You must be logged in to post messages.
Please login or register

Strategy Discussion

Hop to:    
loginhomeregisterhelprules
Bottom
Topic Subject: Unit Rating System - Lots of Math
posted 21 May 2003 06:21 PM EDT (US)   
OK, I've been thinking for a while about how to evenly rate units to account for all factors: HP, attack, armor, speed, range, cost. I think that I've come up with a reasonable system, but before I rate all the units, I thought that I would get people's input on how the system is weighted right now. Here's what I did:

1. Take pierce and hack armor, average them. (Crush armor is, for the most part, insignificant).

2. Next, multiply the HP by the armor average/100 . Add the HP to this number.

3. Take hack attack*1.1 , ranged attack and crush attack. (1.1 hack accounts for buildings hack vulnerability. Crush is also useful vs buildings, but is usually very high anyways.) Add 1/3*range. basically:
1.1*hack + ranged + crush + 1/3*range

4. Take the attack total, multiply it by the HP total. Multiply this by 1/4 of the unit speed.

At the end of step 4, we have calculated the Unit Effectiveness, or basically how powerful the unit is in fights.

5. Now take 1.5*gold cost, .75*wood cost, and 2*favor cost and the food cost.

6. Multiply the total cost by 1/2 times the population
This is the Cost Total.

7. Divide the is Effectiveness by the Cost Total.

So, heres an example: which is better - hoplite or ulfsark?

I'll do ulfsark out step by step:

1. 30+5=35/2=17.5

2. 80*.175=14+80=94

3. 1.1*9=9.9

4. 94*9.9=930.6*.25=232.65*4.8=1116.72

5. 50+1.5*30=45+50

6. 95*.5*2=95

7. 1116.72/95=11.75

Hoplite:

Unit Effectiveness - 1207.5
Cost Total - 112.5
Total Effectiveness=10.73
The Ulfsark seems more effective overall. (although the hoplite is more powerful, it is also more effective) Just for kicks, I calculated the Hersir effectiveness - 7.75

Obviously, this isn't perfect, it fails to take into account unit bonuses, and special attacks. It also assigns arbitrary numbers to resource value and hack/range/crush attacks. What do you think? Should I change any of the coefficients before computing the rest of the units? (I've already written a computer program that calculates the effectiveness using this method, so it isn't too much work.) Thanks in advance for any comments/advice.

EDIT: changed favor variable down a bit

[This message has been edited by Jackson (edited 05-21-2003 @ 06:29 PM).]

Replies:
posted 21 May 2003 07:38 PM EDT (US)     1 / 10  
OP said:

"OK, I've been thinking for a while about how to evenly rate units to account for all factors: HP, attack, armor, speed, range, cost."

lol - what about bonuses?

(even if you answer that, the point obviously generalizes....)

cdj

posted 21 May 2003 07:59 PM EDT (US)     2 / 10  
The problem with all these systems is that they generalize too much, especially as far as counters are concerned. One unit might have a lower rating than a unit it counters, for example.

A unit which has a high effectiveness rating in terms of cost & such, might do pierce damage, and if your opponent is using pierce-resistant units you're kinda screwed.

I think they're probably useful to some degree for comparing the same type of units between civilizations (ie: is a scarab a better siege than a ballista? Is a ballista a better siege than a catapult? etc). But when you extend it to comparing different types of units, it just gets messy.

posted 21 May 2003 08:06 PM EDT (US)     3 / 10  
why dont u go work on something like ur calculus homework instead of showing us ur math skills for a stupid game
posted 21 May 2003 09:10 PM EDT (US)     4 / 10  
*sigh* just ignore nrs mabey he wil shut up
posted 21 May 2003 09:20 PM EDT (US)     5 / 10  
ok, all that seems fair. I've been thinking that this really can't compare specific unit types to each other, but it is useful for determining which units are the best value for each civ, perhaps to find out who you would like to play if you like infantry, etc. I'm hoping of figuring out a way to incorporate bonuses...

and Mrs, I've already finished my calculus homework and I got bored.

posted 21 May 2003 09:34 PM EDT (US)     6 / 10  
no. go do more. theres nothing more pathetic than an RTS mathematician
posted 21 May 2003 09:57 PM EDT (US)     7 / 10  
Did you see Archepterix's post on calculating units effectiveness? It considers specific types of units vs other specific types and it doesn't use any arbitrary mutlipliers, it's all based on in-game stats.

Programmer on 0 A.D., author of Norse Wars, co-author of Fort Wars.
posted 21 May 2003 10:39 PM EDT (US)     8 / 10  
good for him, he was successful in completely wasting an afternoon
posted 23 May 2003 08:20 AM EDT (US)     9 / 10  
shut up killer!
posted 23 May 2003 11:00 PM EDT (US)     10 / 10  
I judge simply based on common sense and experience. I use testing occasionally, at most.
Age of Mythology Heaven » Forums » Strategy Discussion » Unit Rating System - Lots of Math
Top
You must be logged in to post messages.
Please login or register
Hop to:    
Age of Mythology Heaven | HeavenGames