Help - Attack order in battles

[varangy]

Help me figure this out: how does the attack order affect the battle outcome? How can I achieve the best outcome?

Take a look at the following 2 pictures. It is the same battle just with different order of which unit attacks first, second, etc.

[Soar_Slitherine]

The game simulates a round of combat to determine the values used for estimating the odds, so I believe it's just the result of the random number generation during simulated combat that changed here. You can directly see in the screenshots which factors changed, and none of them have anything whatsoever to do with the order the units are added in.

[scottrossi]

i realize it probably is RNGesus playing games like Soar_Slitherine says, but I have found more often than not, it seems that spacing out mechanized units with infantry tends to get the odds higher, so I do much what you do in the 2nd pic. 2-3 buggies, tanks, etc and then 2-3 infantry (foot and motorized)

[zgrssd]

There is some variation within the odds. This is just a bug with the prediciton and does not indicate performance.

Some relevant points:
- concentric attack. Attacking from more direcitons is better. It can give you a bonus of up to +200%. Absolutely nesseary if the enemy is entrenched. However the bonus is halved if you use units that are not all under the same OHQ (or even if you throw in the OHQ itself).
It is better to make two well coordinated attacks, then one poorly and overstacked one. Or even sending in one unit, while the other defends.
Concentric attack also increases the stacking limit considerably.
- Armor. Small arms count as 20mm. But even a simple buggy can easily carry 100mm armor, applying a -80% penalty to small arms fire. While it is primarily a defensive bonus, not receiving damage allows your buggies to attack combat turn after combat turn, strategic turn after strategic turn with minimal readiness losses until the enemy is ground to dust
- leave them a escape. Fully encirceling the enemy makes ordered retreat impossible and panic retreat less likely. It is a classical trick of the mongols to leave the enemy a escape route, so they break quicker. This looks like a minor and their forces despawn the turn after they lost their city.
- avoid sending in troops with poor readiness. They just take damage quicker, causing you more losses and getting you whole attack to stall out earlier. Often it is better to wait 1 turn for recovery then attacking now.
- Infantry is generally poor in attack. They just are cheap enough to replace (from the IP side, manpower is another question) that you can afford to use them

Your primary attack froce are the purple Formation. Split them over as many directions as possible. Then wait for decent readiness - if they try a breakout attack, even better!
Put 2 of those independant buggies under the same OHQ. If you need reinfrocements and can not afford to build them, disband the 3rd unit.
Asuming those are a minor, consider leaving them a escape hex. One of those fields, not the forrests.
You should not need to shell them with artillery. They are only enemy militia level troops and artillery ruines the buildings and population that you want to capture.

[laagamer]

The order they're placed in has nothing to do with attack order or victory ratio, as I understand it.

Page 293 of the manual states that before the first turn, subunits are assigned initiative scores between 0 and the particular units initiative score, which I don't believe is visible.

The RNG of this inevitably makes it impossible to determine exact attack order.

Though, of course, there are still front line and backline units i.e. infantry vs artillery. Frontline units needing to breakthrough to attack backline units.

Basically, the RNG provides a bit of randomness to every combat encounter, much like real life.

[BlueTemplar]

ORIGINAL: Soar_Slitherine

The game simulates a round of combat to determine the values used for estimating the odds, so I believe it's just the result of the random number generation during simulated combat that changed here. You can directly see in the screenshots which factors changed, and none of them have anything whatsoever to do with the order the units are added in.

I'm pretty sure that the odds display doesn't simulate a full combat round ?
Actually it *does* change with the order of units.
I don't buy that it's a question of randomness, because you get the same odds with units in the same order.
Or then it's random with a fixed seed ?

Also, can someone explain to me why Commanders and experience would give *negative* bonuses ?

[BlueTemplar]

.

[BlueTemplar]

Yeah, so as I expected, there seems to be no combat simulation going on :
- For vehicles (including walkers) the (unmodified) Offensive Power just seems to be the sum of Combat values of all the troops (subunits).
- For APCs with infantry it's the sum divided by √2.
- For just infantry it's divided by 2.

