Total War only has a handful of combat stats. Attack, Defense, Morale, and one or two others. Victoria 2 also has a handful; Attack, Defense, Discipline, Morale, Movement, and Teamwork (or something). All these stats are also broad and generalized while addressing factors other than pure equipment; they measure skill and mood. But they mimic combat far more accurately and satisfyingly than Elemental at the moment. The point I'm trying to make is complexity can build from a few simple stats; that it's better to have a few generalized stats rather than trying to account for every little detail. Having a multitude of stats is one extreme. Having too few stats is the other extreme; in this case, with only a few ways to influence the system, the system can be uninteresting and prone to mood swings, as it were.

I also did not talk about the way stats are utilized. 

My point wasn't that stats and abilities were wrong. My point was that simply adding them does not magically make things better. My point was to avoid adding in stats that are intuitive but also add too much complexity and obtuseness.

I'm basically saying exactly what you seem to think I'm not - a fewer, broader stats.

Actually, Total War does have a few more stats. In addition to different attack and defence values based on type of attack and type of armour (which Elemental also has), the most important difference I think is the skill stat. Instead of simply doing (weapon attack - armour defence), the calculation starts with (attacker skill - defender skill). The result of this generates the probability of a hit or a miss, not how sharp the sword is or how think the armour is. If it's a hit, then the (attack - defence) calc is done for damage with, I'm assuming, standard normal probability. When units gain experience (level up), their skill goes up. It's more complicated than this - they use rather complicated calculations which result in more realistic battles and damage - however the player never worries about the details; you can tell by the way the 3d models are going about their combat how the calculations are being done.

Doing it this way means that simply giving a mace to a peasant doesn't mean he'll beat an unarmoured veteran foot soldier with short sword. It's no matter that a good blow with a mace would probably kill an unarmoured man, if his chance of laying the blow is only 10%.

The combat system should be simple. Look at famous games like Civilization, Alpha Centauri, Master of Magic even and Colonization. Give units a base value dependent on their type (and have LOTS of TYPES in the game) then add modifiers from equipment like +2 for magical weapons to +5 like in dnd. All other things being equal except for magic and ranger. Give bonuses for length of weapon, give first strike advantages even if on defense. Give some units counterattack abilities before they die if they suffered a death blow effect (meaning simultaneous strike). DnD combat mechanics have always been my favorite. I like THACO if anyone remembers that. Get the ATK and the DEF with all their modifiers and then just do the simple rule ATK minus DEF = Negative Hit Points with the DEF counterattack modifer causing damage to the attacker as well. Combat encounters should be simultaneous at the same time for both sides. Not side A attacks side B and side B takes this much damage. Then side B attacks and side A takes this much damage. That kind of combat has never been fair to the defensive player. Combats must be/remain simultaneous.

The ai must also be improved not to advance when on defense without surveying the enemy and its missle power.

total war has charge bonuses, units with both melee and ranged attack, separate defences versus ranged and melee and quite a few others (some of which are hidden). it also does not have magic to deal with. it also gets to average out results over hundreds of combatants, so individual odd results go unnoticed. nor does it have to represent heroes, which arguably requires a much more developed, d&d style stat system to make fair and interesting.

elemental needs more stats. you can't represent a war of magic with 4 stats.

i agree with the analisys but imo its not a problem of combat speed

imo its a problem in overall balance

like max dmg being better than fast and low dmg

ofc  you 1 hit enemies that cant  counterattack

it was the same problem as the first combat system

imo when we see balanced weapons and RELIABLE not random armor combat speed MAYBE works

so i wouldnt shut down combat speed but first try to balance dmg and spells

when you cant 1 hit enemies and their armor actually defend them attack 3 times and getting 3 counterattacks is not going to be so  great like it is now where you can give opponents 3 times 123232313 dmg with no CA

As (many) others have said, the issue is more than damage distributions.  Certainly, that needs to be fixed.  More important, however, is the need for more complexity and useful options on the battlefield and, therefore, back to unit design.  This does not need to become a Total War game, but you do need to include most or all of the following:

1) Damage types and resistances (even it it's just physical, elemental and spirit)

2) Flanking effects on morale

3) Troop Placement (or make the default placement clever enough to put the bow-users and mages behind the infantry)

4) Tactical spells with a greater variety of effects, particularly effects that cannot be achieved without magic

Finally, as Frogboy has mentioned, integration of various systems is critical.  You can't call either magic or tactical combat finished until both of them are.

