Reconstructing My EDU System

Hey guys,

So I lost all the files on which I kept my old EDU system. I liked it a lot, so I am attempt to recompile it bit by bit. I am going to copy-paste an email I sent to ryoga in the next post and post more things I managed to reconstruct in this thread. Please let me know if you have any questions - this stuff is very important and I'm very keen on getting it right.

Unit cost and stats were determined based on three characteristics: culture, skill, and equipment. Culture is something that is specific to the unit - i.e. if I say it's from a "Germanic" or a "Steppe" culture then it gets XYZ bonuses and penalties, almost all of which pretty much only affect the unit's cost. These will make sense in due time.


The basic formula was this:

[[CONSTANT 1] * (aggregate unit skill stat) + (aggregate equipment cost)] * CONSTANT * NUMBER OF MEN IN UNIT * ANY OTHER RELEVANT MODIFIER

I actually managed to find the table I used to calculate the most basic stat skill: the combat skill relative to the mean. This was calculated based on the unit's change of defeating soldier of baseline skill in combat based on changes to its base attack.

To explain: The base melee attack of each unit (not considering weapons, equipment, cavalry vs. infantry etc.) was 5 attack and the base defense of each unit (not considering armor, etc.) was 10 defense (skill, no armor or shield). Since adding one to attack skill and adding one to defense skill have statistically identical effects on one unit's chance of beating another in a fight (this is axiomatic based on how RTW combat works) then a unit's effective base melee combat strength was called +0 if its base stats added up to 15.

Since the base combat skill cost wasn't determined by category but rather by deviation from the base value, this gives a lot of flexibility compared to the simple EB system. For example, I can give a unit who I want to give a very high defense +3 defense compared to the baseline (this is a lot - it represents about a 33% greater chance of victory in single combat, which adds up rather quickly across an entire unit) but -2 attack (effectively a 21% decrease in single combat victory chance) so that its net strength was +1.

These are the chances of victory in single combat against a unit with no base stat changes (i.e. +0, base attack 5, base defense 10) associated with changes in base melee stats (attack and defense combined):

-4: .406
-3: .429
-2: .452
-1: .476
+0: .500 (50%)
+1: .524
+2: .548
+3: .571
+4: .594
+5: 0.617
+6: 0.639
+7: 0.661

You'd raise these victory chances to the fourth power in order to get the base cost modifier. Therefore, the base skill cost of a unit with +0 skill is .0625 and the base skill cost of one with +6 is 0.1667. This would be added to other skill costs (projectile weapon skill being the most important one) and then multiplied be a constant to determine the aggregate effect on unit cost.

I didn't have a system of morale-based costing, so I just assigned morale values that I thought made sense to various units, usually standardizing them for a given faction. Cavalry units' morale was determined by what mounts they used, though.

I basically never used anything above +6 (that's what I usually gave a faction's best elite units) or below -4 (that's what I gave crappy levy slingers), but you can if you want. Keep in mind the elite units almost always had stuff like better armor and everything too. Again, I'll give you something will all of this once I get it all together again - since you should probably be focusing on conceptual stuff, I will take some time to make sure I get it right.

This all seems rather complicated (trust me it's only the first part), but once you got used to it, it was a very convenient way of statting units quickly. I probably should have written a program to do it for me, but I was about 16 years old when I came up with all of this (I'm 21 now) and didn't have nearly as much knowledge back then.

As for the missile attack, that's a good question. In EBNOM, I took advantage of something I came up when making EDUs for EB multiplayer, and that's missile accuracy. You can actually alter the accuracy of missiles in descr_projectile_new. I think the base accuracy is something very high, like 0.01 or lower, and I thinks this represents something like a standard deviation of accuracy, i.e. how far off the target the projectile tends to be. I'm not sure since I never conducted tests on it. It's best to illustrate with examples:

