What's the probability of disaster occurs?

[melm]

I just played one game that have four disasters occurred during that 24-turn game, either in advantage, slightly advantage or slightly disadvantage. Among the four, two disasters are consecutive. It seems that it happens quite often. In my other games, you may not meet that many but one or two in 24 turns.
I'd like know that how often you will get the disaster result once one loses the fight. Plus how often miracle “hold firm” may happen if one loses the fight.

[Barrold713]

I just played one game that have four disasters occurred during that 24-turn game, either in advantage, slightly advantage or slightly disadvantage. Among the four, two disasters are consecutive. It seems that it happens quite often. In my other games, you may not meet that many but one or two in 24 turns.
I'd like know that how often you will get the disaster result once one loses the fight. Plus how often miracle “hold firm” may happen if one loses the fight.

My entire FoGDL season 1 with the Krappodockians was an improbable disaster!

It doesn't seem to happen often of course for having the quality of infrequency, but who cannot remember a turn or two from a match that did not crush all hopes for a dignified loss let alone a piddling victory?

I saw something today in my Fall of the West tournament where my opponent's Roman cavalry fragmented from good order while falling back. There was nothing I noted that would have been a flanking threat after the move but it was an easy break at that point. The percentages giveth and and the percentages taketh away.

[rbodleyscott]

I saw something today in my Fall of the West tournament where my opponent's Roman cavalry fragmented from good order while falling back.

Easily avoided, of course, by not falling back.

[MikeC_81]

I just played one game that have four disasters occurred during that 24-turn game, either in advantage, slightly advantage or slightly disadvantage. Among the four, two disasters are consecutive. It seems that it happens quite often. In my other games, you may not meet that many but one or two in 24 turns.
I'd like know that how often you will get the disaster result once one loses the fight. Plus how often miracle “hold firm” may happen if one loses the fight.

Depends on what you count as a disaster. Impossible to tell you what happened without screenshots.

[Barrold713]

I saw something today in my Fall of the West tournament where my opponent's Roman cavalry fragmented from good order while falling back.

Easily avoided, of course, by not falling back.

Very true. I actually wondered if he did it by mistake. Either way it was a die and learn moment for him.

BDH

[Ludendorf]

I'm not actually sure on the exact statistics, but double drops happen more often than you'd think, and sometimes even with high quality units. It tends to happen when a unit has taken a few too many hits already, but bolts from the blue can and do happen. Wherever possible, it's always good to have a plan in place for what to do if one of your units suddenly panics.

[melm]

I just played one game that have four disasters occurred during that 24-turn game, either in advantage, slightly advantage or slightly disadvantage. Among the four, two disasters are consecutive. It seems that it happens quite often. In my other games, you may not meet that many but one or two in 24 turns.
I'd like know that how often you will get the disaster result once one loses the fight. Plus how often miracle “hold firm” may happen if one loses the fight.

Depends on what you count as a disaster. Impossible to tell you what happened without screenshots.

It's just "disaster" written in the combat log window. In "disaster" situation, usually the morale drops two levels from steady to fragmented.

[melm]

I'm not actually sure on the exact statistics, but double drops happen more often than you'd think, and sometimes even with high quality units. It tends to happen when a unit has taken a few too many hits already, but bolts from the blue can and do happen. Wherever possible, it's always good to have a plan in place for what to do if one of your units suddenly panics.

Does the program decides the disaster result first then generate a high random casualty, or the program generates a high casualty from combat first then determine whether it is disaster or not?

[Kabill]

Does the program decides the disaster result first then generate a high random casualty, or the program generates a high casualty from combat first then determine whether it is disaster or not?

It generates casualties first. If, after that, one unit fared significantly worse than the other, the game rolls a cohesion check for the loser.

The cohesion check is 2d6 (i.e. 2 six-sided dice), needing 6 or more to pass, but with various modifiers (most of which apply down). However, if the unit lost by a particularly large amount, a final score of 2 or less (I think?) results in a double-drop. Cohesion checks from other sources (e.g. falling back or nearby routed units) can also cause double-drops, with the exception (I think) of missile fire which never can.

Modifiers notwithstanding, then, double-drops should happen in no more than 1 in 32 cohesion checks. But the application of modifiers makes them much more common (e.g. if you're rolling with a -3 modifier, which is not uncommon, the chance of a double-drop is now 10 in 36 or almost 1 in 3).