[Soar_Slitherine]

ORIGINAL: Vic

Those odds are just an estimation guys. And it is baded on simulating 1 combat round of battle. Changing the order of the units could result in slightly different results in that simulation since a lot of random rolls are involved and something like the butterfly effect might occur.

^

[BlueTemplar]

So I thought that "both" was obviously impossible, but thanks to this new very similar thread I'm rethinking this :

Maybe the modifiers that *are* randomly rolled in combat are going to be *also* displayed randomly in the odds screen, rather than their expected value ??
EDIT : And then the pseudo-random seed indeed makes it so that you still get the same result when the units are ordered the same way.
This would make sense for the (up to) 4 leader skill modifiers (but still doesn't explain how you can get negative values).

Also, I'm suspecting some rounding issues, like in my screenshots for the defender "readiness modifier on attack values" being 1% off - note that "modified defensive power" doesn't change ! (though that itself could be a rounding error ?)

Still, this doesn't explain why the "max attacked (attack saturation) modifier for attack values" would change ?? (-7% to -10% in varangy's screenshots)
Ditto for "entrenchment modifier of" (+53% to +58% in varangy's screenshots) ??
Ditto for "readiness modifier on attack values" (-32% to -28% in varangy's screenshots) ??

[nukkxx5058]

It's just that the way the odds are estimated is just conceptually boggus (refereing to the comment of the dev about the way odds are estimated). You cannot do a single combat simulation to estimate the odds. One single round is far too dependent on a single dice roll. We need an average, we need multiple simulations to achieve a convergence to the true value. As a results, odds are near to useless as they don't give you even a rough idea of the true probability of success/failure of your attack. The problem is that I can't find a convenient way to estimate, at least even roughly the battle odds by looking at the other indicators (they are too many factors to take into account) ... This is a major issue in my opinion. Even if the odds indicate 25:1 you cannot click on the start battle button with confidence. I've lost so many battles with similar odds, far more than the implied probability suggests. In reality a 25:1 odd (for example) means that the implied probability is 3.8% chance of losing and 96.2% of winning. With such odds, losing should be a rare event (ie. 3.8% probability). It's far from being true in SE.

This is why in the other thread about the same topic (I didn't noticed this one existed before posting) I suggested to perform a Monte Carlo simulation instead of a single round. Of course, this would sevrely increase the odds calculation time, but at least the odds would be meaningful. And of course it all depends on the number of simulations performed. Lower number leading to a larger standard deviation of the result but faster calculation, and vice versa.

-----
In my example with 25:1 odd, 3.8% = 1/(25+1) - ie you should lose only once in 26 battles, statistically. That is what 25:1 odds means.
100:1 odds means that you should lose once in 101 battles, in average. ie with a probability of 0.99%, less than 1% of the time.
etc.

[JeanleChauve]

These are not probabilities, but an estimate of your base attack power modified by modifiers versus enemy base attack power modified by modifiers.

[BlueTemplar]

I'm not sure that "base attack power" is a good term either, surely the Combat value also takes into account troop "HP" ??

[Soar_Slitherine]

It does take HP into account, as some of the modifiers listed in the estimation are explicitly hitpoint-based.

[JeanleChauve]

I'm not sure that "base attack power" is a good term either, surely the Combat value also takes into account troop "HP" ??

OffensiveDefensiveOower.png (248.11 KiB) Viewed 2175 times

You're right, I didn't have the game open and I didn't use the right terms. It is actually the offensive power for the attacker and the defensive power for the defender.
The same unit will always have the same offensive power value for any combat and at any time or location.
In the case of the IV rover, it will be 5*275 = 1375
Unfortunately, I have not yet discovered how this value is calculated.
I know that the type of weapon comes into play, the number of attacks per turn of combat as well as the number of subunits, probably the soft / hard attack values and perhaps the combat value that we can find in the model design, for which I couldn't find the calculation either.
What is certain is that it is not random

[JeanleChauve]

It does take HP into account, as some of the modifiers listed in the estimation are explicitly hitpoint-based.

Do you know the math for offensive power?

[BlueTemplar]

It does take HP into account, as some of the modifiers listed in the estimation are explicitly hitpoint-based.

I'm specifically talking about the value *before* the displayed modifiers.

