Comprehensive Retreat Study

Testing for this weekend...

I will continue with more testing... I think I just about have this formula nailed down. Factoring in the fact that a 1 hp defender can prevent a retreat is adding some complications. For sake of my sanity :eek: I will assume in my calculator and in testing that the horseman will have an available tile for retreat... I can see no conceivable way to program this into a spreadsheet (or anywhere) for that matter since game circumstances vary greater that the boyancy of a various terds in a punch bowl... I will just have to live with this minor error :(

Also, I will be continuing along the lines of previous testing done by Krayzeenbk and I with different parameters. I will not be able to post all weekend since I only have internet at work :( But Monday I hope to be full of answers on everything-you-ever-wanted-to-know-about-retreating-units-but-were-afraid-to-ask!

:D
 
Originally posted by BomberEscort
At the very least the retreat bonus of 100% is misleading... It means that of all the battles he can retreat from, he will. The rules governing retreat are as follows:

Retreat is not possible (no matter the experience of the unit); these instances are:

(1) When the fast-mover has only 1 HP left (at the beginning of the attack)

(2) When both units are fast-movers

(3) When there is no available tile for retreat

(4) When the fast-mover is fortified in a city (or colony?)

(5) When the Defender has 1hp remaining.

If the battle meets all the above criteria, and the bonus is 100% the unit will retreat...
Click to expand...

I believe I just showed that to not be true. I had horsemen with a retreat bonus of 100%.
1) They all started with 3/3 HP.
2) They were fast movers and the spearmen defenders were not.
3) They had an available tile for retreat (this is really only a problem when the fast unit is being attacked, I think).
4) The fast unit was obviously not fortified anywhere since it was attacking and was definetely not in any city, colony, fortress, on any airbase, outpost, radar tower, or anything else you can think of.
5) The defenders in the valid cases had over 1 HP remaining.

All of these criteria were met, yet the 100% retreat horsemen continued to fight in 29.69% of cases where they had a valid chance to retreat. :confused:



As for the number of cases - yes it is definetely true that I need to do more tests (as I said, probably tomorrow).

EDIT: As for the formula, you need to calculate the chance that the attacker will reach 1 HP before the defender, then multiply that by his actual chance to retreat at the chance to retreat value entered in the editor (which are apparently not the same numbers).
 
Tests on vanilla v1.29f
I have used modded units, elite slow defender with d=1000 and fast regular attacker with a=1 or 2
Defender never lost more than 1hp.

attacker--defender------------attacker
retreat---retreat---attempts--retreats---%retreats
-bonus-----bonus
--100-------51--------500-------496--------99.2
--100-------60--------500-------459--------91.8
--100-------75--------500-------396--------79.2
--100------100--------500-------324--------64.8
---50-------50--------500-------268--------53.6
---50------100--------500-------179--------35.8

Seems like defenders retreat bouns decreases the chance for the attacker if above 50.
 
I've done more tests over the weekend, and to summarize my results... Retreat bonuses 50 or under seem to work. Anything higher that 66% seems to get capped at 66%. I did tests at 100%, 99%, 80%, and 75% and they all capped around the 66% mark. I've never had results like Oystein's in my testing... Something is going on that we are missing. Maybe he has found it...

EDIT: Maybe the bonus has something to do with the ratio between the attack and defense of the respective units...
 
Originally posted by Oystein
Seems like defenders retreat bouns decreases the chance for the attacker if above 50.
Click to expand...

This would seem to go with my observation that bonuses above 50% 'seem to be broken'...

from my first post...

WIN: 5 LOSS: 65 DRAW: 130 (64.5%)... Both are regulars so the 100-100 bonus would apply, this agrees with Oystein's data

WIN: 5 LOSS: 113 DRAW: 92 (46%)... Both are regulars so the 50-50 bonus would apply, this agrees with Oystein's data

I believe Oystein has lead us in a new direction... :goodjob: I think the secret is about to be revealed...
 
I will do tests on the following:
Unit #1 = 2.1.2
Unit #2 = 1.1000.1

Unit #1-Unit #2-Actual Results

66-66 (Elite v. Elite)
66-58 (Elite v. Veteran)
66-50 (Elite v. Regular)
66-34 (Elite v. Conscript)
58-66 (Veteran v. Elite)
58-58 (Veteran v. Veteran)
58-50 (Veteran v. Regular)
58-34 (Veteran v. Conscript)
50-66 (Regular v. Elite)
50-58 (Regular v. Veteran)
50-34 (Regular v. Conscript)
34-66 (Conscript v. Elite)
34-58 (Conscript v. Veteran)
34-50 (Conscript v. Regular)
34-34 (Conscript v. Conscript)
 
Originally posted by BomberEscort
Does a slow defender get a retreat bonus???
Click to expand...

