2h Fixes: Discussion

[Foz]

I've seen numerous threads related to the 2h weapon bug, and an equally large number of suggestions for what to do as a fix to this problem. For those of you not already familiar with this, there is something wrong with the attacks of units using 2-handed axes, hammers, and polearms: they fail to attack cavalry.

What I'd like to see here is open discussion of the problem and its various solutions, as I haven't found anything on the forum yet that appears to be a solution that is endorsed as "the best."

To get things started, I've heard 2 proposed solutions typically:

1. Modifying the unit descriptions to use the animations of a different unit that isn't bugged.
2. Replacing the animations with non-bugged versions. This mostly seems to be animations ported from the demo.

Is one of them clearly better in any of the following senses?:

- Fixes the problem more effectively without modification of unit attack stats
- Presents a nicer or more natural looking fix
- Easier for a typical user to implement

[Musashi]

It's not just the fact that they don't attack cavalry. Sometimes they don't attack infantry either. And even when they do, the animation is easily 10x longer than an attack from a normal unit, and it goes off very rarely as well.

In short they're utterly and completely gimped.

[pevergreen]

zxiang's 2h fix is the most complete. It ports the demo ones, and fixes the then made problem with another group of 2h weapons. Billman can now be a decent part of your army.

Easy to install? Yes! just replace 2 files with the ones downloaded. Done.

[Tristrem]

i agree with the above poster, this is the easiest, and best fix I have come across yet

[hrvojej]

I think the best way to go is to use the fixed animations, but with toned down stats for the fixed units. Fixed animations with vanilla stats turn those units into monster killing machines. I used something similar to Lusted's LTC stats for the two-handed units but with fixed original animations (and not JHI animation that LTC is using), and I find that it works well and in line with how I think those units should perform.

[pevergreen]

Yes, each player will have different ideas on how each unit should perform but even with "high stats" they still get smashed on the charge. Check out the 2h Test thread i made that has results.

Basically mounted sarges will take win by 30-50 every time. With billmen bracing and in deep formation.

[dopp]

The most impressive stat line in the game will still not save a unit of infantry from any proper cavalry charge. Cavalry has double the mass compared to RTW, plus they get very strong charge bonuses against infantry. The only way to stop the charge 'properly' is to have special formation bonuses of your own. Pikemen and halberdiers have spear wall, plus someone at TWC was working on re-enabling shield wall for spearmen, since their normal spear resistance to charges is either not working properly or is insufficient to stop cavalry.

I don't have a major problem with powerful cavalry charges myself but it does make both spearmen AND heavy knights a little redundant when all you need is Mounted Sergeants or even horse archers to break an infantry line that happens to lack pikes.

[Roy1991]

Just increase the mass of the infantry - problem solved :)

[Foz]

I think the best way to go is to use the fixed animations, but with toned down stats for the fixed units. Fixed animations with vanilla stats turn those units into monster killing machines. I used something similar to Lusted's LTC stats for the two-handed units but with fixed original animations (and not JHI animation that LTC is using), and I find that it works well and in line with how I think those units should perform.

So far I am liking the replaced animations. I had been using the quick fix of replacing the bugged unit animations with those of the Zweihander unit, but of course the demo-ported animations look a lot more natural. As for balance, let's look at the following. If you compare for instance English Armoured Swordsmen with French Dismounted Noble Knights, they break down like this:

Armoured Swordsmen:
Attack: 13
Charge: 3
Armour: 8
Defense Skill: 8
Shield: 6
Cost: 610
Upkeep: 150

DNK:
Attack: 21
Charge: 6
Armour: 8
Defense Skill: 5
Shield: 0
Cost: 530
Upkeep: 225

Based on those stats, I'd expect the units to match up very well against each other, having close fights as their frontal attack and defense numbers match each other (13 attack v 13 def, 21 attack v 22 def). In practice, though, the Swordsmen get slaughtered, only taking about 1/6 of the DNKs with them. This tells me the stat rebalancing is in fact necessary. Anyone wanna link me to a well-balanced unit file to use with the replaced animations? It would be much appreciated.

Yes, each player will have different ideas on how each unit should perform but even with "high stats" they still get smashed on the charge. Check out the 2h Test thread i made that has results.