Because we're playing against AI you need to throw balance out the window for the most part when it concerns the ai. If the game is balanced the ai will never be worth playing against. Those crying for balance are usually multiplayers and you've already stated you wouldn't sacrifice the single player experience for the multiplayer one. Balance has always been an issue in most all PC strategy games and it is what ruins most of them that follow that lie that the game must be balanced.

What favors the human must be given to the AI without cost. The Civilization series does this the best by difficulty levels. It's why it's been such a success over the years and the reason MOM is still a success after all these years neither of the games are balanced. Most everything favors the ai advantage as well it should. It's what makes these games FUN & CHALLENGING.

The Warlords series is something you might look into as well for combat mechanics. They are simple yet fun and they don't have to have all that complicated bs that the AI can't handle anyway.

This is why you should just take examples from the MASTER GAMES an implement them and throw balance out the window. Don't worry about the multiplayer element of balance unless you can code it separately from the main game.

I've no doubt that most of what I say here will have been previously mentioned by other posters, but in the effort to give feedback/share thoughts I'm going ahead anyway.

Well, as previously stated, splitting the D will result in a bell curve set of results (2d6 resulting in more 7's for example) and this would serve to take a degree of unpredictability out of the combat while retaining an exiting level of randomness.  Whatever system is eventually implemented, losing too much unpredictability could make combat stale. 

The main thing though, is if combat is to be revamped, then at the same time so do the equipment lists.

Someone already mentioned there simply aren't enough missle weapons, Crossbows could serve as a higher tier weapon or perhaps a slower firing alternate throughout all tech levels.  I do realise that missle weapons could be exceptionally imbalancing to the system if too powerful, so perhaps some kind of range penalty, to stop an army of archers sniping the enemies to death could be implemented.  Preferably in the form of damage output rather than accuracy, so back rank archers would still be capable of providing support rather than a series of lucky blows wiping out the occasional unit.  Also, shields could be tweaked to offer a greater degree of protection against archery, giving a value to the lower damage, single handed weapon types, aswell as the opportunity to create units dedicated to resisting and countering archers.

Differences between weapon types would need to be part of this system aswell.  Not just the damage type of blunt/slash/pierce, but the speed factors being retained/refined and perhaps some of the qualities mentioned by other posters in regards to first strike opportunities.  In the regard of first strike, giving the quality to polearms and having it negated by facing a polearm (or enchanted weapon with the same ability/hero with same ability) would be a sufficient enough representation.

Smoothing out the numbers in weapon tiers is a pressing need, there is a vast degree of disparity between weapon types and the levels at which they appear, also some steps taken to eliminate redundancy in weapons developed could be used.  Perhaps ensuring that no tech development includes two weapons designed for the same purpose with little difference in cost or effect could work.  The most obvious examples of this problem appear throughout the empire tech tree. Trog scimitar/two handed scimitar and the shortbow/longbow leaping to mind. Differences in material cost aren't enough to warrent the inclusion of both.

Finally, I'd say there is a great deal of merit in the suggestions been given regarding the development of heroes.  I'd suggest a system of specialisation similar in concept to that of the city levelling system, though spread out a little.  Allow a choice from a list of abilities if the hero gets to a certain level.

For example an adventurer reaches level 3 and can then select from the following list;

1) Increased movement. (+1 move to unit)    2) Increased vision. (+1 to sight)      3) Punising strike (Once per battle Improved attack)

At level 9 a second tier and at level 15 a third and final tier could be introduced with improved versions of previous abilities or more powerful abilities appearing for the first time (unique summonations for example), perhaps even some of the soveriegn abilities at the final tier (organisation being a helpful one to have).  This could allow for some thematic types of heroes aswell, riffing on the traditional themes of paladin, ranger and warleader, aswell as the more Elemental specific ones such as Fallen, Channeler and Scion.

It goes without saying that each type of hero could have a different kind of list, and would give some incentive to get those merchants, loremasters and researchers out from behind the walls and in the field doing some research or building some trade contacts (with guards of course). 

Now, this would place a lower value on the farmer type of hero/administrator and they would either have to develop in a different fashion, or perhaps have abilities that sync up with the level of the city they manage.  Ideally, the administrator types (production, farming, mining) would be provided with a different system for xp and development based upon the progress made within the city they manage or time served.  One thing for sure, they shouldn't be the handled in the same fashion as heroes.  Beating people up doesn't make you better at farming... well, unless you use the corpses as mulch.

Anyway.  Thanks for taking the time to read this those who did.

