Odds numbers, combat outcomes, calculations

Ask, comment, read.
cnash
Second Lieutenant
Posts: 9
Joined: Sun Jan 04, 2015 3:46 am

Odds numbers, combat outcomes, calculations

Unread postby cnash » Sun Jan 25, 2015 10:23 am

When the game looks at odds numbers to determine combat outcomes, does it refer to the integer values that are displayed on the info sheet, or does it refer to a floating-point (or somehow else continuous) value that's hidden from the user? That is, does a unit that's just shy of making an odds number of +3 have a different outcome profile to a unit that just barely made it into the +2 slot? Likewise, are there partial armor shifts? If not, what's the point of cavalry units having an armor value of one: even with all seven steps active, the unit can never crack the 10-armor threshhold for a +1 armor shift.

Another question: in the manual, there's a chart showing likely combat outcomes: for a +1 odds number, those are 1 attacker loss, 1 defender loss, 3 defender suppressed. But the defender suffers either the loss or the suppression. How is it determined which happens? I ask because in some situations, it's preferable to suppress two steps than to kill one (I'm looking at you, Cottbus, and your defender that can only maybe be dislodged by five armor divisions).

Also, do I understand correctly that specialist-step armor bonuses with an asterisk (like Stug III: +1* armor) are anti-tank, so they only apply to reducing the enemy's armor shift? It is not clear in the documentation that this is what the asterisk signifies, and not all asterisk-bonused specialist step descriptions specify "anti-tank."

User avatar
HolyDeath
Captain
Posts: 47
Joined: Mon Jul 22, 2013 4:25 pm

Re: Odds numbers, combat outcomes, calculations

Unread postby HolyDeath » Wed Jan 28, 2015 5:41 pm

cnash wrote:When the game looks at odds numbers to determine combat outcomes, does it refer to the integer values that are displayed on the info sheet, or does it refer to a floating-point (or somehow else continuous) value that's hidden from the user? That is, does a unit that's just shy of making an odds number of +3 have a different outcome profile to a unit that just barely made it into the +2 slot?

I will quote from manual: "Attacker and defender losses are looked up in the combat results table using the final odds number as displayed on the combat sheet. Note that, in actual combat, this number is slightly randomized on each lookup to simulate the inherent uncertainty of real life combat."

To me it means that combat losses odds (NOT odds themselves) are randomized. If you got odds +3, then you can get +3, or +2 or +4 and roll results for combat losses from one of them (so attacker losses 0, defender losses 1-2 OR defender has 3 suppression).

Additionally, judging from page 33, even a slight difference can yield different results - as you can see the Soviets are losing only 4 men, instead of 5 on prediction screen, although combat sheet suggests that at such odds they should take 5 losses. My take is that's so because the Germans have 23 to 26. They lack 3 points to an ideal tie, so prediction screen is taking that into an account. It means values are not absolute and take numbers into account more than it's suggested.

I hope this answers your questions.

Likewise, are there partial armor shifts? If not, what's the point of cavalry units having an armor value of one: even with all seven steps active, the unit can never crack the 10-armor threshhold for a +1 armor shift.

Good question.

Since totals are divided by 10 I suppose cavalry does not get armor shift. But... what if we give that cavalry a specialist step that grants +1 armor (like T-34)? That would be 2 x 7 = 14 / 10 = 1. With KV-1 it's +2 armor: 3 x 7 = 21 / 10 = 2. Infantry (starting with armor 0) won't get that +1 shock value to odds when you give them T-34 (7 x 1 = 7 / 10 = 0). An infantry with KV-1? 7 x 2 = 14 / 10 = 1. You can see the potential.

Another question: in the manual, there's a chart showing likely combat outcomes: for a +1 odds number, those are 1 attacker loss, 1 defender loss, 3 defender suppressed. But the defender suffers either the loss or the suppression. How is it determined which happens?

It depends on the roll of a digital dice. You'll get one out of two results given the odds. You can't determine which one it will be.

Also, do I understand correctly that specialist-step armor bonuses with an asterisk (like Stug III: +1* armor) are anti-tank, so they only apply to reducing the enemy's armor shift? It is not clear in the documentation that this is what the asterisk signifies, and not all asterisk-bonused specialist step descriptions specify "anti-tank."

It means that the armour shift only applies on defense (but against all units, not only armored). In case of 88mm FlaK it applies only against armored and only on defense.

cnash
Second Lieutenant
Posts: 9
Joined: Sun Jan 04, 2015 3:46 am

Re: Odds numbers, combat outcomes, calculations

Unread postby cnash » Thu Jan 29, 2015 3:12 am

Okay, so, suppose I'm the Axis, and I need to attack a soviet tank corps (4/3/5+1/6), with four active and no suppressed steps. It's in a forest, so there's no armor shift, but it's not entrenched or veteran. No specialist step. The point is, its defense combat total is 12.

