Are Damage Calculations bugged?

[mxb2001]

Yes, sorry about beating a dead horse (re: LUCK) but I was wondering how often one can see a 3.8% chance come true. Yes, 3.8% of the time, I know. But to take another tack, is it possible that while the much inspected randomness has been given a clean bill of health the damage calculator might be off for some units with odd rare abilities. To wit, berserk, drain, backstab, marksman, etc. I just had a drake fighter attack an ulfserker (nicely softened up by ranged first) on a hill and the dam calc claimed a 90+ chance to kill and only a 3.8% of the drake snuffing it. That's what happened though and while it could be chance I must admit I've been seeing the less than 50% odds come out a lot more than 50% of the time for killing attacks and this seems the straw that breaks my camel.

So, maybe dam calc is reporting falsely for units with unusual abilities? Food for thought...

[josteph]

while it could be chance I must admit I've been seeing the less than 50% odds come out a lot more than 50% of the time for killing attacks

Have you actually counted this, or is this just an impression?

What's the setting of the "Allow damage calculation with Monte Carlo simulation" advanced preference?

[Ravana]

Damage calculations are certainly not aware of eventbased abilities, and I suspect they fail with some filter cases too.

[josteph]

I checked. A Drake Clasher attacking a Dwarvish Berserker on flat terrain in daytime has 76.9% chance to kill and 22.7% chance to be killed in the damage prediction dialog. Trying it out, out of 750 iterations, 590 times the Drake Clasher died and 160 times the Dwarvish Berserker died. That's 78.7%.

Here's the code. (Run it with wesnoth -t; I didn't bother writing it as a headless unit test.)

Spoiler:

Code: Select all

diff --git a/data/scenario-test.cfg b/data/scenario-test.cfg
index 18af4e47c66..e9bf9d67df6 100644
--- a/data/scenario-test.cfg
+++ b/data/scenario-test.cfg
@@ -38,13 +38,12 @@ Xu, Xu, Qxu, Qxu, Ql, Ql, Ql, Xu, Xu, Xu, Xu, Xu, Xu, Xu, Xu, Gg, Gg, Gg, Gg, Gg
     turns=unlimited
     id=test
 
-    {DAWN}
     {MORNING}
     {AFTERNOON}
     {DUSK}
     {FIRST_WATCH}
     {SECOND_WATCH}
-    current_time=3
+    {DAWN}
 
     victory_music="loyalists.ogg,underground.ogg,revelation.ogg"
     defeat_music="sad.ogg"
@@ -248,37 +247,6 @@ Xu, Xu, Qxu, Qxu, Ql, Ql, Ql, Xu, Xu, Xu, Xu, Xu, Xu, Xu, Xu, Gg, Gg, Gg, Gg, Gg
             overlays="wesnoth-icon.png~CROP(10,10,20,20)"
         [/unit]
         [unit]
-            x,y=14,6
-            type=Footpad
-            generate_name=yes
-            [modifications]
-                [trait]
-                    id=feral
-                    availability="musthave"
-                    male_name="feral"
-                    female_name="female^feral"
-                    description="Receive at most 40% defense in villages"
-                    [effect]
-                        apply_to=defense
-                        replace=yes
-                        [defense]
-                            village=-60
-                        [/defense]
-                    [/effect]
-                [/trait]
-                [trait]
-                    id=overpowered
-                    name="overpowered"
-                    availability=musthave
-                    [effect]
-                        apply_to=attack
-                        increase_attacks=1
-                        increase_damage=1
-                    [/effect]
-                [/trait]
-            [/modifications]
-        [/unit]
-        [unit]
             x,y=12,8
             type="Thief"
             generate_name=yes
@@ -1436,6 +1404,21 @@ My best advancement costs $next_cost gold and I’m $experience|% there."
 
     [event]
         name=start
+#define X x
+        {NOTRAIT_UNIT 1 (Drake Clasher) {x} 5}
+        {NOTRAIT_UNIT 2 (Dwarvish Berserker) {x} 6}
+#enddef
+        [for]
+            start=1
+            end=1000
+            [do]
+                [label]
+                    x,y=14,6
+                    text="Iteration $i"
+                [/label]
+                {X 14}
+            [/do]
+        [/for]
 
         [gold]
             side=1

[beetlenaut]

how often one can see a 3.8% chance come true

It doesn't sound like you've been playing very long, because that kind of thing happens all the time. You should expect to see something similar in every scenario or multiplayer match. (It's only 1/26 after all, not something like 1/1000.) Personally, I wouldn't risk a 3.8% chance of death with a leader or hero--it's too high.

Damage calculations are certainly not aware of eventbased abilities, and I suspect they fail with some filter cases too.

But not with a mainline ability like berserk.

[mxb2001]

[quote=beetlenaut post_id=638372 time=1549837398 user_id=107794]
[quote=mxb2001 post_id=638313 time=1549649228 user_id=175961]
how often one can see a 3.8% chance come true
[/quote]It doesn't sound like you've been playing very long, because that kind of thing happens all the time. You should [i]expect[/i] to see something similar in every scenario or multiplayer match. (It's only 1/26 after all, not something like 1/1000.) Personally, I wouldn't risk a 3.8% chance of death with a leader or hero--it's too high.
[quote=Ravana post_id=638315 time=1549650264 user_id=129456]
Damage calculations are certainly not aware of eventbased abilities, and I suspect they fail with some filter cases too.
[/quote]But not with a mainline ability like berserk.
[/quote]

5 weeks playing :) and I guess my post is half frustration and half tongue in cheek. I've been wargaming since the 70's so I've had a long time to learn to hate bad die rolls. No, you never do come to accept them... <grin>