Unit Combat Rating

[ShadeHonestus]

Amongst the other new features at the assembly I have been working around the idea of a Unit Combat Rating as it is dubbed in my own thoughts at the moment.

Here is the jest of it:
I want to come up with an equation that would effectively represent a units Overall Strength on the battlefield and create a scale for such a rating.

This is just a thought at the moment and would like to have some serious input.
Of course I would also like to hear if you think this is even a worthwhile task or not.

Thanks...

[Dionysus9]

Thanks for the Compare tool Honestus!!!

I've been thinking hard about a Combat Rating Equation, and here are my conclusions:

1) We must attempt to quantify the value of morale/valor by experiments.

2) We must attempt to quantify the value of the number of men in a unit, by experiments.

3) We must find a way to express Charge bonuses, Armor, and other intangible benefits like fighting in ranks of 2-3, bonuses vs. Cav, etc.
(4) We must also find a way to quantify the value of being able to fire ammunition. This is VERY difficult because you never know what speed/armor the enemy will have.

The above four issues seem to be the biggest barrier to creating a viable unit ranking system.

Melee/Defense values are self explanatory and seem like they can simply be added together. +2 melee, -1 defense = +1. All other things being equal, it would seem that a +1 (overall) unit would generally tie another +1 (overall). Some testing is needed.

So far I have been using an equation like this:

(Melee + Defense) + 1/2(Charge+Armor+[valor+morale]) = rating

Melee and defense are almost always applicable (unless getting shot at), Charge only comes into play for part of the time, as does armor. Morale/Valor only applies if you are losing or being flanked, which is maybe 1/2 the time?

But the above equation is just a guess, and I havent experimented with Custom Battles to see if it holds true.

Massive experimentation is required I think.

Thanks again for the Tool Honestus...its awesome.

[Cheetah]

First of all, thx for the tool Honestus. It is very nice. About your idea: IMHO there is no such thing as Overall Strength because of the Rock-Paper-Scissor relation of the units. What is important -and missing from the tool http://www.totalwar.org/ubb/smile.gif is the cost of units. What you need to know is that which is the best anti-cav unit for your money, which is the best meele unit for your money, etc.

[ShadeHonestus]

The unit cost is found when you click on a unit, and is not yet in the front end and sortable. I will most likely replace the current "morale" field with "cost" as it seems the previous isn't pertinent.

I was thinking of graphing unit costs as well...

As to the Combat Strength Rating. I was thinking in terms of dual ratings. Although there is already an attack and defense bonus represented, I was thinking of an inclusive Attack and Defend Rating that would incorporate the other pertinent rankings such as melee bonus, armor, honor etc...

Just a thought and its is probably not all that useful other than quickly sorting units for comparison.

Thanks all...

[The Black Ship]

Perhaps the biggest "problem" with MTW is that there isn't a gold standard for comparison, as in Shogun. All you'd have to say in Shogun was..."compared to a WM" for foot sloggers, or "better than a YC" for horses, and everyone knew the pertinent value of the unit.

[Puzz3D]

I think a combat rating comparision between two units in a head-to-head encounter might be useful if you can see the effect of upgrades. That will give a good estimate of what beats what and at what cost. During unit selection, you'll be able to make intelligent cost efficient unit choices. During a battle, if you are going to win a straight up encounter, then you can turn your attention to some other less favorable matchup which needs an additional unit sent in to help. Morale is a factor in the cost, but you are going to upgrade valor on low morale units anyway, so that they are useful in battle. In the end, there probably isn't a big variation in morale among the units that people are typically fielding, and you can probably leave it out. I don't see how you can put an accurate cost estimate on morale anyway since it's a threshold effect rather than a proportional effect like all the other combat factors.

You need attack, defend, shield, armor piecing bonus, anti-cav bonus and charge. Based on a lot of 1v1 tests in WE/MI, my own estimate of the value of charge is 5 charge is equvalent to about +1 melee. Of course, you have to actually charge to get the charge benefit. You could add all these numbers for each of the two units and take the difference. Every point of difference is a compounded 20% difference in the chance to win. You would also want to display each unit's upgraded cost to see how much that chance to win or loose is costing.