Basically mounted sarges will take win by 30-50 every time. With billmen bracing and in deep formation.

This should be something more akin to mutual assured destruction, I think. Most of the horses making the charge should die, but then so should all the billmen in the first few lines as wounded/dead horses plow into them at ramming speed. IRL this would probably be one hell of a mess.

The most impressive stat line in the game will still not save a unit of infantry from any proper cavalry charge. Cavalry has double the mass compared to RTW, plus they get very strong charge bonuses against infantry. The only way to stop the charge 'properly' is to have special formation bonuses of your own. Pikemen and halberdiers have spear wall, plus someone at TWC was working on re-enabling shield wall for spearmen, since their normal spear resistance to charges is either not working properly or is insufficient to stop cavalry.

I don't have a major problem with powerful cavalry charges myself but it does make both spearmen AND heavy knights a little redundant when all you need is Mounted Sergeants or even horse archers to break an infantry line that happens to lack pikes.

I also don't have a problem with powerful charges, since as I mentioned the horses, even if wounded/killed on impact, still pose a huge threat to any men that are in their way.

As for increasing the mass of the infantry as Roy1991 suggests, I don't really think it's a good solution. The horses really should be able to knock men around a great deal, they weigh a LOT more and are of course charging. It seems to me a better solution would be to fix the formation bonuses for applicable units, as it's clear that a unit not explicitly bracing and prepared for a cav charge should in fact be completely cut to ribbons by it, while a prepared unit of the correct type should be able to hold its own or at least be assured of slaughtering a good number of charging knights on impact, which does not really seem to happen now in general.

[JCoyote]

Actually the halberds should be handing the swordsmen a beating. They have anti-armor ability, and are armored; swordsmen are armored, but lack anti armor ability. The balance is the swordsmen are less vulnerable to missile fire under all conditions. whereas the halberds don't have shields. I think it would balance a bit closer if both sides were firing in missiles during their fight.

But, point being, there is more to balance than simply how well they hammer each other in melee. There are compensating factors for swordsmen still, and in some situations swordsmen might still have even a melee advantage, like on walls.

I have a lot of training with a sword; even so, given a sword and shield a halberd in modestly coordinated hands still has an advantage against me on an open field.

[Lusted]

I think i managed to balance the 2 handed fix quite well in my LTC mod, using the JHI to replace the bugged one.

I also increased the mass of spearmen so they do better against cav, as spearmen in normal M2Tw are useless. Making spearmen better against cav beings back more of the rock, papaer, scissors dynamic.

[Zatoichi]

I have spent a long time trying to rebalance the billmen, Dismounted Knights etc with the original animations from the demo. I ended up reducing the attack and charge values quite drastically as a result of one-on-one testing against Dismounted Chivalric Knights, basic spearmen and basic cavalry. I've also made minor adjustments to the recruitment cost of castle trained billmen (a very slight increase).

I'm happy with the way these units are matching up now, they feel 'right' to me personally.

However, I still don't know what effect these changes are making to the AI - it all has a knock on effect with the AI's desire to recruit these units in the first place, and the effect on the AI's auto-calc'ed battles. So I can't say with all honesty that I have achieved a more balanced game, but at least when playing as England my billmen don't get chewed up by cavalry, and neither do they decimate units which cost 4 times the amount to train.

[Radioman]

I have spent a long time trying to rebalance the billmen, Dismounted Knights etc with the original animations from the demo. I ended up reducing the attack and charge values quite drastically as a result of one-on-one testing against Dismounted Chivalric Knights, basic spearmen and basic cavalry. I've also made minor adjustments to the recruitment cost of castle trained billmen (a very slight increase).

I'm happy with the way these units are matching up now, they feel 'right' to me personally.

However, I still don't know what effect these changes are making to the AI - it all has a knock on effect with the AI's desire to recruit these units in the first place, and the effect on the AI's auto-calc'ed battles. So I can't say with all honesty that I have achieved a more balanced game, but at least when playing as England my billmen don't get chewed up by cavalry, and neither do they decimate units which cost 4 times the amount to train.

I'm using the XNGfix and did alot of like tests until I matched my Dismounted English Knights against Dismounted Portuguese Knights. In my EDU file they are exactly the same stat wise, except a slightly higher att bonus for the English units...Every test concludes with the Portuguese winning!!! Does anyone know why this is???

