Help to measure up the tactical AI

[Cheetah]

Gentlemen, thanks for the tests and for the detailed battle descriptons. Very interesting.

@ Ars Moriendi, I was thinking about the first kind of rating (your (a) rating), so it is fine. Of course a defeat can have different difficulties too (like losing a close fight or losing it big time) but I expected people to win. Of course, I proved myself wrong as I lost one of these battles ... but let's stick to this rating as I still expect most of the people to win all 4 battles.

Also, late BG not much of a problem, I guess you could have done the same with an early BG (i.e. chasing routed units). Of course early BG is still the "default". What I am interested in though is that whether you could do this (i.e. send your general behind enemy lines to chase routers) without any harassment, or were your BG harassed by enemy cavs?

[Ars Moriendi]

What I am interested in though is that whether you could do this (i.e. send your general behind enemy lines to chase routers) without any harassment, or were your BG harassed by enemy cavs?

Well, they were occasionally intercepted by some none-routing unit, but nothing that wasn't quickly solved with the help of another unit breaking from the main fight to "sandwich" the offender.
The trick is to keep them far enough to be safe, but not too far so the routers have time to rally. Also, as these battles went, most of the HRE cavalry was busy from the beginning - either deeply embedded in my line from the initial charge or chasing my xbows.
The large number of casualties these units scored (~350 for the BG with milan, ~250 one of the huscarls) was also helped by the fact they were in the ideal position to mow down large numbers of enemies in the final mass rout.

Another point : some fast, light cav might be better for this role, although probably the first contact with HRE knights would've been the end of them.

Two questions :
- what is it that you hope to achieve with these tests ?
- when do we get the second scenario ? I haven't played the game in months (still waitin for the darn patch ) but seeing your invitation I just couldn't resist to fire up some custom battles...

[hrvojej]

I haven't played the game in months (still waitin for the darn patch ) but seeing your invitation I just couldn't resist to fire up some custom battles...

Same here

[Cheetah]

- what is it that you hope to achieve with these tests ?

Well, I would like to know which setups are easiest for the human player, and which setups are the most challenging, hoping that the devs will take notes and will try to guide the AI recruitment policy (and perhaps the upkeep costs) accordingly.

The second scenario:

Human player has the same armies as before.

AI army has a slight change as it has 4 more melee troops (2 feudal knight and 2 dismounted feudal knight) instead of the 4 xbows:

AI army, HRE:

1 BG
4 imperial knights
2 feudal knights
2 mounted sergeants
4 dismounted imperial knights
4 dismounted feaudal knights
3 armoured sergeants

[Ars Moriendi]

Results for the second scenario :



Screenshots :

danish1
danish2
danish3

milan1
milan2
milan3

hungary1
hungary2
hungary3

france1
france2
france3

Write-up later, when I get time...

[nikolai1962]

Oops, i did it the wrong way round--kept the same hre army against the four armies you specified as the opponents

[Ars Moriendi]

Evaluation for the second scenario : Harder than the first. If I had not the experience from the first scenario with similar armies, the outcome would have been worse.

1.Danish - identical deployment, similar battle. More difficult, although the results are almost the same.

2. Hungary - identical deployment, battle not so similar. First, the AI just stood there for a while getting shot, whereas in the first battle it attacked right from the start. Second, more cavalry for the HRE allowed the AI to chase my HA more effectively. Horse archers still dominate the battlefield, but they're difficult to manage against large numbers of cavalry so this time more maneuvering was required. I took more than twice as many casualties.

3. France - identical deployment. I started with a frontal charge (4 Chiv Knights), destroying 4 units of dismounted knights. I pulled back to prepare a second, but there was no time for it because the AI countercharged his imperial knights throughout the line. All the rest of the battle was cavalry maneuvers : charges, counter-charges, flanking, surrounding etc. The few casualties my infantry had were probably caused by boredom, since they were never issued any orders...

4. Milan - here's the ugly part : I lost the first time, the results listed in the table are from the second run. I know we're supposed to give the first try only, but I was so pissed I just clicked "Play Again" before thinking about recording the results. Sorry.
...
That being said, the problem here is the uselessness of the foot missile units. The first three battles can be won without much of their assistance, but the Milanese is close to being unwinnable (well, for me at least). Not having any kind of terrain advantage and with a cavalry balance of 3-vs.-9 it's quite difficult to employ them effectively. My usual plans for missile units would be (a) deployed in front in a wide line, fire at the approaching enemy - which in these scenarios was useless considering the AI just ran at me and there was no way to delay them enough for the xbows to cause serious damage; or (b) deployed on flanks to move outward and fire at flanks and rear of the engaged enemy line - not achievable due to significant disadvantage in cavalry. I might as well start the battle with 6 less units.
The way I solved this in my second Milan attempt : a kind of deployment I've never used before. I put the italian spears in schiltrom and placed them with rather wide spaces between, like this :

   *          *           *

   m    *     m     *     m   *

        m          m 

 x     x     x     x     x     x

where * is schiltroms, m is dismounted man at arms and x crossbows.
The AI charged the schiltroms, and while they formed these "circles" around the spearmen it allowed the dmaa to engage in a sort of flanking. More importantly, the wide spaces allowed the xbows to fire between the engaged units long enough to kill significant numbers, while being protected from the initial charge. If you look at the results you'll notice they had the most kills (with the exception of the very special "router patrol", of course). Some of them were eventually caught by enemy cavalry going around the infantry formations, but they did enough of their job by then.
This solution allowed the foot missile units to change their role of "power amplifiers", meaning they just helped reduce casualties by killing from afar in battles were I had the upper hand and would have won anyway without them - to a more active part, being the little something that wins the day.

[Tyrac]

Are there any rules in terms of PAUSE or speed? Being able to hit Pause in a SP battle at any time makes things far far far easier.