Conclusive(?) Testing on Armor Upgrades & Damage Formula

**Kolaris8472** · February 25, 2014, 02:40 PM

Despite M2TW being around for all these years, I still saw a lot of differing opinions on the exact effects of armor upgrades. Setting out to satisfy my own curiosity I performed these tests. If they convince you as well, all the better.

I had heard primarily 3 theories about armor upgrades -
1. They gave +1 per level, just as the UI indicated
2. They gave +2 per level
3. They gave varying hardcoded amounts based on the commented-out table in export_descr_unit. The first upgrade is worth +4, and subsequent upgrades are worth either +1 or +2.

First I would test the kill rates against real armor by modding the armor stat in export_descr_unit. Then after I had a large sample size, I would revert armor to 0, apply upgrades, and see which real armor value they were the closest to.

The Setup:

Spoiler Alert, click show to read:

The results:

Spoiler Alert, click show to read:

Conclusions and Observations:

Spoiler Alert, click show to read:

-Armor Upgrades are +2 Armour per level. The values of 1 Upgrade, 2 Upgrade, and 3 Upgrade are very near the values of 2 Armour, 4 Armour, and 6 Armour. 3 Armour Upgrades is extremely near the results of 6 Armour (rather than 3), so theory #1 is defunct. 1 Armour Upgrade is very near the results of 2 Armour (rather than 4), and there is no difference between the value of those upgrades (upgrade level 3 is the same as upgrade level 7), so theory #3 is defunct.
-There are no separate formulas for "Hit" and "Kill", there is only 1 damage formula. I had personally assumed there was a calculation to see if your attack exceeded the targets defense and, if it did, a separate calculation would fire to see if you had killed the target. That is not the case, and the more your attack exceeds the targets defense, the higher your chance to kill them. This might sound obvious to some, but it basically means that a 0 defense target will take more casualties from Attack 9 arrows than Attack 1 arrows (as the results clearly show).
-As a result of the above point, Missile Accuracy in battle_config.xml has a huge impact. Because there is only 1 damage formula, "Missile Accuracy" (which has nothing to do with actual projectile accuracy) is in fact a straight multiplier to the effectiveness of missiles. Increase this value, and even Attack 1 arrows will have a very high chance of killing targets in full plate armour. I didn't record the test, but with a Missile Accuracy of 10 (rather than 1), Attack 1 arrows were killing Armour 12 targets 100% of the time. This is a very important number that can differ quite a bit from mod to mod, I think without the knowledge of the mod devs.
-XP does not have an effect on missile "damage". Just like the UI indicates and most people probably assume. Accuracy was already set to 100% in these tests so it still may be possible that high XP archers are more accurate.
-Weapon Upgrades do not have an effect on missile damage. Just like the UI indicates.
-Fire Arrows do not seem have an effect on missile damage. I had heard (and believed) that the cons to lower accuracy and rate of fire was higher damage and morale damage, but that doesn't seem to be the case. The variance was high enough that it could be they get perhaps +1 attack, but that wasn't primarily what I was trying to confirm so I didn't test it again.
-Kill rates on 0 Armour targets are extremely low. Just an observation, you would think that if an arrow actually hit a 0 Armour target (which they all did in these tests), you'd have more than a 20-40% chance of killing them. Apparently not.

Thanks for reading and I hope these tests were useful to you. If you've noticed I did something wrong, or if I overlooked a variable, please let me know!

EDIT: I really wish the "Preview Post" actually previewed what the post will look like...formatting was correct in the preview, but it's quite messed up here. Hopefully everything will be still be readable.

**Deceneus** · February 25, 2014, 08:09 PM

You made two modifications for testing. You gave the attacked unit zero movements points and the attacking unit almost 100% accuracy from the arrows. I think both modifications skew your results. And in particular the zero movement points setting appears to completely break your game results.

**Kolaris8472** · February 26, 2014, 07:45 PM

They skew results away from normal game situations, certainly...and towards the test variable, the effects of armor upgrades and the damage formula.

**k/t** · February 26, 2014, 09:38 PM

Great work! Any chance of a few tests to check the effect of accuracy on kill rates?

**Kolaris8472** · February 27, 2014, 06:27 PM

"Missile Accuracy" or Projectile Accuracy?

I think you mean projectile accuracy, but I'm not quite sure how to test it. It seems to be more of a scatter value, with higher numbers meaning more potential spread. The trick is I believe it has diminishing effects the closer you get to the target.

**k/t** · February 28, 2014, 03:37 PM

The number you adjust in descr_projectile.

Can't you shoot at a stationary unit from the same distance with no experience, then 1 bronze, then 2, then 3, then 1 silver and so on? Then we could see if there's a difference.

**Kolaris8472** · March 01, 2014, 05:13 PM

Oh so testing the effect of experience on accuracy? Yeah that's easy, I can do that.

**k/t** · March 02, 2014, 03:36 PM

I thought the number in descr_projectile was a scatter value.

The numbers in battle_config might be a multiplier or divider of sorts, since they're set to 1 for infantry, 1.5 for cavalry and 2 for elephants. Then again, BC had those values switched around, so I don't know if anyone truly knows what they do and how they work.

**Kolaris8472** · March 02, 2014, 04:28 PM

I concur, they're a scatter value

The battle_config is a multiplier, I haven't tested it but briefly (1 test) and I would guess that simply going from 1 to 2 would result in a 100% increase in lethality. I can run more tests on that too.