[pevergreen]

Charge? Random Duels? English General Getting knocked around.

[Zatoichi]

I'm using the XNGfix and did alot of like tests until I matched my Dismounted English Knights against Dismounted Portuguese Knights. In my EDU file they are exactly the same stat wise, except a slightly higher att bonus for the English units...Every test concludes with the Portuguese winning!!! Does anyone know why this is???

It depends who you are controlling in your tests, who charges, whether the general dies early on or not - it's very hard to do one-on-one unit match ups and achieve reliable results. It would be much better to link up 2 PCs and play these fights out against another human opponent so you could model exactly what factors you were looking for, but that's not an entirely practical solution. Oh, and also it's extremely dull!

[Foz]

Actually the halberds should be handing the swordsmen a beating. They have anti-armor ability, and are armored; swordsmen are armored, but lack anti armor ability. The balance is the swordsmen are less vulnerable to missile fire under all conditions. whereas the halberds don't have shields. I think it would balance a bit closer if both sides were firing in missiles during their fight.

But, point being, there is more to balance than simply how well they hammer each other in melee. There are compensating factors for swordsmen still, and in some situations swordsmen might still have even a melee advantage, like on walls.

I have a lot of training with a sword; even so, given a sword and shield a halberd in modestly coordinated hands still has an advantage against me on an open field.

I still wasn't satisfied that results were making sense, so I removed the AP stat from DNKs since it purportedly makes all the difference. The Armoured Swordsmen were still completely destroyed every single time. The AS's should have a 1 point advantage in frontal defense and an even matchup for attacking. As expected they took a bit bigger hit on the charge than the DNKs took in return, but immediately after that they continued to get absolutely beaten into the ground. This in spite of the swordsmen slashing at what appeared to be a considerably higher rate of fire than the DNKs. Anyone know what's going on here?

I'm guessing maybe the shield and defense skill of the swordsmen are never allowed to apply at the same time, meaning they would in practice be at 14 def from the left and 16 from the right, never achieving their claimed 21 defense from the front. Or maybe it's just that the "front" area is so small that in practice they never get the full overlap of all bonuses. If either is in fact the case, it would mean that the 2-handed units aren't overpowered, but rather that the shield units are horribly underpowered due to working not as intended. I sorta hope this is the case, as it would neatly explain the issue I'm seeing as well as the horrible ineffectiveness of spear troops (Armoured Sergeants for instance list 5/3/6 for armor, and would be at 8 from armour+skill or 11 for armor+shield instead of their frontal 14 from all 3 during combat).

It's certainly too early to claim this is the case yet, but I'll do testing (starting with non-AP DNKs set at 15 attack, the hypothetical swordsmen avg value) and share the results as soon as I'm home today.

I think i managed to balance the 2 handed fix quite well in my LTC mod, using the JHI to replace the bugged one.

I also increased the mass of spearmen so they do better against cav, as spearmen in normal M2Tw are useless. Making spearmen better against cav beings back more of the rock, papaer, scissors dynamic.

You mean only by replacing the animation, or by using it plus modded unit stats? If it's the latter I'll have to plug in the file and mess with it soon to see what I think. My list just seems to grow!

I have spent a long time trying to rebalance the billmen, Dismounted Knights etc with the original animations from the demo. I ended up reducing the attack and charge values quite drastically as a result of one-on-one testing against Dismounted Chivalric Knights, basic spearmen and basic cavalry. I've also made minor adjustments to the recruitment cost of castle trained billmen (a very slight increase).

I'm happy with the way these units are matching up now, they feel 'right' to me personally.

However, I still don't know what effect these changes are making to the AI - it all has a knock on effect with the AI's desire to recruit these units in the first place, and the effect on the AI's auto-calc'ed battles. So I can't say with all honesty that I have achieved a more balanced game, but at least when playing as England my billmen don't get chewed up by cavalry, and neither do they decimate units which cost 4 times the amount to train.

You've done a lot of work, so link the file: we might all like it

[Lusted]

You mean only by replacing the animation, or by using it plus modded unit stats? If it's the latter I'll have to plug in the file and mess with it soon to see what I think. My list just seems to grow!

Both, modified animation only results in severly overpowered units. I think i dropped the attack for all units with the bug by at least 8 after changing their animation.