(these are from my current download of the EBNOM files - I don't know if the links I uploaded even work anymore)

projectile arrow_cb_l

;effect default_arrow_trail_set
effect_offset -1.5
damage 0
radius 0.025
accuracy_vs_units 0.07
affected_by_rain
min_angle -60
max_angle 50
velocity 50
display particle_trail
triangle
texture data/textures/arrow_test.tga
tail 0.3
length 2.5
tail_tex0 0 0.0
tail_tex1 0 0.25
head_tex 1 0.125
end

projectile arrow_ha_m

destroy_after_max_range_percent_and_variation 50 220
;effect default_arrow_trail_set
effect_offset -1.5
damage 0
radius 0.025
accuracy_vs_units 0.06
affected_by_rain
min_angle -60
max_angle 70
velocity 70
display particle_trail
triangle
texture data/textures/arrow_test.tga
tail 0.3
length 2.5
tail_tex0 0 0.0
tail_tex1 0 0.25
head_tex 1 0.125
end

projectile javelin_l

damage 0
radius 0
accuracy_vs_units 0.25
min_angle -60
max_angle 70
velocity 32
display aimed invert_model_z
model data/models_missile/weapon_javelin_high.cas, 40
model data/models_missile/weapon_javelin_low.cas, max

projectile bullet_h

effect small_arrow_trail_set
damage 0
radius 0.025
mass 0.01
accuracy_vs_units 0.09
min_angle -60
max_angle 30
velocity 80
display aimed
model bi/EBNOM/data/models_missile/sling_bullet.cas, 40
model bi/EBNOM/data/models_missile/sling_bullet.cas, max

The projectiles are named based on what the projectile actually is, how skilled/accurate the soldier is, and based on whether the user is mounted or not (if not indicated otherwise, the user is an infantryman). For example, "arrow_cb_l" means "low skill user of a composite bow"

There are subtle differences between the projectile speed and other characteristics of the different types of bows (and they have different capabilities in the stats) but in terms of accuracy the different types of bows have identical accuracies. In general, infantry are more accurate than cavalry (which makes intuitive sense) and bows have the highest accuracy, followed by slings and then javelins. I made javelin accuracy vary enormously, though: low-skill javelin users are supposed to effectively be "point and shoot" types where they throw a ton of stuff and hope it hits something, whereas high-skill users will be able to nail their targets. The gradations between other projectile accuracies aren't as high, but since other types of projectiles tend to be .

To emphasize the point of there being no blanket classifications for units - "elite" ones do not necessarily have good accuracy (most of the cataphract horse archers had regular accuracy I think) and otherwise unimpressive units can be freely statted with high accuracy (for example, Balearic slingers). The cost of gaining skill with each kind of weapon is calculated as a skill cost, with javelins being by far the lowest cost and slings the highest (though elongated composite bows are the most expensive weapon and basic stone slings the least).

CONST 1 * CONST is equal to a number around 845, at least according to my calculations.

My guess is that CONST1 = 425 and CONST = 0.2, since I remember using the number 0.2

gamegeek2 wrote:CONST 1 * CONST is equal to a number around 845, at least according to my calculations.

My guess is that CONST1 = 425 and CONST = 0.2, since I remember using the number 0.2

So, I've done some more comparisons. I can't be positive what the actual numbers I used were originally, but I'm pretty positive CONST (i.e. what you multiply skill cost + equipment cost by) is 0.2 and we can work with 400 for now since that's an easy number. I'll start working on equipment cost values tomorrow.

It's no big deal if I don't reconstruct the exact system, as long as we end up with something relatively close to it.

Great news!

by the way, the formula I was working on was like, feel free to use it if you want

TOTAL COST = SOLDIERS * { WEAPON1 + WEAPON2 + DEFENCE + MORALE + TYPE + CLASS}

with:
SOLDIERS: #men * collision_mass * #hitpoint
WEAPON [close-combat]: (Attack_factor * Charge_bonus) * (attack_celerity +1) /10
WEAPON [ranged]: #shot * (Attack_factor * Charge_bonus) /20
DEFENCE: (shield1 + defence_factor1 + armour1 + shield2 + defence_factor2 + armour2) /10
MORALE: morale + COSTANT1 + COSTANT2

• COSTANT1 (discipline): low: -1; normal: +0; disciplined: +1; impetuous: +2
  
• COSTANT2 (training): untrained; +0; trained: +1; highly_trained: +2
  

TYPE: COSTANT3 + COSTANT4 + Culture-bonus

• COSTANT3: Skirmisher: +0; Spearmen: +2; Light: +1; Heavy: +4; Archer/Slinger: -1
  
• COSTANT4: Infantry: +0; Light Cavarly: +1; Medium Cavalry: +3; Heavy Cavalry: +5
  
• Culture-bonus: this is a little tricky, as it was intended for the sake of boosting the production of certain units. For example, for nomad/horse-based factions all the cavarly would got a "negative" bonus, meaning  the units would cost less then they should, romans would have a "positive" bonus on every heavy cavalry, and every "spearman" cavaly, meaning they would pay more for all the good cavarly. I was still working on it.
  

