Some M6 OOS Combat Findings

[Litcube]

Not sure that it belongs here, but S&M is my home, plus most of the discussions involving tweaking OOS combat seem to happen here. Plus, I wrote a script to get this output.

The script pitted an M6 versus an M6. All of them, every combination, 15 matches each (actually 30, as I didn't exclude double matchups). Hope someone finds this info useful. Watched every battle OOS.

I find the results to be pretty consistent, actually. If you look two different matchups, for example, The Vali vs. the Osprey, and then the Osprey Vs. the Vali, The former won 15/15, while the latter won 0/15.

So whoever said don't use ammo based weapons in OOS might find this interesting (see the loadout for the Vali below).

My findings do not match Wyvern1's findings in regards to the best M6 OOS, it seems (his show the Springblossom as the top). Either I'm doing something wrong, or I can't read German (which I can't, I might conclude).


Download the JPEG spreadsheet here

Numbers indicate battles won by ships in the left-most column.

Last column is the sum of all battles won by that ship's row of battles out of 15.


Loadouts were equipped with highest ranking weapons based on Wyvern1's laser findings in the German forum, thusly:


------------------------------------
Washi Hauler
Concussion Impulse Generator x 10
Fragmentation Bomb Launcher x 2
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 950
------------------------------------
Vidar
Matter/Anti-Matter Launcher x 10
Electro-Magnetic Plasma Cannon x 4
Matter/Anti-matter Warhead x 149
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 500
------------------------------------
Vali
Matter/Anti-Matter Launcher x 14
Matter/Anti-matter Warhead x 209
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 280
------------------------------------
Truelight Seeker
Point Singularity Projector x 6
High Energy Plasma Thrower x 1
Rudder Optimisation x 19
Engine Tuning x 13
Cargo Bay Extension x 1000
------------------------------------
Springblossom
Prototype Starburst Shockwave Cannon x 14
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 30
Cargo Bay Extension x 2600
------------------------------------
Skiron
Concussion Impulse Generator x 8
High Energy Plasma Thrower x 6
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 710
------------------------------------
PX
Fragmentation Bomb Launcher x 8
Particle Accelerator Cannon x 4
Rudder Optimisation x 19
Engine Tuning x 13
Cargo Bay Extension x 800
------------------------------------
Pirate Osprey
Concussion Impulse Generator x 10
Fragmentation Bomb Launcher x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 2000
------------------------------------
Pirate Centaur
Concussion Impulse Generator x 8
High Energy Plasma Thrower x 3
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 980
------------------------------------
P
Fragmentation Bomb Launcher x 9
Particle Accelerator Cannon x 2
Rudder Optimisation x 19
Engine Tuning x 13
Cargo Bay Extension x 1300
------------------------------------
Osprey
Concussion Impulse Generator x 8
Fragmentation Bomb Launcher x 3
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 2000
------------------------------------
Nemesis
Concussion Impulse Generator x 8
Fragmentation Bomb Launcher x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 350
------------------------------------
Military Transport
High Energy Plasma Thrower x 6
Rudder Optimisation x 10
Engine Tuning x 8
Cargo Bay Extension x 500
------------------------------------
Marauder Hauler
Concussion Impulse Generator x 8
High Energy Plasma Thrower x 3
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 500
------------------------------------
Katana
Matter/Anti-Matter Launcher x 10
Electro-Magnetic Plasma Cannon x 2
Matter/Anti-matter Warhead x 149
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 600
------------------------------------
Hyperion Vanguard
Concussion Impulse Generator x 8
Fragmentation Bomb Launcher x 10
Rudder Optimisation x 18
Engine Tuning x 15
Cargo Bay Extension x 1000
------------------------------------
Hydra
Concussion Impulse Generator x 8
High Energy Plasma Thrower x 1
Rudder Optimisation x 14
Engine Tuning x 14
Cargo Bay Extension x 2000
------------------------------------
Heavy Osprey
Concussion Impulse Generator x 14
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 12
Engine Tuning x 12
Cargo Bay Extension x 1700
------------------------------------
Heavy Nemesis
Concussion Impulse Generator x 12
Fragmentation Bomb Launcher x 2
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 320
------------------------------------
Heavy Hydra
Concussion Impulse Generator x 10
Rudder Optimisation x 8
Engine Tuning x 6
Cargo Bay Extension x 250
------------------------------------
Heavy Dragon
Concussion Impulse Generator x 12
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 9
Engine Tuning x 9
Cargo Bay Extension x 350
------------------------------------
Heavy Centaur Prototype
Concussion Impulse Generator x 12
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 500
------------------------------------
Heavy Centaur
Concussion Impulse Generator x 12
Fight Command Software MK2 x 1
Fight Command Software MK1 x 1
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 950
------------------------------------
Dragon
Concussion Impulse Generator x 8
High Energy Plasma Thrower x 1
Rudder Optimisation x 9
Engine Tuning x 9
Cargo Bay Extension x 800
------------------------------------
Corvette
Beta Kyon Emitter x 15
Rudder Optimisation x 5
Engine Tuning x 5
Cargo Bay Extension x 1000
------------------------------------
Centaur
Concussion Impulse Generator x 8
High Energy Plasma Thrower x 3
Rudder Optimisation x 10
Engine Tuning x 10
Cargo Bay Extension x 1000
------------------------------------

