Did some testing of mailed knights (40 unit size) versus spear militia (75 unit size), Spain vs Portugal, with just the basic units. This is not vanilla, but using the Carl 1.13 fix pack. Suspect that the shield fix (by Foz) included in that is the only significant change from vanilla for THIS test. These were all direct one-on-one frontal charges in the grassy plain. In more mixed terrain, the cavalry will perform less well due to greater difficulty in executing a good charge.
I ran the defense to control the spear formations and the difficulty was set to hard. I recorded only the spear kills of cavalry as the chance of a spear win was very low. In one case the schiltrom came very close with 35 kills (of 41 total cavalry including the captain).
I was reading the earlier vanilla tests and can say my results appear to diverge from those some. The schiltrom was stated to be worthless in one of those sets of tests. This suggests to me that the schiltrom issue was partly related to shields. I found the schiltrom pretty reliable against cavalry in a "last stand" situation. But not better than the overpowered loose formations. At least in my sets of 10 test battles, the schiltrom is a little more prone to a very bad result which skews the curve. One the whole it does very well as long as guard mode is left off.
The variations in performance, as others have noted, can be large. This is good. A lot of it centers on how long the captains survive on each side in these tests. In a large battle generals (AI aside) are usually better protected. So, while the commander death issue may be less in larger battles, there will also be more complex morale issues in play that may result in similar ranges. I would, however, expect results to tend more towards the center in straight face offs in a larger battle.
I saw the AI do multiple (3, sometimes 4) charges in the cases where the spears did best. I also noted that it rarely pulled off a good charge after the first. The AI didn't pull back far enough to get range to re-charge properly. It lacked time to reorganize, and charged early and in undisciplined fashion. This was complicated in the non-schiltrom cases where the spears gave pursuit, but the knights didn't even do a good solid followup charge on the stationary guard mode or schiltrom units. In most cases there was a single, successful charge with no followups, just melee. The consistent exceptions were with the columns. The sheer depth of them, and the fact that the first several ranks were wiped out cleanly, allowed the cav to pull back and charge properly at least a second time. On occasion if that one succeeded, they managed a third decent charge. In these cases the spears suffered, unlike in the formations usings more width.
The 2 rank formations were interesting in that the charge smashed through effectively splitting the spears in two, but the remaining spears usually held out long enough to inflict reasonable casualties. I suspect this would be a bigger problem in a larger battle, however, where other morale effects were at work. And a player would exploit it by driving through into the rear to create greater morale issues and probably rear charges. Unlike earlier reports, in this case the loose formation was not better. Those earlier tests were with hobilars that froze and wouldn't charge, so this may be a unit type issue, or it may be due to something in the fixes.
The 4 rank formation seems to hold longer, as does the 2 rank one, though it is pierced. The 3 rank close formations are annihilated in the charge, but they do some return punishment in the process. Oddly, the 3 rank formation is the worst of the three. The 2 rank devolves into a split melee, but lasts a while, effectively pinning the cav, and does good damage. The 4-rank holds out a while too, and is not pierced by the charge. A counterchage by supporting infantry behind the spears will probably annihilate the cav before it can pull out. The same would probably apply to the 2-rank melee, and the charging unit would have a clearer path.
The 3-rank loose formation is very prone to rout. It rarely survives the charge and it's too spread out to do much damage on impact. I suspect it's spread out just enough, and the initial charge does enough damage, that it crosses the threshhold instantly. More than half the unit dies in the charge, and the rest are very divided. Oddly, the one victory was the 3-rank. The 2-rank loses fewer, and the 4-rank has better integrity. The schiltrom, of course, is integrity defined. It tends to die to the last man.
The columns can work out well, but can also be routed instantly. The routs are largely due to the captain's instant death and the significant envelopment. The huge flank exposure makes using it in a larger engagement very risky, however, which mitigates the realism problems this could create. While they sometimes performs very well, they also perform worst in the bad cases. The big issue is how long they are. If they are long enough that no wrap-around cav ends up in the rear, they can hold okay. If the cav gets that far in the charges, like against a unit with 6 files, the rout is almost instant usually.