[Carl]

@the_foz_4:

You might also want to compare the mental stats of the two units and also weather one has better hardiness, (can't be bothered to check ATM as i've just spent 4+ hours digging around myself).

These factors appear to have a BIG affect on kill rates to me as I’m working on my own 2-hander fix and the basic trick I’m doing is to give ALL units equipped with a given weapon type the same animations set. Thus in theory, only training, discipline, special abilities, and stat values actually affect the match-ups between units with the same weapon type.

So my Heavy Billmen and JHI have the exact same animation now, (Halbeard_Militia), but the Billman have 4 better attack 1 better charge and 2 better defence than the JHI yet they STILL get absolutely thrashed by the JHI. The only differences I can spot is that the JHI have better morale, Training, Discipline, and are "Very_Hardy" as apposed to the Billmens "Hardy".

Thus bear in mind that these can make a big difference on their own, it isn't just attack, defence, charge, and animations that determine kill rates.

Finically, someone did a test a bit back and found that attack trumps defence. Apparently if you have 2 units that are identical, EXCEPT that 1 has high attack and low defence and the other is the exact reverse, (i.e. it has a defence value the same as the other units attack value and an attack value the same as the other units defence value), it will get beaten, (i.e. the unit with the high defence). So your results do seem to match up with that in any case and with my own experience with DEK.

[Zatoichi]

You've done a lot of work, so link the file: we might all like it

Well, I've done a bunch of other bits and pieces to my EDU as well, so I'll just post the relevant lines - but please bear in mind this is the culmination of me taking other peoples' ideas from a thread over at TWC (namely Zxiang1983, VaultDweller, Godzilla, Raguen and Lusted to name but a few) and running my own tests and tweaks. The results are pleasing to me, but I'd encourage anyone to have a play with the 'original' demo 2Hander animation and mess with the stats themselves.

Spoiler Alert, click show to read: 

Woodsmen:
stat_pri 8, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2

Croat Axemen:
stat_pri 9, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2

Religious Fanatics:
stat_pri 9, 1, no, 0, 0, melee, melee_blade, blunt, mace, 150, 2

Billmen:
stat_pri 8, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 350, 125, 75, 55, 230, 4, 50

Heavy Billmen:
stat_pri 9, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 460, 150, 75, 55, 300, 4, 70

Bill Militia:
stat_pri 8, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 330, 100, 65, 50, 170, 4, 40

Heavy Bill Militia:
stat_pri 9, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 450, 150, 75, 55, 300, 4, 70

Berdiche Axemen:
stat_pri 11, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 580, 150, 90, 70, 380, 4, 90

Varangian Guard:
stat_pri 12, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 550, 175, 100, 75, 520, 4, 130

Norse Axemen:
stat_pri 11, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2

Dismounted English Knights:
stat_pri 12, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 650, 225, 100, 75, 530, 4, 130

Dismounted Noble Knights:
stat_pri 12, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 650, 225, 100, 75, 530, 4, 130

Dismounted Portuguese Knights:
stat_pri 12, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 650, 225, 100, 75, 530, 4, 130

Galloglaich:
stat_pri 11, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2

Free Company Men at Arms:
stat_pri 11, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 530, 180, 75, 55, 330, 4, 80

Mutatawwi'a:
stat_pri 8, 1, no, 0, 0, melee, melee_blade, blunt, mace, 150, 2
stat_cost 1, 260, 70, 90, 70, 160, 4, 40

Tabardariyya:
stat_pri 11, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 590, 175, 100, 75, 490, 4, 120

Eagle Warriors:
stat_pri 11, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2

Galloglaich Mercs:
stat_pri 11, 1, no, 0, 0, melee, melee_blade, slashing, axe, 150, 2
stat_cost 1, 580, 180, 90, 70, 380, 4, 90

[sivlar]

I think the best way to go is to use the fixed animations, but with toned down stats for the fixed units. Fixed animations with vanilla stats turn those units into monster killing machines...

Just wanted to throw this out there, since lots of folks say using the vanilla stats with the pre-gold animations unbalances units.

In the pre-gold demo with the old animations, the stats were even higher on those units than they are in the final release with the broken animation.