[melm]

Thanks Kabill. Then the probability may depends how often unit may lose large amount instead of small one to go for cohesion test. Right? Then cohesion test is 2d6 problem with some modifier.

[Kabill]

Yeah, a test is only made if a unit loses a melee combat (you see a red number for casualties). Lots of minor losses from melee combat will not cause a test (this is different for shooting, where casualties from different sources stack over the length of a turn). Also...

Plus how often miracle “hold firm” may happen if one loses the fight.

I mean to reply to this before. A unit can only fail one cohesion test from each source (impact, melee, shooting, flanking) in one turn. Subsequent occasions a unit might be required to test display as a "Holds Firm" result but no further roll has been made. E.g. if a unit loses a melee combat against one enemy unit, passes its cohesion check, and then loses another melee combat against another enemy unit it is fighting, it won't roll again and it will just show "holds firm". You might have known this, but I'd overlooked it for quite a while and is the reason why sometimes it can seem that some units pass their check loads of times in a row (but they're not - they're just using the same result as the first roll).

[rbodleyscott]

Yeah, a test is only made if a unit loses a melee combat (you see a red number for casualties). Lots of minor losses from melee combat will not cause a test (this is different for shooting, where casualties from different sources stack over the length of a turn). Also...

Plus how often miracle “hold firm” may happen if one loses the fight.

I mean to reply to this before. A unit can only fail one cohesion test from each source (impact, melee, shooting, flanking) in one turn. Subsequent occasions a unit might be required to test display as a "Holds Firm" result but no further roll has been made. E.g. if a unit loses a melee combat against one enemy unit, passes its cohesion check, and then loses another melee combat against another enemy unit it is fighting, it won't roll again and it will just show "holds firm". You might have known this, but I'd overlooked it for quite a while and is the reason why sometimes it can seem that some units pass their check loads of times in a row (but they're not - they're just using the same result as the first roll).

Actually they can fail subsequent tests, but only if their previous roll was just enough to barely pass, and then they acquire another negative modifier.

The reason it works like this is because it simulates the tabletop game, where only one cohesion test is taken, after all of the close combats have been adjudicated. However, because of the rather more relaxed sequence of play in the computer version, this isn't really possible, or if it was done at the end of the turn it would lose the immediacy that the current system engenders.

Therefore, to get the same effect as one single test, while in fact taking a test each time the unit loses a combat, a single random roll is used for each unit for all close combats in a turn.

If a separate random roll was used for each such test, the failure rate would be much higher than intended, so the whole cohesion test system would have to be rebalanced accordingly, which would take it even further away from the tabletop rules than it currently is.

The "Held Firm" is something of a Catch 22 for the UI design. If we show it repeatedly, those who have not fully understood the system think the unit is getting lots of lucky rolls. However, if it wasn't shown then people would think the unit wasn't subject to a cohesion test when it lost a combat round. Whereas it is subject to a cohesion test, only the random roll stays the same, but the modifiers may be different from the earlier tests. This can cause a unit to fail a test when it passed a previous one. It can also cause it to drop cohesion again if it previously dropped one level, but the additional modifiers (or more severe combat loss) tip it into the double drop bracket, in which case it will drop one further cohesion level. But the most likely result is that it will continue to hold firm if it held firm once.

Note, however, that a test for a Fragmented unit being charged uses a new random roll each time, so it is well worth charging a Fragmented unit even if it became Fragmented as a result of a close combat this turn, because with -2 modifier for being already Fragmented it is highly likely to break immediately without any combat. However, if it passes the cohesion test for being charged, it will use the existing close combat cohesion test roll if it loses the impact combat, so, assuming it dropped cohesion to Fragmented as a result of losing a close combat this turn, it will hold firm unless it has acquired sufficient additional modifiers to put it into the double-drop bracket.

And if all that sounds complicated, it is because it is. That is why we have the computer to adjudicate it for us.

-----------------------------------------------------

"Disaster" is used in the combat log for a double drop.

[melm]

Thanks Richard. It's pretty detailed explanation.

[melm]

To continue the topic.

We can read the chance of win, draw or loss from the tooltip. Then how about the chance of suffering large casualties? Is it related with the win-draw-loss percentage, that is to say, the high win percentage may have high chance to inflict high casualty to force a cohesion test? Or the casualty is generated from a normal distribution with some mean and deviation?

[Kabill]

To continue the topic.