Oh, I forgot you need the rank bonus for spears fighting in depth.

[This message has been edited by Puzz3D (edited 09-12-2002).]

[Dionysus9]

Quote Originally posted by Puzz3D:
Based on a lot of 1v1 tests in WE/MI, my own estimate of the value of charge is 5 charge is equvalent to about +1 melee. Of course, you have to actually charge to get the charge benefit. You could add all these numbers for each of the two units and take the difference. Every point of difference is a compounded 20% difference in the chance to win. You would also want to display each unit's upgraded cost to see how much that chance to win or loose is costing. [/QUOTE]
^^^^^^^^^^^^^^^^^^
This is exactly the type of quantification I am talking about. And let me say, Cheetah, that I understand what you are saying, but there is a way to work around it-- I do not throw my archers at cavalry, and I dont expect my archers to get attacked by cavalry (if they do, I've made a mistake). So, all I am really interested in with regard to a Combat Rating is Scissors v. Scissors. If I take Scissors v. Paper, I'm going to win. Likewise, if I take Scissors v. Rock, I'm going to lose. I know that. What I don't know is what happens if I take Scissors v. Scissors? Trebizond Archers v. Bulg. Brigands? That is where the uncertainty exists, and that is where the Combat Rating can come in handy.

Yuuki and Honestus-- You guys seem to agree that the "morale field is not pertinent", but has anyone done any experimentation to see if it is or not?

I propose an experiment! And a quick review of the Custom Tool provides the perfect subjects:

A. Peasants vs. Muslim Peasants:

These two units have identical stats, with 2 exceptions:
(1) Peasants have -2 morale, Muslim Peasants have 0 morale.
(2) Peasants have armor piercing pitchforks and Muslim Peasants dont.

This makes them good for testing how "pertinent" the morale figure may or may not be. Of course, the test is not "perfect" unless the peasant's armor piercing ability is modded out (since both units DO have 1 armor, Peasants have a slight attack advantage, not due to morale).

So, if we mod Peasants to have No armor piercing bonus, and then we pit the two units against eachother a bunch of times....we can start to see a trend-- and perhaps quantify the effects of the "morale" field.

Or, we can conduct a similar experiment without modding the peasants armor piercing capabilities. Simply stand some peasants at a specified distance from some Longbows and see when they run (as compared to Mulsim Peasants). After 100 trials, or so, a trend should be relatively clear.

This should be done in Multiplayer with two computers, because if you try to perform experiments against Custom Battle AI, you get results that vary based on crazy/silly AI movements. Both units should behave consistently, over many many trials. The only way to guarantee that is in Multiplayer.

Soooooo..... my hypothesis is that, on the average, normal Peasants will run slightly before Muslim Peasants. This makes them slightly less useful (ignoring armor piercing for now), and this "less useful" effect should be capable of quantification.

If we can perform such experiments for every factor (i.e. armor-piercing, attack cav bonus, blah blah blah), then we can BEGIN to formulate a workable Combat Rating.

I think it can be done, and can be useful. The problem is I only have 1 computer, and most people dont want to stand a peasant against longbows 100 times...

So its hard for me to run these tests. Maybe I will see you guys online and we can try a few.

Goood discussion!!!

-----

Also, I've noticed two errors in the Compare Tool (I'm not criticising, but only trying to help http://www.totalwar.org/ubb/wink.gif)-- Armenian Heavy Cav is listed in Infantry (maybe they can dismount?) and Sipahi of the Porte are listed as having 28 ammo (which they dont have).

[Puzz3D]

Dionysus9,

The morale "field" in that table has nothing to do with combat. I'm refering to "morale" which is in the honor column. Morale doesn't matter if you are "winning". If you are loosing, it also doesn't matter except it determines how long the unit will stand and get pumelled before it routs. What I really want to know is how well will my front line do against the enemy's front line. We'll both have very similar units in the line. If you are going to loose the front line battle, then you had better get something behind the enemy and hit the back of his line as soon as possible. Even with this information it's far from all the things you have to account for in a battle to be successful.