I'm at work atm and don't have access to the EDU from the demo but I recall all the 2-handed units in there having an attack rating even higher than what the final has.

I personally think the unbalance comes from the animation itself, not the stats. Since the spin attack and such are very fast compared to what other units have.

Any way to make the animation run slower than what it does now instead of nerfing the stats of the units that use it?

[Radioman]

It depends who you are controlling in your tests, who charges, whether the general dies early on or not - it's very hard to do one-on-one unit match ups and achieve reliable results. It would be much better to link up 2 PCs and play these fights out against another human opponent so you could model exactly what factors you were looking for, but that's not an entirely practical solution. Oh, and also it's extremely dull!

Thanks that makes sense, if you have friends to play with...Nonetheless I guess that it's just one of those crazy things. I wonder if I give my DIs. english Knights exp. points, perhaps that will do. I've done 2 on 2 as well always ending in the same way...

[hrvojej]

I did my tests with e.g. dismounted English knights vs. dismounted gothic knights (no shield, so one less problem to worry about), and even with an attack of 8 (like Lusted's LTC mod is using) DEK thrash the gothics more often than not. When you look at the raw stats though, they actually favor the gothics (DGK 14 att vs. DEK 13 def, 8 DEK att vs. 9 DGK def - after ap is factored in; also, I set the skeleton compensation and mass to 1 and morale to 11 for both units). So, from the front, DGK should have an advantage of one point for both attack and defense.

Since defense value is only taken into account from the front and to the right, it is likely that the value of 9 total def for DGK actually will not hold in a number of situations (it's just the maximum def they can have). The ap of DEK reduces the def of DGK to 5 if DEK hit them from anywhere else other than front/right. However, in the opposite situation the difference is 14 vs. 8, so if the difference is assumed to be linearly factored into the calculations, DGK should now have an even larger 6 vs. 3 advantage when hitting from other directions. Therefore, there's either something going on with the animations or the math is not what I think it is.

With an attack of 21 it's just ridiculous what DEK are capable of, and so I set their attack at what feels right for me (9 att). I tried to slow them down with increasing the delay between attacks, and I tried to make them less lethal by decreasing skeleton compensation, but it didn't seem like it made a tangible difference. However, it also seemed that, with the fixed animations, the 2h axes were less powerful than the maces of dismounted knights, but I didn't feel like doing another round of exhaustive tests to determine that for sure.

I don't know how to mod animations though, nor whether it's possible to do so at all. But yes, I agree that this sort of thing could have an effect on strategic AI, since, unlike the battles, it is most likely still guided by stats as such.

[Foz]

Did that testing I was talking about, and the results are in: shields are totally borked in close combat. Before I discuss, here are the raw numbers listed as "Armoured Swordsmen kills/Dismounted Noble Knight kills":

Spoiler Alert, click show to read: 

In all cases I use the same exact setup and orders, only altering the unit stats in the file. The same DNKs are used each time, with the AP stat removed and attack value set to 15. For a baseline, I start with unmodded AS. AS stats are listed (armor value/defense skill/shield value).

A. AS unmodded (8/8/6) vs. DNK

44/57
42/56
45/56
47/59
22/55
-----
200/283

B. AS 22/0/0 vs. DNK

50/30
40/29
55/29
40/25
45/43
-----
230/156

C. AS 0/22/0 vs DNK

39/17
60/48
60/59
38/23
34/19
-----
231/166

D. AS 0/0/22 vs DNK

14/54
28/58
13/48
13/60
21/58
-----
89/278

E. AS 0/0/0 vs DNK

28/54
22/60
13/41
17/52
22/59
-----
102/266

F. AS 8/14/0 vs DNK

34/21
56/39
35/28
38/12
52/31
-----
215/131

A is our baseline, the way the game operates with vanilla unit settings. Assuming that the units primarily engage each other from the front, this should be identical to B, C, and D if everything is working correctly. They in fact do primarily engage from the front, which I helped assure by lining the swordsmen up in a 20x3 formation like the knights use.

Note that B and C have almost identical results, indicating skill is as good as armour in this battle. D is absolutely horrible by comparison, telling us that the shield does not protect its bearer in close combat as much as his armour or skill do. In fact we can see in E that a unit with 0/0/0 def stats performed slightly better than the one with 22 shield stat only, which means if anything the shield points actually hurt its defense. This is not at all significant though, and should probably be chalked up to random error in testing.