[ADStryker]

Nice ...

Maybe it's worth averaging the wins for X vs Y with the losses for Y vs X.

I notice that the Springblossom beat the Corvette 4 times (implying that the Corvette won 11 times), while the Corvette beat the Springblossom 6 times (implying that the Springblossom won 9 times). Can't really make much of that without averaging, I think (i.e., Springblossom won, on average, 6.5/15, while the Corvette won, on average, 8.5/15).

Looks as though, in general, the heavier the shields the higher the ranking. Two obvious exceptions are the Corvette (which has good shields, but ranked poorly -- probably because of relatively weak weapons) and the Truelight Seeker (which has poor shields but ranked in mid-pack -- probably because it can mount cap-ship weapons).

[Litcube]

Can't really make much of that without averaging, I think (i.e., Springblossom won, on average, 6.5/15, while the Corvette
won, on average, 8.5/15)..

I felt it was important to also illustrate the expected irregularities in combat. These M6s are pretty even matched ships, really. Out of two 15 round matches, 6 wins or 8 wins is pretty even. It's the 15s that provide reliable info.

100 matches would produce better results. But this test itself took about 3-4 hours in SETA.

[Serial Kicked]

Yep, interesting

I always thought the differences between most M6 were overstated. On the other hand, 15 battle is a very low count for consistent results.

[djrygar]

and then there are things like Hmmerhead missile, that can be mounted in most corvettes. WIll devastate any M6 in one round

DId I read properly that you were observing this battle OOS? (using some satellite I guess? Then this was not real OOS. It was semi-OOS

if you watch battle using satellites, each turn/round has 5 seconds, id you dont, 30 seconds

[ADStryker]

...I felt it was important to also illustrate the expected irregularities in combat...

True enough. There is often no easy way to present complex data -- sometimes one must choose a specific presentation for each significant aspect of the data.

[Bobucles]

What you were more likely seeing is that the first ship to fire wins. This is especially true of unsupervised OOS combat, where most ships will simply 1-shot each other.

[Gazz]

If I may suggest a refinement to the test script...

For ideal results you may want to cut out the middle men - the combat scripts.

OOS, ships get the same number of rounds and fire alternatingly like a mexican stand off.

So basically you want a script to...

- spawn ships and alternate who gets the first shot
- Ship 1 -> fire turret null at Ship 2
- Ship 2 -> fire turret null at Ship 1
- repeat until destroyed, log result.

