Author Topic: Real BattleValue - based on calculated Battle Performance  (Read 2664 times)

Hptm. Streiger

  • Lieutenant
  • *
  • Posts: 968
  • 3d artist, spread sheet warrior, KTF
Without further derailing the topic about the house rules for Autocannons
I want to start a second discussion about the BattleValue more precisely the BV for weapons.

(First the result for the eye:
Code: [Select]
final Bench SLAS 3
final Bench MLAS 14
final Bench LLAS 45
final Bench PPC 63
final Bench AC2 9
final Bench AC5 23
final Bench AC10 62
final Bench AC20 125
final Bench MG 2
final Bench LRM5 11
final Bench LRM10 42
final Bench LRM15 103
final Bench LRM20 179
final Bench SRM2 6
final Bench SRM4 12
final Bench SRM6 18

The BV weapon values are based on damage - and to hit probability based on 4 - and some global modifications like "headcap", energy. Cluster.. etc.
This allows lots of loop holes in weapon design.

I wanted to see how the "real" performance in the battle is - well first option i play a thousand games and see the outcome - or i just break it down to
  • target
    • movement
    • armor
    • criticals
  • weapon
    • range
    • damage
    • to Hit Mod
  • gunnery of 4

I did use the intro tech level first and tested them vs "real" Mech values (from 3039U (all that are available in the MegaMech Package)

How do I compute this values:
Loop through Mechs
Loop through Ranges in 30m steps (if range is outside stop)
Calculate modifications (movement, jumping, range) roll a for to hit
do all the calculation for hit (missile, hit zone, floating critical, damage)
see if a destroy even occure (no structure in CT, Hd, 3 Engine, 2 Gyro, Cockpit, 3 criticals in Head)
see how many shots were fired inverse this value and sum it together for all ranges and tests
divide the final inverse/shot benchmark value by the number of tests.... (plus some generic rounding and dividing by 100 to have nice values)

Here a simple "protocol" for the difference between cluster and slug damage of the LB10X (C for cluster; S for slug)
Code: [Select]
Final Value LBX10C
Benchmark: 37.5
Avg Number of Shots: 45.6
Avg Number of Hits: 21.9
Avg Number of Critical: 25.3
Avg Number of Headshots: 3.9
Reasons of Destruction:
Center Torso Destroyed:71.9
2 Gyro Hits:0.9
3 Engine Hits:0.7
Pilot Dead:38.9
Ammo Explosion:1.5
-----------------
Final Value LBX10S
Benchmark: 74.6
Avg Number of Shots: 35.5
Avg Number of Hits: 11.6
Avg Number of Critical: 4.5
Avg Number of Headshots: 0.3
Reasons of Destruction:
Center Torso Destroyed:101.5
2 Gyro Hits:0.6
3 Engine Hits:0.2
Pilot Dead:10.0
Ammo Explosion:1.7
-----------------

a more detailed log is in the attachment.

Pls let me know if something is unclear or if you find errors (thinking or math)


« Last Edit: 30 August 2017, 02:40:39 by Hptm. Streiger »

Sabelkatten

  • Lieutenant Colonel
  • *
  • Posts: 6952
Re: Real BattleValue - based on calculated Battle Performance
« Reply #1 on: 30 August 2017, 14:05:09 »
I'm not certain that's a great way of doing it either, you're missing things like hole-punching/crit-seeking combos... I mean, it's certainly better than the standard BV2 system, but I don't know if it's much better.

Anyway, there was one thing that looked strange in the file :

Quote
Avg Number of Shots: 49.0 with a Minimum of 0 and with a Maximum of 151

I figure this is number of shots needed to kill the target... So how can it be zero? ???

Mattlov

  • Catalyst Demo Team
  • Lieutenant
  • *
  • Posts: 1210
  • Fnord.
Re: Real BattleValue - based on calculated Battle Performance
« Reply #2 on: 30 August 2017, 15:01:50 »
I'm not certain that's a great way of doing it either, you're missing things like hole-punching/crit-seeking combos... I mean, it's certainly better than the standard BV2 system, but I don't know if it's much better.