We can read the chance of win, draw or loss from the tooltip. Then how about the chance of suffering large casualties? Is it related with the win-draw-loss percentage, that is to say, the high win percentage may have high chance to inflict high casualty to force a cohesion test? Or the casualty is generated from a normal distribution with some mean and deviation?

I don't know exactly how casualties are determined, but the win/loss % are estimates based on simulated casualty rolls. I.e. there is no roll to win or to lose; each unit inflicts x casualties on the other and one is considered to have won if the number of casualties suffered by one side is sufficiently higher than those suffered by another. E.g. two units might fight, one causing 12 casualties and another 10, which would be a draw. But another round, one might cause 30 and the other 8, which would be a win for the side which caused the most casualties. So the tool-tip % you see is an indication of the proportion of combats won, drawn and lost out of 1000 simulated rolls (meaning there will be a small amount of error, on average in the region of ~2 percentage points I think)

[rbodleyscott]

The higher the lose chance, the higher the chance that the loss will be a "bad loss".

A "bad loss":

1) Gives an extra -1 CT modifier.
2) Qualifies the loser for a possible double drop if his CT score is low enough.
3) Triggers a break off for infantry vs non-shock infantry or shock infantry who did not originally initiate the combat.
4) Triggers a push back for infantry vs shock infantry who did originally initiate the combat.
5) Triggers a break-off for cavalry vs non-shock cavalry, or shock cavalry who did not originally initiate the combat.

The unit which originally charged is the one that "originally initiated the combat". It has nothing whatsoever to do with whose turn it is or which player did or did not click the Resolve Melee icon.

In short, the higher the lose chance, the worse the overall result is likely to be.

Since the game is primarily a morale-based system, what really matters is Cohesion drops. The actual casualties suffered are relatively unimportant except in their attritional effect on the unit's future combat power, and the possibility of eventually autobreaking. It is possible for a unit to break from Cohesion drops after suffering relatively few casualties.

It is in fact possible for a unit to break before suffering any casualties at all. e.g. Unit sees friendly unit break in an adjacent square, and suffers a double drop CT result, then is charged by the enemy pursuers, fails the CT test, and breaks before contact.

[melm]

To continue the topic.

We can read the chance of win, draw or loss from the tooltip. Then how about the chance of suffering large casualties? Is it related with the win-draw-loss percentage, that is to say, the high win percentage may have high chance to inflict high casualty to force a cohesion test? Or the casualty is generated from a normal distribution with some mean and deviation?

I don't know exactly how casualties are determined, but the win/loss % are estimates based on simulated casualty rolls. I.e. there is no roll to win or to lose; each unit inflicts x casualties on the other and one is considered to have won if the number of casualties suffered by one side is sufficiently higher than those suffered by another. E.g. two units might fight, one causing 12 casualties and another 10, which would be a draw. But another round, one might cause 30 and the other 8, which would be a win for the side which caused the most casualties. So the tool-tip % you see is an indication of the proportion of combats won, drawn and lost out of 1000 simulated rolls (meaning there will be a small amount of error, on average in the region of ~2 percentage points I think)

Interesting. You mean the percentage of win-draw-loss is simulated with samples of combat casualty result(with certain criterion to determine win or not). I thought before that the percentage is estimated from POA with some modifier.

[rbodleyscott]

Interesting. You mean the percentage of win-draw-loss is simulated with samples of combat casualty result(with certain criterion to determine win or not). I thought before that the percentage is estimated from POA with some modifier.

No, the program runs through the actual combat calculation code, then applies the random part 1000 times, and counts the number of wins, draws and losses (using the same criteria as for the actual combat calculation) then divides by 10.

Despite the statistical variation caused by the relative small (!) sample size of 1000, this removes the risk of much greater error that could occur if we tried to work the chances out from the POAs etc. mathematically and got the logic slightly wrong.

Doing it the current way we can be 100% sure that all of the relevant factors are correctly taken into account, and given their correct weightings. We just have to tolerate the small degree of statistical variability.

[Morbio]

This explains why I've noticed different win-draw-loss percentages for a given melee combat! I look to see the chance of winning and get 1 set of predicted results... I go back a few minutes later and see a different set of predicted results even though nothing has changed... I thought I was going mad!

[melm]

I don't really understand the multiple opposing result.

In the snapshot, one is our unit +35% versus the legionary and the other is enemy unit +2% against my warband. In my understanding, multiple opposing for my warband shall have debuff as my troops have to split to face different direction. Thus, the bonus shall be added to the enemy. I really don't understand why my unit got the bonus against the legionary.