This way you get 100% accurate "OOS results" - excluding missile use - only 1000 times faster because there is no 5-30 sec wait.
With a LOT of results you can get far more accurate average results.
Statistically speaking, 15 points of data aren't worth the paper they are printed on. At 30+ you slowly get into the range of useful results. A few thousand are preferrable.

fire turret null is the only instruction that matters for OOS combat.

[Nafensoriel]

If someone makes a script that does this automatically I have a spare machine to run whatever for days if needs be.

[Litcube]

if you watch battle using satellites, each turn/round has 5 seconds, id you dont, 30 seconds

True. Timestep changes, but the outcome is the same. OSS combat is the same regardless of timestep.

[Litcube]

- spawn ships and alternate who gets the first shot
- Ship 1 -> fire turret null at Ship 2
- Ship 2 -> fire turret null at Ship 1
- repeat until destroyed, log result.

I'm a do this now.

[Litcube]

What you were more likely seeing is that the first ship to fire wins. This is especially true of unsupervised OOS combat, where most ships will simply 1-shot each other.

This is not happening here. Most of these ships went through several rounds or more.

If that was the case, this test shouldn't have taken more than 10 mins.

[Litcube]

With Gazz's suggestion:

Download the JPEG spreadsheet here


Alot faster! However, different results. The rankings are very close, but the amount of wins is decreased or averaged out a little more.

I suspect this has to do with the initiative not being given to the fastest ship anymore, but my my forced alternation of attacker always getting the initiative.

Gazz, any comments?

[ADStryker]

A few interesting threads with some relevance to this comparison:

[SCR] OOS Combat Rebalance (alpha 0.31 - 10.07.09)

OOS combat - how SHOULD it work?

OOS Combat Resolution (Was: OOS Missile firing test.)

(edit ...)

I asked Wyvern11 about his work on this, and he graciously contributed the following...

From: wyvern11
To: ADStryker
Posted: Mon, 28. Jun 10, 08:14
Subject: Re: M6 OOS combat rankings and Weapon rankings
First of all it's nice to hear, that my findings still run around in one form or the other.

Secondly, I have not played X3 for some time and I am not sure if I will again in the near future, this mainly because of poor OOS (and poor forecasts to 2.5++). Maybe I could piece it together from memory. because some threads were english, you need no google translate:

First hook was one of may "my one beats your one" discussions OOS-wise and the question "do missiles work OOS" by jlehtone

OOS Combat Resolution (Was: OOS Missile firing test.)

here I got the notion, how OOS basically works (and why it sucks most of the time).

In x3r I began basically like litcube to mount ship against ship, but i scripted it to do *a battle each sector and all bystanders erased* litcubes problem probably is, that attacker always wins, because defender seldom fights back, if there is lots of universe population left.

in short, outcomes where absurdly different from IS-expectancy, because high span in each hit (some 100 Points -> some 100000 Points) for some ships in same class.

Based on Q-superiority lucike, gazz and me began a more earnest research during which gazz modifed the oos-damage-script to log out each hit into a textfile. now we pieced both together: identical battles and each hit gets logged, crunch the numbers through linux-grep sed-awk (interesting stuff with tables begins about page 15)

OOS-Sektorwachen im Reich der Xenon

gazz found the findings such awful that he began this :

[SCR] OOS Combat Rebalance (alpha 0.31 - 10.07.09)

development stopped because of new development in OOS for 2.5++, you need L3+ for these to read I did some tests on new style and found it even worse. main thing was that (at my time of discussion-participance) only one random turret was summed up for damage for each combat round.

