Melee Power Rankings Chart

[Pillin]

Ahem.. looking over the chart i see some errors i've done.
replace these two in creatures and that should help

name=AR_Hoplite hp=11 mrl=12 mrst=10 bEnc=3 str=11 att=11 def=11 prec=10 ap=12 gold=11 res=0
name=AR_Hypaspist hp=11 mrl=14 mrst=10 bEnc=3 str=11 att=11 def=12 prec=10 ap=13 gold=15 res=0

[Targa]

Originally posted by Saber Cherry:
I'm running the sim now for Arco units versus all other units (a rectangular matrix). This will give you the relative values of arco units, but the full (square) matrix will take longer (maybe 30 minutes versus 6 hours) so I won't do it until tomorrow.

I'm looking forward to the results of the triangular, hexagonal, octagonal, and the reverse-trapezium matrix tests.

Wake up Neo...

[Saber Cherry]

Originally posted by Pillin:
Ahem.. looking over the chart i see some errors i've done. Replace these two in creatures and that should help

Got it=) AP doesn't affect the actual results, though, until I add trample. Or mod-output capability. So the previous results still stand

[Goad]

The combat simulator tells me that the number of ulm guardians or knights I can get for the first two turns of gold/resource will do far better against indies than the same amount spent on pikeneer_1s. But in play, the pikeneers seem to do dramatically better.

This seems to be for three reasons:

1, Most important, the repel causes indies to rout sooner.
2. The greater size of the starting out army with pikeneers causes my army to be less likely to rout.
3. The guardians/knights get bunched up in the default starting forces and don't actually do as much. On the other hand, if two or three knights are deployed such that they engage by themselves, they have a too high chance of dying.

[Saber Cherry]

Originally posted by Goad:
The combat simulator tells me that the number of ulm guardians or knights I can get for the first two turns of gold/resource will do far better against indies than the same amount spent on pikeneer_1s. But in play, the pikeneers seem to do dramatically better.

This seems to be for three reasons:

1, Most important, the repel causes indies to rout sooner.
2. The greater size of the starting out army with pikeneers causes my army to be less likely to rout.
3. The guardians/knights get bunched up in the default starting forces and don't actually do as much. On the other hand, if two or three knights are deployed such that they engage by themselves, they have a too high chance of dying.

Repel is important, and the morale pool is important. However, there's another factor: Being swarmed.

The power ranking chart is generally 1-on-1, while in-game combat occurs many-on-many. This has two MAJOR effects:

1) A defense decrease each time a unit is struck, which is important versus Groups, and irrelevant in 1-on-1 combat.

2) Every action causes fatigue. 6 Pikeneers with 6 encumbrance could attack 7*100/6 = 119 times before passing out. 2 Black Knights with 5 encumbrance can attack only 2*100/5 times = 40 times, plus 2 lance attacks and 40 some 10-damage hoof attacks that won't really kill anything. Total 82 attacks, but considering that half of them are "hoof" attacks... let's say 60 net attacks. So against a horde of weaklings, Black Knights would pass out before killing as many units as Pikeneers. Fatigue also decreases unit stats and so forth.

So, to see whether you're better off with pikeneers or Black Knights, simulate EXACTLY what you want to do:

code:

---

C:\Projects\Java\Dominions>java RectMatrix 7*ulm.pikeneer.1 2*ulm.black.knight 

vs 20*light.inf 10000

..



Deathmatch Power Ratings for 10000 rounds:



7*Ulm Pikeneer 1 ------------ Wins: 70.8%   Power:  1919   Overall:  1397

2*Ulm Black Knight ---------- Wins: 12.0%   Power:   189   Overall:   347

20*Light Inf ---------------- Wins: 58.6%   Power:  1293   Overall:  1000



Average --------------------- Wins: 47.1%   Power:  1133   Overall:   914

Median ---------------------- Wins: 58.6%   Power:  1293   Overall:  1000

---

This fights 7 ulm pikeneers against 20 indy LI, then 2 knights against 20 indy LI. And as you can see, the pikes do better!

For reference, the syntax is:

java RectMatrix #*a1 #*a2 ... vs #*b1 #*b2 ... battles

where
# means the number of that unit type
a1, a2, ... mean the name of a unit
b1, b2, ... mean the name of a unit
"vs" is between the "a" units and "b" units.
"rounds" is the number of rounds for each engagement

This will make each unit on side "a" fight each unit on side "b" in a rectangular matrix. For example, in this scenario:

java RectMatrix 10*a1 10*a2 5*a3 vs 1*b1 1*b2 100

Each engagement would be fought to the death 100 times, and there would be 6 engagements, as follow:

ten a1's versus one b1
ten a2's versus one b1
five a3's versus one b1
ten a1's versus one b1
ten a2's versus one b2
five a3's versus one b2

[ March 05, 2004, 22:44: Message edited by: Saber Cherry ]

[Saber Cherry]

Incidentally...

Note that the big chart in this thread has "Deathmatch" rankings and "Gauntlet" rankings. The difference is that 1-on-1 "Gauntlet" is more like a real battle - units keep replacing the dead until 50 rounds are up, whereas "Deathmatch" ends as soon as one side or the other dies.

So the power rankings in "Gauntlet" mode are much more realistic for normal scenarios -1v1 "Deathmatch" is better for assassinations, Arena combat, and battles between very small armies (like a scout versus some local militia).

However, to really accurately model a large battle (homogenous on each side) you can't use the 1-on-1 numbers from the "Battle Rankings"... instead model the battle with RectMatrix or Fight using unit multipliers as shown below. The power rankings are just a sort of guideline.

If you look at Black Knights and Pikeneers in the Deathmatch mode, Black Knights are 20X as good... but in Gauntlet mode, they are only 2X as good.

Black Templars with a Nature/Earth blessing (reinvigoration/berserk/regeneration) would do much better, though. You can simulate blessing effects in the combat sim, but you have to do it in the unittypes file, not in the command line.

[ March 05, 2004, 23:28: Message edited by: Saber Cherry ]

[Endoperez]

I am bumping the old (ancient) thread having the unit listings Saber Cherry created with the help of her Java-based Combat Simulator. I think a link is somewhere in this thread, and it can be easily found by searching for Combat (not Battle!) Simulator. It produces numerical stats for units fighting. It might not support all the abilities of units, but it has most of them.

This was Last updated in (edit March, so it is a bit outdated... One more reason to learn how to 'pay for the free things'. Check the 'Time to pay up' thread for more information about that.

[ June 11, 2004, 19:08: Message edited by: Endoperez ]