He carries the retreat bonus parameter... so you could say he does, but the parameter is never used (unless testing shows defender retreat bonus to be a factor, as I suspect it will).
 
New tests:

attacker--defender------------attacker
retreat---retreat---attempts--retreats---%retreats
-bonus-----bonus
---50--------0--------100-------100-------100.0
---49--------0--------500-------485--------97.0
---10--------0--------500-------106--------21.2
----1--------0--------500--------14---------2.8

If attackers retreat bonus if 50 points higher than the defender, he will always retreat.
 
I am almost finished with my tests above, now all we need is a formula to quantify the retreat ability, so I can incorporate it into my calculator...
 
Not to confuse the issue, but......

It SEEMS like whenever one of my tanks is hit by artillery, he NEVER retreats. In my last battle, I had about 30 tanks attacking a city. Every tank that was hit by artillery didn't retreat at 1hp (and the defender had 2+ hp), but those that did NOT get hit (or there was no more artillery to fire) almost ALWAYS retreated at 1hp. Has anyone else seen this? How does artillery fire affect the situation? In other words, does this USUALLY happen or was I just the "lucky" one...... :rolleyes:

Great work, all!

-- From The Cellar :smoke:
 
Retreating does not apply to bombardment, since bombardment is a free attack for the bombarding unit...
 
Sorry..... I must not have been clear. What I meant was this. I send a tank into a city, the defender fires artillery, my attacker is knocked down 1 hp, and after being knocked down 1 hp, it never seems to retreat, even with the enemy having 2hp +. Instead, this tank gets killed. Now OTHER tanks that don't get hit with "defending" artillery will retreat from the same stack.

GO CHIEFS!!!!! :goodjob: ;)

-- From The Cellar :smoke:
 
It should still have the possibility of retreat as long as it meets the rules for retreat given earlier in this thread... (e.g.- has more than 1 hp, has a tile to retreat to, etc...) I have had retreat after defensive bombardment and normal combat. It must be bad luck...
 
Originally posted by BomberEscort
It should still have the possibility of retreat as long as it meets the rules for retreat given earlier in this thread... (e.g.- has more than 1 hp, has a tile to retreat to, etc...) I have had retreat after defensive bombardment and normal combat. It must be bad luck...
Click to expand...
You mean you've seen a 2HP fast unit get shelled by Arty and seen it retreat when it drops to 1HP?

According to the way retreating works this is the only time it gets to retreat - when it passes the 2HP/1HP boundary. This is when it is being shelled by Arty, not when the unit attacks it later (when it is on 1HP).

Hmm, I am sure I have never seen that... :confused:
 
Originally posted by anarres
You mean you've seen a 2HP fast unit get shelled by Arty and seen it retreat when it drops to 1HP?

According to the way retreating works this is the only time it gets to retreat - when it passes the 2HP/1HP boundary. This is when it is being shelled by Arty, not when the unit attacks it later (when it is on 1HP).

Hmm, I am sure I have never seen that... :confused:
Click to expand...

From what I understand he was saying:

He had Tanks (Regular or above, since I don't see how you can get Conscript Tanks), Attacked a city, Defensive Bombardment by arty reduced him by 1 hp (Now he is 2, 3 or 4hp), the Tank continues to fight (since it was attacking the city in the first place), is reduced to 1 hp and never retreats hence winning or losing the encounter... As opposed to other Tank units, that are not defensively bombarded, attack the city, subsequently are reduced to 1 hp and always retreat. I believe he was trying to make a correlation between the Tank being defensively bombarded and its ability to retreat... and no correlation exists. :eek: If there are questions about 'retreating' please address them here. I would like to keep this thread related only to comments on the statistical analysis of the retreat ability. Thanks... ;)
 
bleh! ok, I posted in the linked thread, even if I did have to quote 4 seperate posts. :rolleyes:
 
You must log in or register to reply here.

Similar threads

S
Capital Defense Bonus?
Replies
4
Views
715
Quintillus
Quintillus
S
Citizen and Buildings Defense Bonus?
Replies
7
Views
568
Spoonwood
S
C
Min City Distance?
Replies
5
Views
828
zRoxas
Z
Acken
Early tips and Theorycrafting
2
Replies
32
Views
6K
Acken
Acken
T
Better City-Town Ecosystem Mod Proposal
Replies
2
Views
386
tman2000
T
Back
Top Bottom