Attack Rating Calculations

[trevius]

Tonight, I wrote a change for #showstats to try to make it show the correct attack rating for players. I used the exact same code that the source uses to calculate attack to factor into damage. Here is the calculation:

		if(attacker->IsClient())
			attackRating = attacker->GetATK() + ((attacker->GetSTR() + attacker->GetSkill(OFFENSE)) * 9 / 10);
		else
			attackRating = attacker->GetATK() + (attacker->GetSTR() * 9 / 10);

I found that this calculation is WAY off from what the character window shows.

After much testing of different values, I have found what I think to be very accurate numbers for the equation. We were also missing some major factors into it as well, such as skill for the item equiped.

Strength - 66 * 0.9
Attack from items * 1.342
skill in use (item type equiped in primary slot) * 2.69
offense skill * 1.342

I haven't tested this yet, but I think this may work to do the calculations above. I did the math manually a few times and it was extremely accurate within maybe 2 or 3 points off or so.

   if(this->IsClient())
      attackRating = (((GetATK() + GetSkill(OFFENSE)) * 1.342) + ((GetSTR() - 66) * .9) + (GetSkill(skillinuse) * 2.69));
   else
      attackRating = GetATK() + (GetSTR() * 9 / 10);

To test this, tomorrow night I will probably put this into #showstats like this:

void Mob::ShowStats(Client* client) {
   float attackRating = 0;
	if(this->IsClient()) {
		attackRating = (((GetATK() + GetSkill(OFFENSE)) * 1.342) + ((GetSTR() - 66) * .9) + (GetSkill(skillinuse) * 2.69));
		//Need to add a way to calculate the skill of the weapon in the primary slot instead of using skillinuse
		int32	AR_Cap = 0;  //Attack Rating Cap
		AR_Cap = (GetSkill(OFFENSE) + GetSkill(skillinuse) ) * 5;  //This isn't correct, but putting it here until I have the correct formula for caps
		if (attackRating > AR_Cap)
			attackRating = AR_Cap;

		if (attackRating < 10)
			attackRating = 10;
	}
	else
	attackRating = GetATK() + (GetSTR() * 9 / 10);

	client->Message(0, "Name: %s %s", GetName(), lastname);
	client->Message(0, "  Level: %i  MaxHP: %i  CurHP: %i  AC: %i  Class: %i", GetLevel(), GetMaxHP(), GetHP(), GetAC(), GetClass());
	client->Message(0, "  MaxMana: %i  CurMana: %i  ATK: %i  Size: %1.1f", GetMaxMana(), GetMana(), attackRating, GetSize());

If that proves to be accurate, we might want to consider altering all other parts of the code that rely on calculating attack properly.

***EDIT: Editing the code as I refine it to make it more accurate***

[KLS]

Attack rating isn't supposed to be what the character window shows, this is by design. Atk is composed of two parts. One that overwhelms mitigation and one that overwhelms avoidance. It would be cool for showstats but overall I'm pretty happy with the formula for the mitigation formula.

[trevius]

I am still working to refine the calculations to have them as accurate as possible. Currently, I still need to figure out the equation to calculate attack rating caps. I also need to write some code for it to check the skill value of the weapon type in the primary slot (using skillinuse isn't exactly how it works).

I will be pretty happy if we can just get #showstats working more accurately. But, I do think that this might be worth considering for actual mitigation calculation once the formula is all done and tested. If there was only a slight difference between the actual formula and the formula currently used in the source, then it wouldn't be a big deal. But, in some cases, the server is using half or less of what the character screen shows. It really has a huge range and doesn't scale the same as the calculations from the client at all. Plus, there currently doesn't seem to be anything that sets attack rating caps, which means that worn attack can be making a huge impact on actual mitigation. I also don't see where weapon skill is being used at all when calculating mitigation, and if that is true, you could have 1 in the skill or 300 in the skill and still hit for the same amounts (though you wouldn't land nearly as many hits). Let me show a couple of example equations.

Player A Stats:
100 STR
100 Offense Skill
100 Weapon Skill (using 1hs or whatever)
0 Worn/Spell attack

Player B Stats:
200 STR
200 Offense Skill
200 Weapon Skill
50 Worn/Spell attack

Player C Stats:
300 STR
300 Offense Skill
300 Weapon Skill
200 Worn/Spell attack

Player D Stats:
380 STR
252 Offense Skill
1 Weapon Skill (using a weapon they don't have skill in yet)
500 Worn/Spell attack

Player E Stats:
380 STR
252 Offense Skill
300 Weapon Skill
500 Worn/Spell attack

Current Formula:
Worn Attack + ((Strength + Offense) * 0.9) = Total Attack Rating

PlayerA: 0 + ((100 + 100) * 0.9) = 180
PlayerB: 50 + ((200 + 200) * 0.9) = 410
PlayerC: 200 + ((300 + 300) * 0.9) = 740
PlayerD: 500 + ((380 + 252) * 0.9) = 1068.8
PlayerE: 500 + ((380 + 252) * 0.9) = 1068.8

