The Complete EDU Guide - Discussion

**DimeBagHo** · 01-04-2008, 06:37

Originally Posted by Makanyane

I was wondering if anyone has managed to come up with any conclusive information about how unit sizes and stats affect auto-calc (other than just auto-calc sucks!) - is there anything you can do within EDU to help an AI faction with smaller unit sizes?

With regular cavalry units you can use the second value on the stat_health line to improve auto-calc performance. It has no effect on performance in a normal battle, it might have some effect on AI recruiting preference (it looked that way to me but I didn't test it rigorously), but it definitely has an effect on auto-calc. I have used this to improve the performance of horse archer factions in XGM with great success.

**Dol Guldur** · 01-04-2008, 12:10

That's very interesting, Dime. Thank you. Something for Aradan to check out and update ;)

**Red Spot** · 01-04-2008, 12:49

you may also have noticed that even though they dont use it, vanilla does set secondary armour stats for cavalry as well ...

Code:

stat_sec_armour  0, 1, flesh

Though I dont know if it may influence anything, I just use it everywhere or nowhere at all ...

G

**Dol Guldur** · 01-04-2008, 14:13

Yes, I have noticed that. This is definitely something for Aradan and his army of reliable, committed beta testers to look into

**Aradan** · 01-12-2008, 18:34

Here are the results of some of my recent tests. I give only the average percentages (fought some sixty-sth battles), because formatting the entire table would take a good while...
Difficulty is set to Medium-Medium, and there are no generals involved. The first row is using stat_health increase only, the second using stat-sec_armour increase only and the third is normal (both values are 0).
The first column is using 4 units of riders (2 groups of 2 couples of the same unit-type), the 2nd using just the first group, the third the other one, and the 4th and 5th using 1 unit of each troop-type.
The "*" sign means that I think the result here was so out of the 'acceptable range" that it shouldn't be taken into account as it is.

Increasing stat_sec_armour's second value from 0 to 2 didn't seem to have any significant result, that cannot be explained by the randomness of the results.
Increasing the second value stat_health from 0 to 2 might have some effect, but given the range of results I am not yet 100% certain. It might well be a fluke of luck, as auto-resolve seems to be very random; using the exact same armies I got results ranging from Clear Defeat to Clear Victory. Still, there is a decrease to enemy kills in most occasions that is significant and needs further research.

Other than that, the results *suggest* (though the number of tests is still quite small) that the auto-calc is using some sort of logarithmic function, that lacks balance when armies are very small of very big.

Dime, what values do u use for your horse-archers stat_health? 1,2 or way bigger?

EDIT: I tested some more using stat_health 1,10 for one unit type and 1,20 for the other. Indeed now the units did much better. Whether using all four units together, or just a couple of them or even a single unit, they always inflicted sth like 53% casualties to the enemy. It seems the auto-resolve has a cap somewhere there for this particular combination of armies. The big deal is that now my cavalry was taking far less casualties than before.

one unit 20 sec-hps: now 6% - 45% before
one unit 10 sec-hps: now 8% - 39% before
two units 20 sec-hps: now 9% - 35% before
two units 10 sec-hps: now 10% - 21% before
four units 15 sec-hps: now 5% - 15% before

So it seems that indeed stat_sec is used in auto-calc and it can make quite a difference in the amount of sustained casualties. And it actually makes no difference in real-time battle!! Now we just got to make sure that it doesn't affect AI-recruitment... :) If it doesn't then this can be a great tool for campaign balance! Thx DimeBagHo for spotting that!

EDIT 2: stat_sec_armour's second value seems to do nothing at all for horsemen and infantry. No idea why vanilla sets it as 1 for cavalry.

**Aradan** · 01-13-2008, 01:01

Previous post updated, it seems we're on a good road here.

**Dol Guldur** · 01-13-2008, 01:03

Well done to DimeBagHo for pointing this out, Lgk for spotting it originally and for Aradan for the research/confirmation. At last, a way to have some control over that darn autoresolve!

**Aradan** · 01-13-2008, 01:07

It would be interesting to see if infantry can be affected by this... I doubt it, but who knows. If anyone has any info on that, it could save lots of time.