CLASS: Militia: +0; Regulars: +1; Elite: +3

There is no reason to divide units into specific classes; it allows for way more flexibility if you do not. You can think of things as being in "Tiers" with higher tiers being better (unlike usual) if that's convenient. These are what I usually used

Tier -4 = -4 below base: crappy light archers, etc. that don't like melee (Sphendonetai, Persian Archers)
Tier -2 = -2 below base: crappy skirmishers, low-quality levy spears
Tier 0 = no change from base: barbarian/rural militias, basic not amazing melee soldiers (Thureophoroi, Germanic/Celtic spear militias)
Tier 1 = +1 above base: "bread and butter" melee combat units (Legions, Machairophoroi)
Tier 2 = +2 above base: above-average skill professionals (Thorakitai, Boii Swordsmen)
Tier 3 = +3 above base: veteran professionals, nobility (Neitoi, Evocatae, Lonchophoroi)
Tier 4 = +4 above base: skilled nobility, high quality professionals (Spathaphoroi, Azat Cataphracts, Celtic noble infantry/cav)
Tier 5 = +5 above base: mid-tier elites (hellenic cataphracts, Praetorians)
Tier 6 = +6 above base: best of the best, faction signature units (Elite cataphracts, bodyguard units, royal foot guards, etc)

It's a sliding scale and there's plenty of room to change things.

ok, I like it

Put a lot of values into a spreadsheet today for convenience. Going to calibrate things a bit so they don't seem too expensive.

I don't think that's really a problem, if the stats and the costs are balanced, we can give or take money with bonus or malus on the buildings (or through script, for the AI).

Well, I do my EDU designing such that it's fair in stand-alone custom battles as well as in campaigns.

of course, I meant all-around balanced, not faction-wise ^^

OK. By the way, do you guys have any particular balance ideas faction-wise? Go ahead and throw out suggestions for what you'd like to see. I'll put up some of my own in a bit.

You mean something like "roman infantry should be the more reliable, celts and germans should have high charge bonus and the parthian cavalry should have high resistance" or something more definite, like base stats and alghorithm?

I mean definitive cost suggestions, stat changes, and stuff. I don't know why you think Celts should have this giant charge bonus, though - that seems much more based on stereotype and not really based on reality/historical accounts.

Here is the secret edu and cost formula of EB 2:
A: values 1 through 4
- 1 = levy
- 2 = regular infantryman and light cavalry
- 3 = elite infantryman and regular cavalry
- 4 = elite cavalry

B:Addition of armor values depending on unit
- helmet +1
- 1 greave +0.5
- 2 greaves +1
- soft armour +0.5
- scale shirt, and linothorax +1
- mail +1.5
- breatplate +2
- cheires + 1
- small shield +0.5
- large shield +1.0

C: weapons
- sword +1
- spear +1 (including kontos, sarissa, xyston)
- dagger +0.25
- longsword, kopis +1.5
- sling +0.1
- simple bow +0.25
- recurve bow +0.5
- javelins +0.5
- axe +0.75
- club +0.5

D: 2 through 3 based on green through elite
2 = levy/green and skirmishers
2.25 = militia-type units, drafted regionals, or semi-professionals
2.5 = regular
2.75 = veteran
3 = elite

F: Number of men in unit on normal setting

H: horse type
Pony = 6
Horse = 7
Nisaean = 8
1/2 barding = +2
3/4 barding = +3
full barding = +4

Equation:
cost = F[D(A+B+C+H)]

Considering that I worked for EB2, I might be able to get access to it but I haven't been an active member for a while. Regardless, the system to my recollection did not give great balance, but I haven't played the actual released game yet.

A message from Ibrahim about accuracy:
It's a measure in radians of the scatter of missiles round a target. You should let him know that, since I'm sure it will help him as it helped me

Right, but radians of what circle?

gamegeek2 wrote:I mean definitive cost suggestions, stat changes, and stuff. I don't know why you think Celts should have this giant charge bonus, though - that seems much more based on stereotype and not really based on reality/historical accounts.

of course, it was an example

ok, then here's a couple of suggestions
they're not directly related to a stat-defining alghorithm, but maybe someone can find them useful.