It is actually the offensive power for the attacker and the defensive power for the defender.

Yet again, this term is (most likely) misleading - which I guess is not new for ShEmp (consider "Defense" and "HP"...)

Do you know the math for offensive power?

We all do, I've figured it out more than a year ago, and posted in this very thread, why do people keep ignoring it ?!? :

Yeah, so as I expected, there seems to be no combat simulation going on :
- For vehicles (including walkers) the (unmodified) Offensive Power just seems to be the sum of Combat values of all the troops (subunits).
- For APCs with infantry it's the sum divided by √2.
- For just infantry it's divided by 2.

(I'm not 100% sure of course, as I barely had the time to test it before I stopped playing ShEmp, so more testing appreciated - especially on the defensive side, which might be different, but sadly I'm not aware for a way in ShEmp to load a game as a non-human-player Major ?)

[Soar_Slitherine]

Hmm, it does appear to be based on the 'Combat value' listed under MNG -> Models. However, I am seeing a multiplier of around 0.71, rather than 0.5, to the value on my laser Infantry, while Quad MGs do get 0.5. Which makes sense since Infantry retain a greater proportion of their firepower on the offense compared to MGs.

[BlueTemplar]

However, I am seeing a multiplier of around 0.71

1/√2 = 0.7071...

(In my case, IIRC when I said "infantry", I meant MGs, and didn't check for basic infantry...)

----

Ah, interesting, I guess I might have been wrong about the terms "Offensive/Defensive" Power being improper :

I had assumed, (because of where it's shown), that the aggregated Combat value already took into account the model modifiers to the combat values, such as :
1.) Halved Offensive Attack ("Soft and Hard Attack") for Foot/Immobile units (including MGs)
2.) Halved Offensive Attack ("Soft and Hard Attack") and Hard Defensive Attack ("Hard Defense") for MGs ?
3.) Halved Hard Offensive Attack and Hard Defensive Attack ("Hard Attack" and "Hard Defense") for small arms ??
(in case it's unclear, they all stack, resulting in a 1/8 modifier for MG's Hard Offensive Attack ("Hard Attack"))

But it looks like that the calculation of the "Combat" value might ignore those modifiers, and they might be applied during the calculation step "Combat" => "Power" ?
(Which would be good, meaning the odds calculation is less dumb that I was thinking.)

So, I am guessing that in an Offensive Attack situation :
- basic infantry gets a √2 divisor because of 1.)
- MGs get an extra √2 divisor because of 2.), resulting in a total divisor of 2

Ideally, since 3.) depends on enemy unit composition, since it only applies against units with Vehicle ("Hard") Armor, it should show up in the odds modifiers... but AFAIK it doesn't !
(Is it already integrated in the "Combat" value itself, or is it completely ignored ?)

Which makes sense since Infantry retain a greater proportion of their firepower on the offense compared to MGs.

Yeah, and also on Hard Defense, maybe that this has an effect on the "Combat" value could be tested in a MP game ?
(Though it might also have the same fate as 3.) ?)

[zgrssd]

Those predictions are a mess, and you should not overvalue any small difference. We are honestly not even sure how they work Some general rules are:
1. Attack from as many directions as possible. This gives you a Multiple Direction bonus of up to 200%. It also increases Stack Points, so more units can attack without penalty.
2. Try to only attack with 1 OHQ. Once you include mixed units, the bonus from point 1 is halved.
3. Consider leaving the enemy a escape route. If they have nowhere to run, they can not orderly retreat and the morale for panic retreat (rout) is doubled. But leaving a hex (and thus attack direction) open does mean you won't attack from that direction - so it is a tradeoff.
4. The prediction massively undervalues the Callibre Calculation. Small arms only count as 20mm, so even 100mm armor can give a massive penalty. Light tanks with Howitzers are best, but even a buggy can carry substantial armor.
5. Don't attack with depleted units, or those that lack have poor supplies and poor organisation.

Regarding this situation:
The pure infantry nature of the enemy forces.
The lack of Ruins on the City Hex.
All indicates this is a minor faction.

The trick of leave a escape route works best against them, as their army despawns the turn after loosing their last city.
They are also particularly vulnerable to armor, as they got few or no AT weapons.