drop-lists:
Code: [Select]
Picked the following assets (8/8 in total) after 26 failed tries (4.606s):
        80 t,  82665 CBills, 1    BV, Demolisher E (Inner Sphere) - GECM, 1 LAMS - 2 UAC10
        80 t,  88665 CBills, 5    BV, Demolisher Prime (Inner Sphere) - GECM, 1 LAMS - 2 AC20
        25 t,  42632 CBills, 2    BV, Harasser Prime (Inner Sphere) - GECM - 2 SSRM6
        85 t,  95731 CBills, 2    BV, Huitzilopochtli A (Clan) - GECM, EOptics, 6 HS - 2 CArrowIV, 2 CLBX10
        75 t, 137748 CBills, 4    BV, Madcat B (Clan) - EOptics, 1 LAMS, 9 DHS - 2 CERLBL, 4 CERMBL, 2 Mgun, 1 CGauss
        80 t,  52830 CBills, 4    BV, Partisan A (Inner Sphere) - GECM, EOptics - 4 LBX5
        60 t,  77899 CBills, 2    BV, Rifleman A (Inner Sphere) - BAP, GECM, EOptics - 2 Mgun, 4 UAC5
        25 t,  30299 CBills, 2    BV, Solitaire D (Clan) - Radar - 4 Mgun, 1 CLBX20
Total: 510 t, 608469 CBills, 2.75 BV

Picked the following assets (8/8 in total) after 345 failed tries (58.473s):
        45 t,  50648 CBills, 1    BV, Hollander II D (Inner Sphere) - BAP, EOptics - 1 LtGauss, 4 AC2
        75 t, 111173 CBills, 4    BV, Madcat G (Clan) - 4 DHS - 1 CDSRM4, 2 CHML, 4 CHSL, 2 CLBX5, 1 CLBX20
        60 t,  71006 CBills, 2    BV, Oro B (Clan) - 3 DHS - 1 CUAC20, 1 CHLL, 2 CERSBL
        30 t,  41199 CBills, 3    BV, Osiris C (Inner Sphere) - ImprJJ, EOptics, 2 DHS - 2 Mgun, 2 PPC
        80 t,  58180 CBills, 5    BV, Partisan B (Inner Sphere) - EOptics, 1 DHS - 4 RAC2
        85 t, 111615 CBills, 2    BV, Shiva Prime (Inner Sphere) - 9 DHS - 1 SSRM6, 4 LPL, 1 LBX20
        70 t, 113799 CBills, 3    BV, Thor E (Clan) - StndJJ, EOptics, 8 DHS - 2 CERPPC, 1 CGauss
        70 t,  82400 CBills, 2    BV, Warhammer C (Inner Sphere) - 1 LAMS - 1 DSRM6, 2 LBX10, 2 MBL, 2 SBL
Total: 515 t, 640020 CBills, 2.75 BV

Picked the following assets (8/8 in total) after 31 failed tries (5.459s):
        80 t,  97845 CBills, 2    BV, Awesome B (Inner Sphere) - GECM, EOptics, 5 DHS - 2 LPL, 2 AC10
        35 t,  43448 CBills, 2    BV, Chevalier D (Inner Sphere) - BHP, C3, 2 DHS - 1 LXPL, 1 MXPL
        35 t,  38698 CBills, 3    BV, Owens D (Inner Sphere) - BAP, MASC, 2 HS - 2 SRM6, 3 SPL
        80 t,  47830 CBills, 2    BV, Partisan Prime (Inner Sphere) - GECM, EOptics - 4 AC5
        85 t, 124015 CBills, 5    BV, Shiva A (Inner Sphere) - BAP, 2 DHS - 2 ERLBL, 2 HGauss
        60 t, 110497 CBills, 3    BV, Visigoth F (Clan) - EOptics, 10 DHS - 3 CERPPC, 2 CeATM3
        70 t,  88250 CBills, 1    BV, Warhammer D (Inner Sphere) - EOptics - 1 LRM15, 1 UAC10, 1 ERPPC, 2 MBL, 2 AC2
        70 t,  92350 CBills, 3    BV, Warhammer Prime (Inner Sphere) - BAP, GECM, 1 LAMS, 10 HS - 1 SRM6, 2 PPC, 2 MPL, 2 SPL
Total: 515 t, 642933 CBills, 2.62 BV

Picked the following assets (8/8 in total) after 0 failed tries (0.186s):
        70 t,  93446 CBills, 2    BV, Avatar A (Inner Sphere) - Radar, StndJJ, EOptics, 3 DHS - 2 LRM5, 2 ERMBL, 2 Mgun, 1 Gauss, 1 ERPPC
        55 t,  75000 CBills, 3    BV, Bushwacker E (Inner Sphere) - GECM, EOptics - 1 SRM4, 1 Gauss, 1 ERPPC
        65 t,  83997 CBills, 2    BV, Catapult E (Inner Sphere) - BHP, StndJJ, EOptics, 1 LAMS, 4 HS - 2 ELRM15, 4 MBL
        85 t, 114731 CBills, 4    BV, Huitzilopochtli C (Clan) - GECM, EOptics, 14 HS - 2 CArrowIV, 6 CERMBL
        65 t, 102500 CBills, 3    BV, Loki B (Clan) - 11 DHS - 2 CHLL, 2 Mgun, 1 CUAC20
        35 t,  41548 CBills, 2    BV, Owens F (Inner Sphere) - GECM, 3 HS - 1 SSRM6, 3 SPL, 2 MPL
        80 t,  39430 CBills, 1    BV, Partisan E (Inner Sphere) - EOptics - 2 Mgun, 4 UAC2
        55 t,  81996 CBills, 4    BV, Ryoken A (Clan) - GECM, EOptics, 5 DHS - 4 CMPL, 1 CLRM15, 1 CDSRM6
Total: 510 t, 632648 CBills, 2.62 BV

