Wierd and funky discovery

[Swamp Thing]

Okay, it seems the attributes in terms of attack and defence work in multiples of 0.5 rather than 1.

Here is a test I conducted. (note, all units were identical in attributes unless stated otherwise, plus it was all conducted on mostly flat terrain without major inclinations).

I had one unit of 60 men with 3 defence, attack and armor fight a unit of 100 men with 2 att, 2 def and 3 armor.

In 3 out of 4 test runs, the 60 man unit wiped out the 100 man unit and won.

Next I took the original 100 man unit with 2 att, 2 def and 3 armour, and matched it against a unit with 100 men and stats of 2.5 attack, 2.5 defence and 3 armor.

4 out of 4 times the unit with 2.5 att and def won.

Next I took the original 60 man 3 att 3 def 3 armor formation and had it fight against the 100 man 2.5 att 2.5 def and armor 3 unit.

3 out of 4 times it was the 2.5 unit that won.

How about that

[CBR]

What about 2 units of same size but one with 2.5 in att/def versus a unit with 3 in att/def?

CBR

[barocca]

Quote[/b] (Swamp Thing @ Sep. 21 2003,08:55)]Okay, it seems the attributes in terms of attack and defence work in multiples of 0.5 rather than 1.
...
Next I took the original 100 man unit with 2 att, 2 def and 3 armour, and matched it against a unit with 100 men and stats of 2.5 attack, 2.5 defence and 3 armor.

4 out of 4 times the unit with 2.5 att and def won.
...
if the game did not crash using 2.5 attack then it is quite likely that your supposition is correct,

you do need to conduct some more tests using units of the same size,
also try att/def of 3.5 vs 3, and 3.5 vs 4, etc...

[rory_20_uk]

A very worthwhile discovery - it would allow a lot mre fine tuning of units stats. Good on yer http://www.totalwar.org/forum/non-cgi/emoticons/smile.gif

http://www.totalwar.org/forum/non-cgi/emoticons/smokin.gif

[Swamp Thing]

As far as I can tell the result is stable, there was not a single crash.

Cheers for the complement Rory.

http://www.totalwar.org/forum/non-cgi/emoticons/wave.gif

I'll be doing some more testing tomorrow. Right now I'm taking a break from Total War. I've been doing a lot of unit modding and its in my head like an addictive song.

Could some of you run some tests, see if you get the same results?

[Puzz3D]

Swamp Thing

It could simply be that the 2.5 is getting rounded up to 3.0. That is also consistent with your results. I tested this carefully in STW a long time ago, and found the combat values were converted to integers, although, you could enter them as decimal numbers. I don't remember if they were truncated or rounded. Remember, this number appears as an exponent in the chance to kill formula and raising a number to a real exponent is a lot more computationally intensive than raising it to an interger exponent. This computation has to be done every second for every man who is engaged in combat. I'll try some tests on MTW and see what I get.

[Puzz3D]

These are the results of a test I ran in custom, flat map, arid, fine day. I used Italian Light Infantry vs Order Foot because they have the same stats. The only thing I had to do at the beginning of the battle was reform the Italian Inf into 25x4 deep to match the Order Foot. Just to make sure I was adjusting the correct values I made a run with the Italian Inf set to 2/5 and they did win easily killing 55 and loosing 11. I could also see that the ai uses an indirect approach when its unit is weaker by 2 or more combat points, but uses a direct approach when its unit is equal or only down 1 combat point.

I made 10 runs with the Italian Inf set to 0.5/3.5. The ai has the Order Foot 0/3, and it charges directly ahead as do I with the Italian Inf. I recorded if I won or lost and kills/losses.

Run, W/L, kills, losses

1) L 51 80
2) L 74 80
3) W 80 56
4) L 62 80
5) W 64 55
6) L 63 79
7) W 76 62
8) L 56 80
9) W 65 47
10) L 58 56

Totals: 6 lost and 4 won; 649 kills and 675 losses

Now I set the Italian Inf to 1/3 which gives them a 1 point advantage over the Order Foot to see how much a 1 point difference (20% advantage in chance to kill) makes in this test.

1) W 76 53
2) W 67 44
3) W 81 36
4) W 76 37
5) W 69 55
6) W 65 33
7) W 74 63
8) W 66 57
9) L 73 80
10) W 76 38

Totals: 9 wins and 1 lost; 723 kills and 496 losses

The difference in kills is 45% whereas the difference in chance to kill going by the combat factors is only 20%. Some factors that could increase the kill ratio once a numerical superiority is attained could be 2 on 1 attacks and less fatigue due to more men just standing. Fatigue wasn't a factor in the one replay I watched except my Italian Inf had to overcome a slight fatigue disadvantage incurred when it walked to the ai's Order Foot. This disadvantage had been erased by the time both units lost 2 bars of fatigue. Once the units were equal in fatigue the Order Foot was gradually able to build a numerical superiority, but this didn't happen until the units were already down to about 1/2 strength. In most of the fights, the Order Foot maintained good numerical parity with the Italian Inf until the latter half of the fight.

[CBR]

And one more factor when spears are involved: rankbonuses.


When a spear unit starts to lose men from the third rank the men in the front (fighting) rank loses 1 attack and 1 defense.

CBR

[Puzz3D]

I guess I have to do the test where they are exactly equal to complete the series.