F further illustrates what happens if the unit's shield points are converted into the skill stat - it performs about on par with the 22/0/0 and 0/22/0 unit, further affirming that armour and skill are good while shield is broken.

What this all tells us is that the shield stat at least in this case (I'm guessing in all cases) doesn't do a damn thing to protect the unit in melee combat. It does work for ranged protection from what I've seen: I set shield in 3 different runs to 0/6/26 and took as corresponding losses 60/48/13 from an unmodded unit of french crossbow militia.

So then the main problem now is that all the shield units are operating at less melee defense than intended, by an amount equal to their shield value. For knights this is a 4 mostly which is probably why they're easy kills in melee. Lots of infantry have 6, which explains why shield infantry of all sorts seem so horribly underpowered. The easiest fix I see is to add the shield value right into the skill field, because the problem is the shield isn't affecting melee, which points in skill ONLY do, so will fix the problem. I'd expect it to throw the recruitment AI a bit of a loop though since according to the computer the defense stats of all shield units will have gone up. Maybe it'll make shield troops look more tempting than buttloads of artillery. I'd rather face stacks of armoured sergeants any day, it's more interesting

[Musashi]

I doubt that the shield stat is actually hurting the unit in melee. More than likely you're just seeing the small variances that come from random chance (There is a random element in unit combat you know).

My guess it the shield does absolutely nothing in melee combat. It's just meant to provide cover from ranged fire.

[hrvojej]

Actually, I plugged the results that foz posted in the shield problems thread into a stats program, and there appears to be a negative effect of the shield in melee. Of course, if I had larger sample size (more runs) I would be able to tell whether this is the case with more certainty, but even as it is now it seems likely that the shield is in fact hurting the units in melee.

[Musashi]

Yes, we can just look at the numbers and see that the test went more poorly with the unit having a shield. However, since there is random chance involved, it's entirely possible that 5 battles could just go worse simply due to that chance. To have a reasonable basis for claiming the shield detracts from their performance you would need to run a couple hundred battles and get the same results, and even then if it's a 5% difference or less it's probably just a fluke.

The survivors in 0/0/0 battles for the AS average at 20.4... with 0/0/22 stats they average 17.8. That's such a tiny margin of difference for 5 battles that it probably doesn't mean anything at all.

Especially considering how high the shield stat was set. If it was actually hurting them, we'd likely see a much wider gap.

[hrvojej]

I was actually refering to his results in this post: https://forums.totalwar.org/vb/showp...76&postcount=1

Thought the sample size is only 5, the effect size appears to be large enough that it at least warrants further testing. Btw, means of armoured sergeants killed with shield 0 is 39.4 (+/- 5.6), and with shield 26 the mean number killed is 70 (+/- 2.5). Either they don't rout with the shield and keep on fighting, or there's something weird going on with the calculations.

[Musashi]

I'd still argue it needs a bigger sample size, but yes, that's certainly more worrisome than the results posted here. The problem is that the results here show a negligible difference if any, so maybe shields are only a hindrance when defending against cavalry?

There's also the question of whether animation enters into it... By giving them a shield stat do they attempt to use a "blocking" animation that they wouldn't otherwise use? If so this could decrease their attack rate, meaning higher casualties in the long run (Since kill rate is king).

[Carl]

It's worth noting that Shields can have separate CC and ranged values too so if you wanted to transfer the Shield to defence skill, then setting the shield to still do it's job at range whilst having zero in melee would hopefully get around the computer acting like a dummy and treating them as better than they are.

To me the numbers also look VERY decisive, ignore the individual ones and look a the totals, their are MASSIVE variances their, the only way that could happen is if luck was having very big effect in each group of 5. That’s not impossible, but the chances of this happening in such a consistent way, (bear in mind that only result D is the one where considering anomalous, look at it like that and only where they have no defence value or where they have a lot of defence in their shield does it give similar results, thus it's actually anomalous when compared to 15 other results and matches loosely up with another 10). The numbers defiantly worry me although I too would prefer to see a larger sample size.

[Revenant]

This shield problem can be also the reason why some late era heavy cavalry (gendarmes) are actually better in combat vs. earlier knights more than their stats indicate.