New Formula:
((Worn Attack + Offense) * 1.342) + ((Strength - 66) * 0.9) + (Weapon Skill * 2.69)

PlayerA: ((0 + 100) * 1.342) + ((100 - 66) * 0.9) + (100 * 2.69) = 433.8
PlayerB: ((50 + 200) * 1.342) + ((200 - 66) * 0.9) + (200 * 2.69) = 994.1
PlayerC: ((200 + 300) * 1.342) + ((300 - 66) * 0.9) + (300 * 2.69) = 1688.6
PlayerD: ((500 + 252) * 1.342) + ((380 - 66) * 0.9) + (1 * 2.69) = 1294.474
PlayerE: ((500 + 252) * 1.342) + ((380 - 66) * 0.9) + (300 * 2.69) = 2098.784

So, basically, the 2 are fairly close in scaling. For the current formula, Strength is scaling the same as the Offense skill, which isn't quite accurate since Offense should be adding more and scaling a bit faster than strength. Also, worn/spell attack is having basically double the effect that it should be, since weapon skill isn't being factored into the equation to make worn/spell attack less of a factor. Another problem is that a level 1 is starting off with way too much attack since the 66 isn't being subtracted to account for starting stats. So, a level 1 high elf wizard with 55 strength (lowest you can make) is starting out with 49.5 attack, when they should only have 10 (10 is the min value). Using the new formula, that wizard would have -9.9, which would cause them to be set to 10 attack until their attack was raised to be over 10.

In reality though, other than the lack of weapon skill check, and lack of attack caps, the current formula works fairly similar to the new one. But, I do think using the new one would make it a bit more accurate. I think it would be very simple to use the new one and have it output about the same mitigation as it does now. All we would need to do is divide the total by about 2 to 2.4 or so. That divider could be set as a rule to allow people to tune it to lower or higher end encounters/equipment. I also think that attack caps should be set as a rule to give people the option to use them or not.

Ultimately, the system works pretty well as it is. I am not complaining about that. But, if we are striving for accuracy and to emulate live as close as possible, I think this should at least be considered.

[trevius]

To get the skill value of the item being used in the primary spot, I am trying to write a function to handle it. I am not sure if this will work as it is, but might be worth a shot. Basically, it is supposed to get the item ID from the primary slot, and then determine the item type of that item. Finally, it checks the skill value of the player for the skill that item type uses.

I guess this could go in mob.cpp:

int Mob::GetPrimarySkillValue(SkillType &PrimarySkill, const ItemInst* Item_ID)
{
	int16 Item_ID = CastToClient()->GetItemIDAt(13);

	if (Item_ID = NULL)
		PrimarySkill = HAND_TO_HAND;

	else {
		const Item_Struct* item = Item_ID->GetItem();

		switch (item->ItemType)
		{
			case ItemType1HS: // 1H Slashing
			{
				PrimarySkill = _1H_SLASHING;
				break;
			}
			case ItemType2HS: // 2H Slashing
			{
				PrimarySkill = _2H_SLASHING;
				break;
			}
			case ItemTypePierce: // Piercing
			{
				PrimarySkill = PIERCING;
				break;
			}
			case ItemType1HB: // 1H Blunt
			{
				PrimarySkill = _1H_BLUNT;
				break;
			}
			case ItemType2HB: // 2H Blunt
			{
				PrimarySkill = _2H_BLUNT;
				break;
			}
			case ItemType2HPierce: // 2H Piercing
			{
				PrimarySkill = PIERCING;
				break;
			}
			case ItemTypeHand2Hand:
			{
				PrimarySkill = HAND_TO_HAND;
				break;
			}
			default:
			{
				PrimarySkill = NULL;
				break;
			}
		}
	}

	uint16 skill = GetSkill(PrimarySkill);

	return skill;
}

If anyone has any corrections or suggestions on how to do it better, I would be very happy to hear it. I am fairly certain that this will fail to even compile lol.

Once that function is working, all I will need to figure in is the attack rating cap formula, which should be fairly simple. And the new code for #showstats with this function would look like this:

void Mob::ShowStats(Client* client) {
	float attackRating = 0;

	if(this->IsClient()) {
		attackRating = (((GetATK() + GetSkill(OFFENSE)) * 1.342) + ((GetSTR() - 66) * .9) + (GetPrimarySkillValue() * 2.69));
		int32	AR_Cap = 0;  //Attack Rating Cap
		AR_Cap = (GetSkill(OFFENSE) + GetPrimarySkillValue() ) * 5;  //This isn't correct, but putting it here until I have the correct formula for caps
		if (attackRating > AR_Cap)
			attackRating = AR_Cap;

		if (attackRating < 10)
			attackRating = 10;
	}
	else
	attackRating = GetATK() + (GetSTR() * 9 / 10);

	client->Message(0, "Name: %s %s", GetName(), lastname);
	client->Message(0, "  Level: %i  MaxHP: %i  CurHP: %i  AC: %i  Class: %i", GetLevel(), GetMaxHP(), GetHP(), GetAC(), GetClass());
	client->Message(0, "  MaxMana: %i  CurMana: %i  ATK: %i  Size: %1.1f", GetMaxMana(), GetMana(), attackRating, GetSize());

If I get this attack rating calculation working smoothly, I may try to just make a function to handle the whole thing, so we could just use GetTotalATK() where needed instead of doing the same code over and over.