I'm a balance nut but my definition of balance is not necessarily yours. =P

I want the systems in a game balanced. Troops vs Magic. Fire vs Earth vs Life. Kingdom vs Empire. Diplomacy vs Warfare vs Economy.
And that does absolutely not mean making each school of magic a carbon copy of any other. Much to the contrary.

That a difficulty level is called "normal" doesn't mean that I expect the AI to play as well as a "normal" human player.
Yes, the GalCiv AI is awesome but compared to WOM's complexity the game isn't even on the same order of magnitude.

I'd be perfectly happy with a "balanced" AI that balances some intelligence with blatant cheating. =P
Just don't make it as obvious, please. No -312817 Gildar treasury in the diplomacy screen...

Expecting a perfect AI is a nice dream and it would be great if it happened but I doubt that Brad would promise that...

Some thoughts:

1.  Make the tactical battle maps larger - 2x2 should do it.  That actually gives some meaning to range and maneuver.

2.  Give all spells a range limit in conjunction with the above.

2a.  Add some higher level bow/missile options.

3.  Give the AI some intelligence! 

• Don't leave high ground to attack when on the defensive.  The AI ALWAYS moves to contact.  It should take up a good position and stay there if on defense.
• Use high ground when possible on the attack.
• Exhaust spell/missile options prior to moving to attack unless facing spell/missile units.
• Attempt a retreat if possible when defeat is likely.  I know the AI retreats, but only when OVERWHELMLY out-numbered.
• Think about "the big picture", not just one unit at a time.  The AI seems to base decisions on the current unit, not the overall picture.

4.  Stuff that Rune_74 said:

5.  ALMOST FORGOT:  Defender should move first!!!  Not the attacker.

I still stand by my first response, but I'd like to throw an additional idea out there that's kind of "one step up." The new idea is based on the following idea: The point of the game isn't to realistically simulate fantasy combat. The point is to have a fun game where the mechanics encourage the players' interactions to develop in a way that could be related in story form as a fantasy combat. Wow - karma for anyone who can state that more clearly. But anyway:

Reactionary Combat.
The goal of this design is to keep both players engaged more continuously in the combat with interesting decisions to make both offensively and defensively. I want the highly interactive feel of Chess or Go (as mentioned earlier).

CRUNCHY DESCRIPTION
Each unit has two types of tactical actions: Actions and Reactions. An Action provokes a Reaction from one unit in range.

- Ability Type tagslabel the abilities of a unit in a meaningful way. Each unit ability can have multiple tags. Examples would include: Melee, Ranged, Attack, Defensive, Spell, Counterspell, Fire, etc.

- Active abilities (Actions) are what they have right now. I cast a healing spell, I attack that unit, I move around.
    * Actions can have a "First Strike" modifier. When marked "First Strike," the Action is resolved before the Reaction.

- Reactive abilities (Reactions) are abilities a unit can use when provokedby an opponent unit's Action. Currently, all units in the game have the (automatically executed) Reaction "Counterattack." In this model, however, additional Reactions are possible. 
    * Just like Actions, Reactions have ranges, target types, etc. Reactions use Ability Type tags (above) to determine which Actions types provoke them. 
    * A Reaction is one of these types:
        a. Defend. Occurs when a friendly unit is the target of a provoking ability. Your reaction targets the ally unit.
        b. Counter. You react when a friendly unit is the target of a provoking ability. This reaction targets the provoking unit.
    * Reactions may be flagged "Personal." A Personal Reaction may only be used when the provoking Action targets this unit.
    * Reactions may be flagged "Ally Only." An Ally Only Reaction may only be used when this unit is not the target of the provoking Action
    * Reactions, like Actions, may have a "First Strike" modifier. If a Reaction is marked "First Strike" it will be resolved before the provoking Action.
    * Some unit abilities may be used as both Actions and Reactions (internally, it may be best to code them as two abilities, though)

New combat sequence. On your turn, do the following:
   1. YOU: Pick a unit that hasn't acted this turn.
   2. YOU: The chosen unit may move and "declare" one Action.
   3. OPPONENT: When the Action is declared, all units that have Reactions provoked by the Action who are within range to use their Reaction ability on a legal target (Personal, Assist, Defend, etc) are highlighted. Choose ONE of those unit's Reactions. If only one unit has a valid reaction (i.e. only the attacked unit can counterattack), that Reaction is selected automatically.
   4. COMBAT ENGINE: Resolve the Action/Reaction in this order: First Strike Reactions AND First Strike Actions, Actions AND reactions.
   5. YOU: Repeat while you have units that haven't acted. If there are none remaining, it's the opponent's turn.