Anyway, there was one thing that looked strange in the file :

I figure this is number of shots needed to kill the target... So how can it be zero? ???

I've seen an Enforcer fail the PSR for rubble terrain and fall on the back, critting the AC 10 ammo.  But that shouldn't effect weapon Battle Value.
"The rules technically allow all sorts of bad ideas." -Moonsword


Hptm. Streiger

  • Lieutenant
  • *
  • Posts: 968
  • 3d artist, spread sheet warrior, KTF
Re: Real BattleValue - based on calculated Battle Performance
« Reply #3 on: 31 August 2017, 02:42:58 »
Anyway, there was one thing that looked strange in the file :

I figure this is number of shots needed to kill the target... So how can it be zero? ???
I've seen an Enforcer fail the PSR for rubble terrain and fall on the back, critting the AC 10 ammo.  But that shouldn't effect weapon Battle Value.

Yup I already have seen it - it is the "wrong terminus"  0 Shots = Can't hit the target because the targetMod > 12
But I have seen some "mistakes" - I did wonder why the LRM20 got so good values were in TT the duel Archer vs Warhammer mostly goes in favor of the Warhammer
I had mixed the "rules" for the Bundle with the damage (if a LRM20 hit with 12 missiles they did 12x4 damage - instead of 5,5,2)

now with some added SDLF weapons:

Quote
I'm not certain that's a great way of doing it either, you're missing things like hole-punching/crit-seeking combos... I mean, it's certainly better than the standard BV2 system, but I don't know if it's much better.
In a later "update" i want to do that - consider weapons and heat of a mech and see if there are some synergy effects
For the beginning I could test some combos however.... (for example HPPC and LB10X vs ERPPC and Gauss on the Falconer)
Although I will do that after I added the "extra" Tech (CASE, XL Engine...) and a "Loadout Capacity" Value Based on crits, heat, weight and ammunition


NameBV
Large Laser47.6
ER-Large Laser62.9
Large Pulse Laser61.2
Medium Laser15.4
Medium Pulse Laser22.9
Small Laser3.5
Small Pulse Laser5.4
Flammer2.1
PPC67.6
ER-PPC94.4
Autocannon/29.3
Autocannon/524.2
Ultra AC/5 - Single Mode25.9
Ultra AC/5 - Ultra Mode37.4
Autocannon/1067.5
LB10-X Slug Shot80.0
LB10-X Cluster Shot38.6
Autocannon/20127.5
Gauss Rifle175.2
Machine Gun2.1
LRM/512.0
LRM/1026.6
LRM/1540.8
LRM/2049.3
LRM/5 + Artemis IV15.3
LRM/10 + Artemis IV32.8
LRM/15 + Artemis IV49.5
LRM/20 + Artemis IV61.1
SRM/26.9
SRM/413.0
SRM/620.1
SRM/2 + Artemis IV8.4
SRM/4 + Artemis IV15.3
SRM/6 + Artemis IV23.5
Streak SRM/29.8

« Last Edit: 31 August 2017, 04:22:27 by Hptm. Streiger »

Sabelkatten

  • Lieutenant Colonel
  • *
  • Posts: 6952
Re: Real BattleValue - based on calculated Battle Performance
« Reply #4 on: 31 August 2017, 13:33:40 »
Some of those values does seem a bit off... IME the AC/20 isn't ~60% more powerful than a GR (when both hit). 40% maybe, but the extra cluster size usually only make a big difference if the target has 15-19 armor.