That code would be something like this:
EDIT: Change the code below for GetTotalATK() a bit.
client.cpp

uint16 Client::GetTotalATK()
{
	int16 AttackRating = 0;

	if(IsClient()) {
		AttackRating = (((GetATK() + GetSkill(OFFENSE)) * 1.342) + ((GetSTR() - 66) * 0.9) + (GetPrimarySkillValue() * 2.69));
		int16	AR_Cap = 0;  //Attack Rating Cap
		AR_Cap = (GetSkill(OFFENSE) + GetPrimarySkillValue()) * 10;  //This isn't correct, but putting it here until I have the correct formula for caps
		if (AttackRating > AR_Cap)
			attackRating = AR_Cap;

		if (AttackRating < 10)
			attackRating = 10;
	}
	else
	AttackRating = GetATK() + (GetSTR() * 9 / 10);

	return AttackRating;
}

Then, the #showstats would just need to have GetATK() replaced with GetTotalATK(), and it should show more accurate values. For the actual attack code to use this, it could just replace the current Attack Rating calculations in mitigation, and just do (GetTotalATK() / 2) instead to get it back to about the same as it is now. NPC DPS wouldn't be effected by this, just players. And by dividing it by 2, it should get them back to about the same as they are now.

Of course, again, if anyone has any corrections or suggestions on how to code this stuff properly, I would be glad to hear them. My coding skills are still really bad and there are certain things I don't understand well or at all yet. I mostly just work by examples I find elsewhere in the source.

[AndMetal]

Originally Posted by trevius

mob.cpp:

int Mob::GetPrimarySkillValue(SkillType &PrimarySkill, const ItemInst* Item_ID)
{
	int16 Item_ID = CastToClient()->GetItemIDAt(13);

	if (Item_ID = NULL)
		PrimarySkill = HAND_TO_HAND;

	else {
		const Item_Struct* item = Item_ID->GetItem();

		switch (item->ItemType)
		{
			case ItemType1HS: // 1H Slashing
			{
				PrimarySkill = _1H_SLASHING;
				break;
			}
			case ItemType2HS: // 2H Slashing
			{
				PrimarySkill = _2H_SLASHING;
				break;
			}
			case ItemTypePierce: // Piercing
			{
				PrimarySkill = PIERCING;
				break;
			}
			case ItemType1HB: // 1H Blunt
			{
				PrimarySkill = _1H_BLUNT;
				break;
			}
			case ItemType2HB: // 2H Blunt
			{
				PrimarySkill = _2H_BLUNT;
				break;
			}
			case ItemType2HPierce: // 2H Piercing
			{
				PrimarySkill = PIERCING;
				break;
			}
			case ItemTypeHand2Hand:
			{
				PrimarySkill = HAND_TO_HAND;
				break;
			}
			default:
			{
				PrimarySkill = NULL;
				break;
			}
		}
	}

	uint16 skill = GetSkill(PrimarySkill);

	return skill;
}

Per trev and my collaboration, corrections to the first function:

client.cpp:

uint16 Client::GetPrimarySkillValue()
{
	uint8 type = m_inv.GetItem(13)->GetItem()->ItemType;	//is this the best way to do this?
	SkillType skill = HIGHEST_SKILL + 1;	//because NULL == 0, which is 1H Slashing, & we want it to return 0 from GetSkill

	if (!type)
		skill = HAND_TO_HAND;

	else {

		switch (type)
		{
			case ItemType1HS: // 1H Slashing
			{
				skill = _1H_SLASHING;
				break;
			}
			case ItemType2HS: // 2H Slashing
			{
				skill = _2H_SLASHING;
				break;
			}
			case ItemTypePierce: // Piercing
			{
				skill = PIERCING;
				break;
			}
			case ItemType1HB: // 1H Blunt
			{
				skill = _1H_BLUNT;
				break;
			}
			case ItemType2HB: // 2H Blunt
			{
				skill = _2H_BLUNT;
				break;
			}
			case ItemType2HPierce: // 2H Piercing
			{
				skill = PIERCING;
				break;
			}
			case ItemTypeHand2Hand:
			{
				skill = HAND_TO_HAND;
				break;
			}
		}
	}

	return GetSkill(skill);
}

One thing to note is that we'll probably want to put this in the Mob class, then create a virtual function in the NPC class to always return 0, since as far as we know, NPCs don't have "skillz", plus a virtual function (above) for the Client class.

[KLS]

NPCs have skills.