EXAMPLES AND NARRATIVES

EXAMPLE 1: Reaction: Counterattack. Range: 1 tile. Provoked by:MeleeAttack. Type: Counter (Personal). Effect: You attack the unit that's attacking you.
    This is what happens now. They attack your unit in melee, and your unit counterattacks. Note that the personal modifier keeps this ability from being used when a nearby ally is attacked. As opposed to...

EXAMPLE 2: Reaction: Harrying Fire. Range: 2 tiles. Provoked by:MeleeAttack. Type: Counter (Ally Only), First Strike. effect: Target's Attack score is reduced for this action.
    If you have a group of archers directly behind your swordsmen, you could choose this Reaction (instead of letting the Swordsmen counterattack) to reduce the damage done to your swordsmen as they are attacked. a variant of this reaction might also do minor damage. The First Strike setting makes sure this takes effect before the attacker's damage is calculated.

EXAMPLE 3: Reaction: Counterspell. Range: Unlimited, Provoked by: Spell, Type: Counter, First Strike. Effect: Target's spellcasting effects are reduced for remainder of the turn based on relative INT scores (or something). Mana cost: X
   Anytime an enemy unit used an Action tagged "Spell," this unit could use this ability. Since it's "First Strike," you would spend mana to reduce the opponent's spell effectiveness.   

EXAMPLE 4: Reaction: Moment of Glory (a spell). Range: Unlimited. Provoked by:MeleeAttack, RangedAttack. Type: Defense, First Strike. Effect: Target's Defense is set "very high" until end of turn.
    Cast this spell as a Reaction to keep a unit alive when they're being attacked (instead of letting the unit counterattack)

EXAMPLE 5: Reaction: Blink (a spell). Range: 0. Provoked by: MeleeAttack, RangedAttack, Spell. Type: Defense (Personal), First Strike. Effect: This unit is moved to a random square within X squares. 
    Cast this spell as a Reaction to an attack or spell to move randomly to a nearby tile before the attack can take place (short circuiting it)

With those examples, I hope it's clear how cool and strategic this system could be. With a decent user interface, it wouldn't play out nearly as complex as this might sound from the written description.

Gnilbert

I strongly echo the sentiments regarding more specialised units and special abilities. So far, apart from being cosmetically different, the created units do not feel unique.

I would like to see units be able to receive active and passive abilities. These abilities would be unlocked by researching the appropriate techs, and would be tied to the Warfare/Conquest and Adventure/Domination trees.

For example, when researching the first level of Adventure/Domination, a "Frontiersman" ability can be unlocked. This would be a passive ability which gives the unit with it a 25% bonus against creatures. This would encourage more early exploration through the creation of military units, since they would be less fragile against random encounters in the wild.

The player can train a unit to acquire these active or passive abilities. Training an ability will require a monetary or resource cost, and active abilities will have a mana (or "energy") cost.

These abilities can be learnt by the Sovereign when unlocked. Active abilities would also require mana to use. However, the number of abilities the Sovereign can have will depend on his/her level (e.g. 3 at Levels 1-10, 4 at Levels 11-20 etc). This would allow players to create martially-oriented Sovereigns with character, without making them overpowered.

These are just some of the abilities I would like to see:

Mounted Combat -- Received upon unlocking "Mounted Warfare", this would give the trained unit a bonus to movement speed on the map and in battle. However, they would be weak against units with the Cavalry Killer ability.

Cavalry Killer -- Units with this ability deal 50% more damage against mounted units.

First Strike -- Unit with this ability always attacks first, even when being attacked.

Far Shooter -- Unit with this ability deals more damage the further the target is from it.

Shield Breaker -- Active ability, this enables the trained unit to reduce the defence of target enemy unit by 1/2 of its attack score. Can stack.

Shield Other -- Active ability, this enables the trained unit to take half of the damage dealt to target ally unit.

Apart from passive and active abilities, I would also like an overhaul of the tactical magic spells to make each school unique. Right now, most of the spells seem interchangeable. There is just no reason for me to learn a particular spell from the Water school when there are other spells in the other schools that do the same thing.

I suggest making each magic school unique. Fire spells would emphasise doing large amounts of ranged damage, while Earth spells would focus on defence and close quarter damage. Water spells would be geared towards healing and crowd control, while Air spells would aim at affecting the battlefield. Enchantment spells are your basic buff/debuff.