I have two panzer divisions (10/5/5+3/6), both veteran, one with three active steps, the other with four. Their attack combat totals are 30 and 40, respectively.

If I apply the odds-number formula from the manual, (3/ln3)*(ln(attack combat total)-ln(defense combat total)), I'll get an odds number of 2.502 for the three-step tank, and 3.288 for the four-step. Both of those round to +3, so that's the number that's displayed on the combat info sheet.

What I want to know is, do those two units each have the same odds of dislodging or damaging the soviet tank? Or does my four-step panzer have better chances, even though the info sheet shows the same prediction?

User avatar
HolyDeath
Captain
Posts: 47
Joined: Mon Jul 22, 2013 4:25 pm

Re: Odds numbers, combat outcomes, calculations

Unread postby HolyDeath » Thu Jan 29, 2015 7:07 am

Technically both divisions have odds +3, but odds numbers for combat losses are slightly randomized, so the 4-step panzer division should have better chances of the odds staying at +3 (although unlikely reaching +4) and inflicting an extra off-sheet losses on the enemy (due to ratio), because it has 3,3 while 3-step panzer division has 2,5 (and it's much more likely to drop below +3, to +2 odds).

cnash
Second Lieutenant
Posts: 9
Joined: Sun Jan 04, 2015 3:46 am

Re: Odds numbers, combat outcomes, calculations

Unread postby cnash » Thu Jan 29, 2015 8:44 pm

So you think the program calculates the odds number (as a float), applies a random modifier, and then rounds the result to the nearest integer and looks up the results on a chart. And, importantly, that it doesn't round the odds number off until after applying the random modifier. The odds number displayed in the info sheet is, then, the number most likely to apply after randomizing.

User avatar
HolyDeath
Captain
Posts: 47
Joined: Mon Jul 22, 2013 4:25 pm

Re: Odds numbers, combat outcomes, calculations

Unread postby HolyDeath » Fri Jan 30, 2015 9:44 am

Here is how I see it:

Before an attack:
The game calculates ratio of forces and modifiers (entrenched, experienced, etc.) into odds and shows the expected results in-game for you.

After an attack:
When you actually attack it applies a random modifier off-screen (we don't know how slight it is, nor we know in what direction) and re-calculates the odds without you knowing them. Then it looks into the combat losses chart and applies a result.

However, I can be wrong. It's only my interpretation of rules that are in the manual.

cnash
Second Lieutenant
Posts: 9
Joined: Sun Jan 04, 2015 3:46 am

Re: Odds numbers, combat outcomes, calculations

Unread postby cnash » Sat Jan 31, 2015 4:57 am

No, that makes a lot of sense. And it lines up better with the manual than what I'd been imagining, which was [a clumsy system that won't benefit anyone by being typed out here].

What I get from this is that it's not especially useful to finesse my attacks by killing or suppressing one or two steps to improve the odds for the second attacker. Even if the odds number shown in the combat predictor changes, because a +2.4 (rounds to +2) and a +2.6 (rounds to +3) are actually pretty similar. (But for a final assault, it's still best to start with whatever unit has odds number -1 or 0, because two or three steps suppressed is more useful than one killed.)

User avatar
Tomislav Uzelac
2x2 Games
Posts: 2211
Joined: Mon Apr 04, 2011 11:24 pm
Location: Zagreb, Croatia

Re: Odds numbers, combat outcomes, calculations

Unread postby Tomislav Uzelac » Sun Feb 01, 2015 10:08 am

Hi guys, I wet and looked this up in the source code, here is how it works precisely:

1. raw odds (attacker/defender ratio, before terrain etc.) are calculated, and then immediately rounded. Distinction between +2.501 and +3.499 gets thrown away at this point.

2. odds (as displayed in the combat sheet) are then calculated by adding weather, terrain, entrenchment etc. That number is again an integer.

3. defender losses index (used to display losses in the combat sheet) is calculated by taking odds and adding any specific shifts (green unit, NKVD, etc.)

4. in actual combat, a random modifier is applied to this "defender losses" index, which yields actual defender losses (KIA).

5. in actual combat, if defender got 0 KIA and hasn't retreated, it incurs suppression. This is looked up using odds (no randomization).

Example: suppose there's an attack with odds 0. The combat sheet predicts 1 KIA for the defender. However, due to randomization, defender losses turn out to be 0 KIA in actual combat. In that case, defender incurs suppression which is looked up using odds 0, which comes out to 3 suppressed.