Hptm. Streiger

  • Lieutenant
  • *
  • Posts: 968
  • 3d artist, spread sheet warrior, KTF
Re: Real BattleValue - based on calculated Battle Performance
« Reply #5 on: 01 September 2017, 02:33:25 »
Some of those values does seem a bit off... IME the AC/20 isn't ~60% more powerful than a GR (when both hit). 40% maybe, but the extra cluster size usually only make a big difference if the target has 15-19 armor.
The current calculation only considers damage and range. you see it with the MG vs Flammer -same range same damage.
When I would add the Clan ERPPC it would get equal if not higher values as the Gauss simple for the fact of 23 max range and no minimal range

I only tested it against pre SDLF tech mechs - only few have max armor and can withstand a single AC20 hit.
Things might differ the more mechs i will add

Hptm. Streiger

  • Lieutenant
  • *
  • Posts: 968
  • 3d artist, spread sheet warrior, KTF
Re: Real BattleValue - based on calculated Battle Performance
« Reply #6 on: 01 September 2017, 09:29:11 »
As suspected when run against Mechs from the TRO 3050u the BV is smaller - reasons are more armor, and AmmoExplosions are not that devastating.
Plus Clan Mechs. Also the Mechs with 7 or less head armor are not that common (in the test 50% of all Rifleman died from a shot in the head. When the mech can take two hits the number drops .
Of course its extreme - because only one weapon is used and when the other armor and structure is though enough there will be a head shot - at least maybe once every 30 hits
Oh noes - have some mistakes there - the LB20X is not sound.....

NameBV
Binary Laser82.5
Large Laser33.1
Medium Laser11.4
Small Laser2.7
-------
ER-Large Laser43.4
ER-Medium Laser16.3
ER-Small Laser4.2
-------
Large Pulse Laser40.9
Medium Pulse Laser17.1
Small Pulse Laser4.2
-------
Large X Pulse Laser58.6
Medium X Pulse Laser23.8
Small X Pulse Laser6.3
-------
Large Variable Speed Pulse Laser58.1
Medium  Variable Speed Pulse Laser27.0
Small Variable Speed Pulse Laser11.2
-------
Flammer1.7
PPC45.2
ER-PPC63.8
Light PPC17.4
Heavy PPC102.7
Snub Noose PPC53.0
-------
Autocannon/27.2
Ultra AC/2 - Single Mode8.1
Ultra AC/2 - Ultra Mode11.8
Rotary AC/2 - 4 Shot Mode19.0
Rotary AC/2 - 6 Shot Mode28.9
LB2-X Slug Shot8.3
LB2-X Cluster Shot9.1
-------
Autocannon/517.5
Ultra AC/5 - Single Mode20.5
Ultra AC/5 - Ultra Mode29.3
Rotary AC/5 - 4 Shot Mode47.3
Rotary AC/5 - 6 Shot Mode71.2
LB5-X Slug Shot20.7
LB5-X Cluster Shot17.0
-------
Autocannon/1045.6
Ultra AC/10 - Single Mode53.6
Ultra AC/10 - Ultra Mode76.3
LB10-X Slug Shot54.1
LB10-X Cluster Shot33.2
-------
Autocannon/2082.9
Ultra AC/20 - Single Mode90.7
Ultra AC/20 - Ultra Mode120.2
LB20-X Slug Shot31.5
LB20-X Cluster Shot41.4
-------
Gauss Rifle152.2
Light Gauss Rifle47.2
Heavy Gauss Rifle133.2
Improved Heavy Gauss Rifle151.1
Machine Gun1.7
-------
LRM/59.0
LRM/1019.6
LRM/1530.1
LRM/2036.7
-------
LRM/5 + Artemis IV11.4
LRM/10 + Artemis IV24.2
LRM/15 + Artemis IV36.7
LRM/20 + Artemis IV45.4
-------
SRM/25.5
SRM/410.3
SRM/615.7
-------
SRM/2 + Artemis IV6.6
SRM/4 + Artemis IV12.1
SRM/6 + Artemis IV18.7
Streak SRM/27.9
-------
Clan ER-Large Laser68.8
Clan ER-Medium Laser28.7
Clan ER-Small Laser8.2
-------
Clan Large Pulse Laser90.6
Clan Medium Pulse Laser36.0
Clan Small Pulse Laser6.9
-------
Clan ER-PPC141.2
Clan Flammer1.7
-------
Clan Ultra AC/2 - Single Mode9.8
Clan Ultra AC/2 - Ultra Mode14.1
Clan LB2-X Slug Shot9.3
Clan LB2-X Cluster Shot10.1
-------
Clan Ultra AC/5 - Single Mode24.1
Clan Ultra AC/5 - Ultra Mode34.5
Clan LB5-X Slug Shot23.3
Clan LB5-X Cluster Shot19.2
-------
Clan Ultra AC/10 - Single Mode53.6
Clan Ultra AC/10 - Ultra Mode75.6
Clan LB10-X Slug Shot53.3
Clan LB10-X Cluster Shot33.4
-------
Clan Ultra AC/20 - Single Mode106.9
Clan Ultra AC/20 - Ultra Mode140.7
Clan LB20-X Slug Shot37.4
Clan LB20-X Cluster Shot41.7
-------
Clan Gauss Rifle134.4
Clan Hyper Assault Gauss 2064.1
Clan Hyper Assault Gauss 30105.8
Clan Hyper Assault Gauss 40140.5
Clan Machine Gun1.7
-------
Clan LRM/514.1
Clan LRM/1030.2
Clan LRM/1545.9
Clan LRM/2055.6
-------
Clan LRM/5 + Artemis IV17.8
Clan LRM/10 + Artemis IV36.9
Clan LRM/15 + Artemis IV55.7
Clan LRM/20 + Artemis IV68.4
-------
Clan SRM/25.5
Clan SRM/410.3
Clan SRM/615.7
-------
Clan SRM/2 + Artemis IV6.7
Clan SRM/4 + Artemis IV12.2
Clan SRM/6 + Artemis IV18.6
-------
Clan Streak SRM/29.9
Clan Streak SRM/410.3
Clan Streak SRM/630.3
« Last Edit: 01 September 2017, 09:31:46 by Hptm. Streiger »