[Dionysus9]

Ummm....im confused. I guess I don't understand what the "morale" field in the Compare Tool signifies. Surely it signifies something relevant to combat? I can't imagine it is meaningless or applies only to SP. I think the morale field is a bonus modifier for morale checks....added to valor.

You of all people know that there is value to quantifying the effects of morale, your contributions to the Academy are proof of that.

My 'experiment' is just an example of how to quantify these different variables--I hope others will be interested in testing out all these variables and trying to pound out a "scissors v. scissors" type Combat Rating.

If we get enough reliable people out there testing these variables and sharing results, we will quickly narrow down each part of the equation. We just need to test scissors v. scissors. I am mostly interested in ranged troops v. ranged troops, and then infantry v. infantry, and the side battle of cav v. cav.

The morale system is an important frontier, as the cav bonuses are probably just straight modifiers to attack/defense.

Morale has a more subtle effect on combat, but its as important knowing you will win a matchup. I'm thinking of morale as an ending point for the matchup between two units. Routing archers are worthless, even if you have more than your enemy.

There must be an intrinsic quantifiable florin value to archers that have 2 honor +2 morale bonus, as opposed to 1 honor +1....etc. Ideally we can creat an equation that accounts for the morale/valor of a unit.

[Dionysus9]

Honestus,

I've been thinking, we should start our hunt for the UCR by concentrating on three (or four) values.

UCR v. missle troops
[for non-ranged this would be primarily armor based]
[for ranged units this would combine armor and the quality of their ranged weapon]

UCR v. infantry
[attack+defense, +1 for every 5 charge, an allowance for armor piercing, etc.]

UCR v. Cavalry
[Cav bonuses, silimar to infantry, speed?]

Then once we have solid figures for those basic components, we can try to assimilate the results for a Ye Grande Olde UCR.

I'm convinced it can be done at least well enough to be of some use, even if not perfect or adapted to every situation. You have all the variables there in the tool....

We just need experimental results and Yuuki's brains...

[Puzz3D]

Dionysus9,

One of the CA programmers posted that the morale cloumn was something used by the campaign game, and he wasn't sure if it was still actually used.

[tootee]

Fish say stats are just numbers. A sensible general will bring sensible units to the battlefield. I'm not gonna worry too much about the chances of my V4 AUM vs opponent V4 Viking. They are close and their meeting outcome will be close.

It is more important to make they dance like how Magy makes his cav dance around the foolish enemies who are busy figuring out numbers and equations.

Bring 4 missiles. Bring 4 cav. Bring 4 spears. Bring 4 shock troops. Make cav, spears and shock almost equal in morale. Gah! if u sensible enof the entire front will have a balance combat strength (melee+defence). Then its back to what is more important, controlling the armies to the state where ya scissors will hit the enemy paper, ya stone hit their scissors while ya paper on their stone! Even if stone hit stone, its not gonna be decisive, so the rest of army can still dance around while the enemies fool around with numbers http://www.totalwar.org/ubb/biggrin.gif

However, if I have time I will be interested in the tool too http://www.totalwar.org/ubb/biggrin.gif. So good work!

------------------
tootee the goldfish,
headmaster of Shogun-Academy (http://shogun-academy.tripod.com)
loyal roach of Clan S.G. (http://thesilvergazwa.tripod.com)
'Pa Si Buay Chao! Si Liao Ka Song!'
------------------

[Arkatreides]

See here for official morale/honour reply:

http://www.totalwar.org/ubb/Forum7/HTML/001435.html

[Dionysus9]

Thanks for the answer Yuuki and Arka.

I did some experiments with Muslim Peasants and Peasants vs. Byz. Inf., and after 10 trials each, they averaged the same breaking point -- they normally ran at 67 men remainining.

This seems to confirm that the "morale" field is in fact useless (which boggles my mind to some extent). I suppose Honestus can replace that column with "pref rows"...

and it looks like nobody is really interested in the UCR but Honestus and myself.... oh well...

[ShadeHonestus]

I'll probably bring unit cost to the master list and leave pref rows in the more detailed unit data, seen when the unit is clicked.

[Cheetah]

PAF