The loose broad formation is still way too powerful against a charge. It shouldn't stand against one at all, not with militia-level morale. The implications in single play are ugly for anyone who likes something resembling realism. Using loose formations lowers effects of missles and charges. One wonders how it will play out against other infantry. Loose spear formations may be an answer to the hordes cav archer based armies. Though I didn't track it, it also seemed to me I had more survivors from loose formations, though if the AI gave pursuit in a larger battle they would undoubtedly all be prisoners.
But on the flip side, loose columns rout early and often when charged head on. And that ignores the issue of large flanks.
The low rank formations are also overpowered. I suspect there isn't enough bonus for ranks in a receiving charge situation. Nor, clearly, is there any penalty for loose formations in those situations.
Guard mode is problematical. If you want to imflict maximum damage you want it off. But it has larger implications if you allow a line of battle to disintegrate in pursuit. If you can handle the micromanagement, I suspect, as with skirmish, it's best left off. If a unit starts pursuing, order it back into formation and line (unless the situation justifies having it pursue a routing unit). We need separate "hold formation" and "hold position" commands like we had before. Staying formed is good in most situations. (Maybe using the group formation will "hold" them into line?)
The raw data:
4 ranks, guard on: 10 20 20 15 17 13 10 23 17 9 (avg 15.4, med 17, min 9, max 23)
4 ranks, guard off: 21 23 18 12 15 33 17 14 18 20 (avg 19.1, med 18, min 12, max 33)
4 ranks, loose: 27 23 26 25 12 22 23 29 30 35 (avg 25.5, med 25, min 12, max 35)
3 ranks, guard on: 16 17 16 24 15 19 19 15 20 9 (avg 17.0, med 16, min 9, max 24)
3 ranks, guard off: 21 16 11 23 18 13 21 17 18 11 (avg 16.9, med 17, min 11, max 23)
3 ranks, loose: 6, 6, 7, 31, 13, 31, 24, 22, 39!, 15 (avg 19.4, med 15, min 6, max 39)
2 ranks, guard off: 15 24 29 18 13 28 13 14 17 14 (avg 18.5, med 15, min 13, max 29)
2 ranks, loose: 16 13 16 13 11 29 20 16 27 17 (avg 16.8, med 16, min 11, max 29)
schiltrom, guard on: 19 22 29 15 12 11 14 26 18 19 (avg 18.5, med 18, min 11, max 29)
schiltrom, guard off: 21 35 32 8 27 28 6 28 32 16 (avg 23.3, med 27, min 6, max 35)
4 file column: 21, 17, 6, 28, 18, 17, 31, 6, 4, 21 (avg 16.9, med 17, min 4, max 31)
4 file column, loose: 17, 14, 6, 5, 5 (stopped tests, not promising, all early/instant routs)
6 file column: 10, 8, 7 (stopped tests, instant routs)
6 file column, loose: 10, 11, 16, 17, 7 (stopped tests, not promising, all early/instant routs)
If you can stand the anti-realism, a 4-rank loose formation gives you good frontage and the best results against cavalry (at least using these units!) A higher quality anti-cav unit can afford to spread out a bit more into fewer ranks if facing lower quality cav, I suspect. But probably not a lot to gain. Might also increase the exposure to simultaneous frontal attacks. Of course, how this will stand up to infantry is another question. The basic default 4 rank formation is solid. The schiltrom may be the best formation to park in the city square if the assault is mostly cav and you can eliminate the infantry before the last stand.
From the attacker's view, if you're charging cav into spears you must maneuver to disrupt them. But you need a solid charge to have much chance. On hard your success won't be as good as the AI's was above and the base formation of 4 ranks is solid, if not the optimal formation.
Of course, we should see a patch soon, and all this may be irrelevant. More testing.