Also, two droplists without any restrictions in mass or price, just BV:
Code: [Select]
Picked the following assets (8/8 in total) after 6 failed tries (0.404s):
        90 t, 141598 CBills, 3    BV, Blood Asp E (Clan) - EOptics, 9 DHS - 4 Mgun, 2 CGauss, 2 CHLL
        55 t,  69700 CBills, 3    BV, Bushwacker G (Inner Sphere) - 1 DHS - 1 SRM6, 2 Mgun, 1 MPL, 1 AC20, 1 PPC
        70 t, 143446 CBills, 3    BV, Novacat A (Clan) - EOptics, 13 DHS - 6 CERLBL
        35 t,  43398 CBills, 2    BV, Owens B (Inner Sphere) - C3, MASC, 1 HS - 2 SSRM6, 2 SBL
        35 t,  59795 CBills, 3    BV, Raven C (Inner Sphere) - C3, BHP, GECM, MASC, EOptics, 2 DHS - 1 ERPPC, 2 MBL
        55 t,  87396 CBills, 5    BV, Ryoken B (Clan) - 4 DHS - 4 CERMBL, 1 CUAC20
        45 t,  74046 CBills, 4    BV, Shadowcat C (Clan) - ImprJJ, EOptics, 1 DHS - 1 CGauss, 1 CERPPC
        70 t,  88250 CBills, 1    BV, Warhammer D (Inner Sphere) - EOptics - 1 LRM15, 1 UAC10, 1 ERPPC, 2 MBL, 2 AC2
Total: 455 t, 707629 CBills, 3.00 BV

Picked the following assets (8/8 in total) after 4 failed tries (0.312s):
        80 t, 109247 CBills, 3    BV, Awesome E (Inner Sphere) - BAP, EOptics, 4 DHS - 2 ERPPC, 2 LtGauss
        90 t, 145946 CBills, 4    BV, Blood Asp Prime (Clan) - EOptics, 6 DHS - 2 CMPL, 2 CGauss, 4 CHML, 1 CSRM4
        65 t,  81997 CBills, 2    BV, Catapult Prime (Inner Sphere) - StndJJ, EOptics, 1 LAMS, 4 HS - 2 LRM20, 4 MBL
        90 t, 125398 CBills, 3    BV, Madcat MKII A (Clan) - ImprJJ, 2 DHS - 2 CLBX20, 2 CERMBL, 1 CLBX10
        80 t,  58180 CBills, 5    BV, Partisan B (Inner Sphere) - EOptics, 1 DHS - 4 RAC2
        30 t,  49333 CBills, 2    BV, SparrowHawk B (Inner Sphere) - C3, EOptics - 4 LBX2
        70 t, 100222 CBills, 2    BV, Thor F (Clan) - StndJJ, 6 DHS - 1 CDSSRM6, 1 CHLL, 1 CLBX20
        60 t,  86097 CBills, 2    BV, Visigoth D (Clan) - EOptics, 1 DHS - 2 CUAC5, 1 TAG, 4 ATM6
Total: 565 t, 756420 CBills, 2.88 BV

I'll have a look at these lists now. :)

EDIT2: Well, the lists would probably need a couple additional restrictions. They are pretty random right now.^^

Offline Starch1ld

  • Lance Captain
  • ***
  • Posts: 688
  • l33tp0intz: +47/-25
  • hi
Re: Chaos March - Feedback Thread
« Reply #23 on: January 21, 2015, 02:42:38 PM »
I would assume that BV is not a good measure to generate droplists by - they probably stem from the boardgame, right? Out of tonnage, price and battle value I think the price is the one that gives the best chance of producing a balanced list of assets.
Not sure, the BV system for MWLL was originally ideated by Invictus. I just tweaked some values and implemented it in the Drop Calculator.
We've been using BV+Ton restrictions for months, and I can tell you that it does work instead, and pretty well.

Offline Starch1ld

  • Lance Captain
  • ***
  • Posts: 688
  • l33tp0intz: +47/-25
  • hi
Re: Chaos March - Feedback Thread
« Reply #24 on: January 21, 2015, 03:51:11 PM »
I don't really know how to use the BV, though. I just assumed "8 players -> 24 total BV", even though some other approach might be better.
That's maybe too much.
With the current ruleset, the battle supervisors decides the BV value for each droplist, and then the drop commanders build them.
For these random generated drop decks, I was thinking about a fixed value of 2 or 2.25 points per player. maybe 2 is better.