I looked up the thread and found it closed by admin, due to upcoming shipment, so i'm not sure how this worked out in the end

2.5 and earlier:
damage = sum of damages of all guns times no of turrets (so more turrets and more guns rule by a factor)

plans for 2.5++:
damage = sum of damages of a random turret (problem : take the centaur - it has a one in 4 chance to score with his front guns -> average damage is lower than that of a dragon with a 1 in 2 chance -> more guns less damage ????)

ammo-wielders work, as long as they have ammo. but there is a bug: if you have only one ammo-wielder on mounted and it runs out of ammo, then the ship will deal zero points of damage with each hit literally, even if it has 13 guns with ammo or ammo-free.

my mass driving katana-oos where always empty only a few fights after remunitioning

sincerely hope, this helps. if you don't care for modified-flag, I could pass gazz's and my scripts to you, if i still find them

see you

wyvern

_________________
Redest du noch - oder denkst du schon ?

[LV]

someone needs to remember we have an edit button don't they !

[Serial Kicked]

Who attacks first? i guess it's the horizontal line according to the results, but just in case i'd ask.

Those Same ship vs Same Ship in the chart are nearly always giving a result of 1 or 2 victories. I find that somewhat weird, the previous chart was way more balanced.

http://members.shaw.ca/litcube/X3/OOSCo ... attles.jpg
vs
http://members.shaw.ca/litcube/X3/OOSCo ... ttles2.jpg

[Gazz]

Speed DOES play a role in getting the first shot because of how the timing and combat scripts work.

The likelyhood of getting the first shot is Ship1speed / Ship2speed.

The reason is simple. OOS, all lasers have a range of 3 (or 4?) km. The faster ship has a higher chance to MOVE into firing range and execute it's combat script.
The defender is hit and it's signal_attack event triggers it's own combat script.


The previous chart was also based on very few points of data so it was a good guess. =)

I suspect this has to do with the initiative not being given to the fastest ship anymore, but my my forced alternation of attacker always getting the initiative.

Random initiative would reduce the usefulness of your data.

If you want to test the speed effect on initiative, let identical ships duel but randomly assign initiative to the attacker based on a fictious speed ratio.
Each set of such fights would be a new set of data with the ships having for instance 100:120 speed and therefore a 100:120 chance of getting the first shot.
Takes a lot of data to build a table but you can prove once and for all how useless or important initiative is.

[Alan Phipps]

Following on the point the OP made way back on ammo-based weapons provided with ammo now being very effective OOS, I think that while that is true of his all-ammo loadouts, the latest thinking is that you should not mix ammo-based and energy weapons on the same ship used OOS.
Bn did a lot of good research on this here: http://forum.egosoft.com/viewtopic.php?t=275357

[Litcube]

Who attacks first? i guess it's the horizontal line according to the results, but just in case i'd ask.

Those Same ship vs Same Ship in the chart are nearly always giving a result of 1 or 2 victories. I find that somewhat weird, the previous chart was way more balanced.

True. And I think it's because Gazz's suggested method isn't allowing for shield recharge time which can skew results if you had two ships with similar weapons, shields, speed, hull, etc., and very different shield generators. But that's another guess. And you're correct about who attacks first.

Really, to reproduce real world results, you'd have to recreate the conditions in which players would actually experience combat; fight.attack.object.std.

That said, the results are very similar in terms of averages and rankings.

I'd like to know, Gazz, how you might suggest implementing missiles into the script that would accurately reproduce fight.attack.object.std. I've examined the file, but it's unclear, and it sounds like you done this before, or can read German .

Code: Select all

$Died = [FALSE] 
      while not $Died 
        = $Attacker->fire lasers on target $Defender using turret null
        = $Defender->fire lasers on target $Attacker using turret null
        = wait 1 ms
        skip if $Attacker->exists
          $Died = [TRUE] 
        skip if $Defender->exists
          $Died = [TRUE] 
      end

[Gazz]

Missile use is VERY random and any accurate representation of "random" ships is hopeless. You might see a trend with several thousand fights.
Missiles also use the ship's fightskill when used in the default combat script (I think) so you'd have to watch that variable, too.

I don't recall the exact code but it could be that firing a missile would skip laser fire for this turn. Usually a very very bad tradeoff.

Shield recharge does not affect OOS combat much. Bigger ship vs smaller ship is practically always an instakill and equal ships rarely see more than 2 or 3 combat rounds.