Col Toda

  • Captain
  • *
  • Posts: 2943
Re: Real BattleValue - based on calculated Battle Performance
« Reply #7 on: 02 September 2017, 06:25:41 »
While your approach to derive your numbers is internally accurate the fact that your number of shots to derive your BV values never happens in the course of a single battle . I understand that you have encountered a real discrepancy between given battle value and what you see in a given combat . Everyone has . It is just a tool , a guideline for Solaris VII duels but in a Campaign when you are doing battles it should be a minimum factor . In a campaign when you plan to do an objective raid you use intelligence gathered for it to determine how much force is required to get the job done efficiently . No plan survives contact with the enemy but how close you are to accurate barring bad luck you succeed more often than not . The things that bite you the most is not compared BV of what you think you are going to encounter . The things that upset the expectations the most
. 1 Just bad luck . 2  Out dated info and the opposition force is composed of different elements . 3  the militia response times is shorter than expected . 4 Ghost or other unaccounted forces are present . 5 The target gets moved wholesale before your arrival and you have no chance to succeed from the moment you arrive at your initially intended target so a secondary one should be arranged for that contingency . In order of what can go wrong . Long term victory is as much or more about loss management as winning and achieving short term success . Napoleon won every battle in his Russian campaign but lost the war and his army by losing supply in the field by being over extended . As long as BV is known to be an imperfect guideline no one should be surprised by an encounter's result . Play testing will result in radically different results due to playing style as much as the numbers . The existing numbers only closely work if 10 percent of your battlefield has terrain be it hills or woods . It breaks down as soon as that changes . In open nearly all clear terrain favors long range or fast units city or highly congested terrain favors jumping mechs with short range weapons . All given BV depends on ideal terrain conditions . If you truly want an accurate BV you should have a multiplier for every stock map or a means to derive one . For instance any non jumping mech gets an 1 to a 0 .6  multiplier to BV . For instance an Awesome 8 Q BV   should be multiplied by  0.6  to get  a true value . Why most fast or jumping targets can get in mininum range and full defensive movement modifier . The Awesome has 1 arm mounted PPC against back attacks with a minimum range of 3 . So a regular gunner that did not move is likely shooting for 11 target numbers more or less .