Offline Silvercraft

  • Recruit
  • *
  • Posts: 133
  • l33tp0intz: +43/-0
Re: Chaos March - Feedback Thread
« Reply #25 on: January 21, 2015, 05:37:14 PM »
EDIT2: Well, the lists would probably need a couple additional restrictions. They are pretty random right now.^^

Here are some that I think should be considered:

ASF/VTOL asset class (not all maps have air bases)
Shiva A asset variant (some things just are unbalanced)
Longtom (not suitable for all games)

Weapon specific restrictions
(no Arrow4s on Thunderrift, thank you!).


User interface options to exclude unwanted things would be good (if it is not too hard to make).


Offline cest73

  • Lance Captain
  • ***
  • Posts: 733
  • l33tp0intz: +21/-17
  • there are links in my signature -->
    • MW:LL game status online (kudos: Spooky)
Re: Chaos March - Feedback Thread
« Reply #26 on: January 21, 2015, 09:05:21 PM »
i would also "wish"  ::)
the BV to be editable and "outside" on the UI:
we could make drops with special BV  classes?
(as low as 2 and as high as 4)?
I do agree the lower BV do increase the level of fun in the games and the level of experience by using less popular assets (kthx ELH for "sidetracking" my experience - i enjoyed it a lot!)

and as i wrote earlier:
I do like handicap scenarios:
1. extraction (sadly none so far)
2. hold line with handicap (low speed, low ammo, low armor or else)
3. capture/abduction (capture/don't kill specific player/mech)
4. assassination (kill only one mech and extract w/o engagement)

this would provide ton of fun :D imo (players of mil-sim airsoft games?).

and we need the mega rant threads:
Code: [Select]
A. Chaos March: what annoyed me most
and

Code: [Select]
B. Chaos March: why I don't play it anymore
and maybe
Code: [Select]
C. Chaos March: when I will come back to play again
cross-posting will be sanctioned  >:(!
so we have a thread to went (mandatory for organized events)
and a thread to feedback player loss from organizational failures/shortcomings

why:
every serous enjoy-er of any M.W./B.T./FASA  game knows by the hart:
Code: [Select]
either [i]rant[/i] or [u]self improve[/u].doing one makes you kill-off the other
so we will both know: -who the rant-crows are and -we will improve upon them ;)

And we do should (and do need) to have evidence of why players leave if ever?  ???
We lost enough already

also
we could make alternate weekends where we would rest Chaos March and test options like the drop-auto-calc/drop-list-gen (DroLGen?)
so we don't wreck Chaos March, and still develop towards the higher level of organization?


 

Offline thEClaw

  • Star Captain
  • ***
  • Posts: 976
  • l33tp0intz: +75/-0
Re: Chaos March - Feedback Thread
« Reply #27 on: January 21, 2015, 09:22:25 PM »
User interface to make the criteria editable (or add new ones, delete others etc.) is in the works, but due to my inexperience with QML it will take some time and most likely won't be too pretty. Editing the .ini-file for the criteria manually is kind of painful, even for me.

Offline xInVicTuSx

  • Apprentice Dev
  • Living Legend
  • *
  • Posts: 3683
  • l33tp0intz: +193/-5
  • Knight of the Inner Sphere
Re: Chaos March - Feedback Thread
« Reply #28 on: January 22, 2015, 12:50:56 AM »
BV was just a term I used form the boardgame to rate each variant by its usefulness.

If you go by tonnage/price alone there are still plenty of variants that will slip through the cracks, its probably best to try to use all three criteria.

@Star

Either a APC/BA or a sparrowhawk... I'm serious. I want to see more sparrowhawks god damn it.

@VIVI

The first regiment system was created before the BV system so it had problems, however I tried to balance them somewhat. What happened was I too rigidly separated clan and IS assets into their own regiments. IT was really designed for 8v8 and above but got used for 4v4 and 5v5 and just didn't work at all. I retooled it with the BV system but by then duelist and star had gone with cherry picking from regiments to try and squeeze the best list possible. This was a nono but people were complaining that they were stuck with suboptimal assets while sometimes the still enemy had superlists. Finally they took the BV and tonnage system and generated a pool of assets for each side to pick from and create a droplist from. But that has gotten to be far too much of a pain to set up and away from the original intention of the Chaos March which was to keep things simple and fluid, which is what the regiment system was designed to do.

It seems now we're coming full circle with the difference that rather than being handmade, regiments (droplists) will be randomized. It may work this time if there are enough quantitative criteria to generate a "balanced" list within reasonable limits for most games with minimal if any tweaks to them.
-Invictus ne Vindicetur-

Offline Starch1ld

  • Lance Captain
  • ***
  • Posts: 688
  • l33tp0intz: +47/-25
  • hi
Re: Chaos March - Feedback Thread
« Reply #29 on: January 22, 2015, 03:59:22 AM »
Either a APC/BA or a sparrowhawk... I'm serious. I want to see more sparrowhawks god damn it.
AC2 sparrowhawks with UNLIMITED respawns. We must do that once.  ;D