-Attack with short weapons (ie. from knives to gladii) should be way more fast than bulkier weapons

-Infantry units without shield and/or armor were made like this because they should effectiverly run a lot on the battlefield (archers, slingers, but even some shock infantry), therefore we should make them swiftier or at least give them the "hardy" / "very hardy" attribute

-Militia:
"Civilized" nations (Rome, Pontus, Egypt, etc, maybe even Aedui/Arverni) effectively trained levies, +1 bonus on Morale on said units as effect of the standardized training.
Furthermore as most levies were standard peasant, we should consider of giving to some of them a "hyde" attribute; for example, militia levies of celts, germans, getai, kushans and the like can have the hide_forest attribute, while sarmatian maybe even indo-saka infantry militia and can have the hide_long_grass

-Cavalry-based nations: +1 bonus on Morale for elite cavalry units, most of them should be "impetuous"

-"Berserker" units: as EB1 had a few, probably even EBNOM have some; as berserkers, the way they are definite in EB, were fanatic warriors floaking into battle, but that should be honored, I think they should be cheaper on recruiting and maybe more costly on upkeeping (a standard deviation of -25% of training and +10% on upkeep should do).

-Sea-fearing units: some cultures had an historical distrust against sea and boats (nomads, but not only), theremore their factional units (and some local units) shouldn't have the sea_faring attributs

-Nomad nations:given the great commitment they had on cavalry, -1 on the cost/man on each and every factional cavalry unit. Also, the cheaper cost should coax the AI in recruiting cavalry instead of infantry (the EB1 staff did something like this, if I recall correctly)
-Nomad-founded nations: like above, -0.5 on the cost/man on each and every factional cavalry unit.

-Standard heat malus and ground bonus for factional units (based on the original factional area):
Romani, Romani Liberati, Sertorians (roman troops): +1 heat, - scrub, - sand, - forest, - snow
Sertorians (hispanic troops): - heat ,  - scrub, - sand, +1 forest, - snow
Aedui, Arvernians, Suessiones: +1 heat ,  - scrub, -1 sand, +1 forest, - snow
Dacia, Lugia, Suebi: -2 heat ,  - scrub, -1 sand, +1 forest, +1 snow
Sauromatae: +1 heat ,  +1 scrub, -1 sand, -1 forest, +1 snow
Pontus, Armenia: +1 heat ,  -1 scrub, -1 sand, - forest, +1 snow
Parthia, Parthia Rebels: - heat ,  +1 scrub, +1 sand, -1 forest, -1 snow
Numidians: -2 heat (min. 0) ,  +1 scrub, +2 sand, -1 forest, -2 snow
IndoSaka, Kushan Empire: - heat ,  - scrub, - sand, +1 forest, +1 snow
Ptolemaic Kingdom, Hashmoneans, Nabateans: -1 heat (min. 0) ,  - scrub, +1 sand, -1 forest, -1 snow

gamegeek2

The answer is a unit circle.

Say a missile scatters 10 degrees to the right or to the left, then you convert this to radians, which is to divide by 180 and multiply by pi.

Paedric and Ibrahim at the totalwar center worked this out IIRC back in 2013.

I don't think you know what I am saying...I know how math works. I'll just figure out other stuff.

gamegeek2 wrote:I don't think you know what I am saying...I know how math works. I'll just figure out other stuff.

Well i don't know math, the answer was from Ibrahim.

I'm going to start figuring out how to program a spreadsheet to do all the cost calculations for me. This will have the very large advantage of making everything neat and easy to understand for everyone else, too!

UPDATE: This has been very successful so far. Expect a fully functioning, easy to use unit cost generator very shortly!

UPDATE 2: From now on, I'm going to program all constants into the program. I can keep them documented there for convenience and transparency's sake. Give me feedback as you wish.

I've successfully costed every single unit in the new spreadsheet! It seems we only have 203 units in total in the EBNOM EDU so far, though this does not count mercenaries that I am sure will be added at a later date. I'm not sure of the number of unique models this represents, but I think we can fit in a reasonable number of more units to this mix to represent various regional units, etc.

For example, we actually don't have any Yuezhi horse archer units in the game yet. After doing a bit of reading on the Kushans, it seems that their early army comprised a substantial body of horse archers, so just having the riders unit seems totally insufficient. The mercenary version of this unit's model could be the horse archer unit for the Wusun, who we might be able to make a relevant political entity via rebels, etc. to introduce a more interesting element to the eastern regions of the game.