Sabelkatten

  • Lieutenant Colonel
  • *
  • Posts: 6952
Re: Real BattleValue - based on calculated Battle Performance
« Reply #8 on: 02 September 2017, 13:33:13 »
First, please try to use line breaks. Your post is almost unreadable as is.

Second, it's also irrelevant. Not only is it a major misunderstanding of how point balance systems like BV work, this thread is only about weapon BV, not unit BV.

theagent

  • Master Sergeant
  • *
  • Posts: 343
Re: Real BattleValue - based on calculated Battle Performance
« Reply #9 on: 08 September 2017, 14:42:31 »
Congratulations.  You managed to prove that:
  • faster-moving targets are harder to hit/require more shots be fired in order to get the required # of hits
  • the more armor a target has, the more damage it can take before critical hits are inflicted/the target is destroyed
  • the more damage a weapon does per shot, the less it has to fire to destroy the target

Except we already knew all of that.

What is missing from these figures is:
  • how the various benches were used to calculate the "final bench" for each weapon.
  • the justification for saying that a weapon's bench # should drop just because a target has 'advanced tech', even though the weapon itself has not changed
  • why weapon benches do not take range into consideration anymore -- actual ranges, not just the abstract range brackets.

The first item is more of a completeness thing (I.e. 'always show your work').

The second matters because the 3050 models are not the same as their 3039 counterparts.   Some have improved mobility, some have improved armor protection, & some have items that reduce/increase their vulnerability to critical hits.

The last item is because range matters.  Yes, an AC20 does twice as much damage as the AC10...but the AC10 can hit targets nearly twice as far away as the AC20, & through most of its range has better modifiers to shoot.

Sabelkatten

  • Lieutenant Colonel
  • *
  • Posts: 6952
Re: Real BattleValue - based on calculated Battle Performance
« Reply #10 on: 08 September 2017, 18:46:33 »
how the various benches were used to calculate the "final bench" for each weapon.
Unless I misunderstand completely...

1/number of kills at each range added together over the whole range, divided by X (to give a practical number). Essentially effective average damage * range.

Quote
the justification for saying that a weapon's bench # should drop just because a target has 'advanced tech', even though the weapon itself has not changed

Err, I'm fairly certain he means that more complete testing will change the values a bit as the test sample becomes more complete...

Quote
why weapon benches do not take range into consideration anymore -- actual ranges, not just the abstract range brackets.

As noted above I'm fairly sure the calculations are based on the exact ranges.

It can be argued that isn't a completely fair way of doing it, thought. Especially very short-ranged weapons are extremely dependent on the speed of the mounting unit. The MGs on a Locust are far more valuable than the SL on an Awesome, while the LRMs on a -1M Locust have only a little more value than the LRMs on an Atlas.

In a BV discussion a year or two ago the example of a 5-ton (IIRC) range 1/2/3 AC/20 came up. BV says it's worth 1/3rd of a standard AC/20 - but since you can mount it on a 7/11 unit without difficulty it's only marginally harder to bring into range.

Hptm. Streiger

  • Lieutenant
  • *
  • Posts: 968
  • 3d artist, spread sheet warrior, KTF
Re: Real BattleValue - based on calculated Battle Performance
« Reply #11 on: 19 September 2017, 01:41:39 »
Thx for the replies - yes at the current state there is nothing "new"
however the placement of ammunition and the armor of the head location has some impacts - and hey of course we knew but was this already included in the BV calculation?
Take the Hunch 4G for example or what about the duel PHX vs Panther or vs Hollander.... the Hollander gets a 1.2 for head capping on its gauss the Panther not. But both mechs can head cap the PHX with a single hit.

however this is only the first step - from here i could start to consider the speed of a unit, heat, range brackets, firing pattern - maybe even some "movement" that might simulate a couple of turns (for example the Falconer 9R - firing HPPC and Slug at medium range and switching to cluster and MLAS at close)