Worst combat roll?

[DABegley]

I have had a recent series of really out there combat rolls from the RNG, so what are your worst rolls? Here are a few of my recent combat rolls.

Losing 5 straight 81%+ rolls in a single attack was annoying but I still took the city so not that bad an outcome.

C1 Cuirassier attacked my C3, D1 Rifleman in a city, on a hill behind a river. The Rifleman dieing wasn't that far out there but the Cuirassier not getting a single iota of damage counts as one of my worst rolls.

2 Barb archers attacking 2 Praetorians (1 a Combat2) in a hill city behind a river, yup both died for me to lose the city and earning a place in my worst roll list. I didn't attack them because they were on a forest-hill hex and were behind a river. I was waiting for them to move to an open hex for the 99+ attack, foolish me.

Now none of these rolls cost me much, took city, held city and retook city so they aren't "military disasters" but they are bad rolls.

 

[Kawalimus]

I throw things across the room any time I don't do damage when I feel like I should.

I don't know my exact worse roll but I know I get really angry when the odds are like 75%+ in my favor and my opponent takes no damage.

 

[Admiral Armada]

I saw a screenshot a while ago of the combat log, the odds were 100%, and it was a loss

 

[say1988]

Yep, it is because of rounding. Even if it is 99.9999% chance to win, you will lose in in a million rounds, and think how many millions of battles are fought just by the people on this site. The odds displayed while attacking were changed from rounding to 100% to >99.9% (or whatever the decimal) because of this. There is never a 100% certain battle in this game, even a Modern Armour with every promotion will have a chance to lose versus a crippled warrior with no promotions or defensive bonuses, granted the odds would be virtually non-existent.

 

[Yonez]

98.2% was the worst loss I've seen.

As far as the worst string of results I've seen was, a barbarian galley took down four of my galleys before I could get rid of it with a fifth. Ridiculous.

 

[JP1]

I've lost a 98.X% once or twice, and with the number of people posting on this site, I wouldn't be surprised if some people had lost some 99.8% and 99.9% battles.

 

[Scoottr]

I've lost tons of 99.x combats. Happens very frequently really. Wish I could see what the odds are when AI takes out my highly promoted unit with a spearman....

Suppose if I tried more fights where my odds were 1.0% or less I might feel differently because supposedly I should be winning those as frequently as I lose the other.

 

[say1988]

Wish I could see what the odds are when AI takes out my highly promoted unit with a spearman....

Check the combat logs afterwards.

 

[Otaku-sama]

I think my worst was 89%. It was Prat versus Barb Archer. I don't know what happened. Then again, I rarely care about losing units because I have the Stack Attack option on, so I don't see the battles, just a swarm of victory/defeat horns and a result. I also enjoy SoDs, so a loss of an experienced unit is unheard to me.

 

[shulec]

Most of us tend to not see the flip side, because we are not often silly enough or desperate enough to attack with terrible odds.

What really peeves me is when my missionaries fail. I have had five in a row fail in what was a hugely critical situation.

 

[PieceOfMind]

Yep, it is because of rounding. Even if it is 99.9999% chance to win, you will lose in in a million rounds, and think how many millions of battles are fought just by the people on this site. The odds displayed while attacking were changed from rounding to 100% to >99.9% (or whatever the decimal) because of this. There is never a 100% certain battle in this game, even a Modern Armour with every promotion will have a chance to lose versus a crippled warrior with no promotions or defensive bonuses, granted the odds would be virtually non-existent.

I could have posted in this in the top myths thread.

Actually it is possible to have 100% odds. You have to be the attacker and be 1000 times more powerful than the defender (e.g. modern armor vs. injured warrior - I know, it's unrealistic!). Essentially it's a consequence of the combat die having only 1000 sides and the attacker being the benficiary of the rounding error.

 

[The Rook]

I've lost plenty of fights with 99.9% winning odds. My worst ever combat luck (or at least the worst I remember) was losing two war elephants in a row to unpromoted barbarian axes (barb uprising bad luck) with 99.4% and 99.1% winning chances respectively! So what are the chances of that, about 1 in 18500? Even worse, this was on the eve of a war against Monty, AND we were semi isolated.

 

[mechaerik]

Losing not one, but 2 SoTL against a wounded Frigate.

 

[Ramesses]

I could have posted in this in the top myths thread.

Actually it is possible to have 100% odds. You have to be the attacker and be 1000 times more powerful than the defender (e.g. modern armor vs. injured warrior - I know, it's unrealistic!). Essentially it's a consequence of the combat die having only 1000 sides and the attacker being the benficiary of the rounding error.

Shouldn't the defender be the beneficiary of the rounding error?

 

[PieceOfMind]

Shouldn't the defender be the beneficiary of the rounding error?

No. What makes you think so?

 

[Infantry#14]

I once lost defending a full health infantry on a hill against a cavalry. The combat log says i have 99.99% chance of winning and the ai cavalry was unscratch.

 

[Ramesses]

No. What makes you think so?

I once lost defending a full health infantry on a hill against a cavalry. The combat log says i have 99.99% chance of winning and the ai cavalry was unscratch.

This is why.

 

[say1988]

No, according to PieceOfMind, if you have an advantage of 1000 times on each role, then it is guaranteed. But each battle consists of several roles. So, if you had a 1 in a thousand chance of winning each round, needing two rounds for victory, and the enemy needed to win just one, then you would have 1 in 1 000 000 odds of winning the battle.

Keep in mind that with both units full strength and no other bonuses even Modern Armour vs Scout is just 40:1, so getting 1000:1 would be rare

 

[PieceOfMind]

I once lost defending a full health infantry on a hill against a cavalry. The combat log says i have 99.99% chance of winning and the ai cavalry was unscratch.

That's strange, considering the combat log only lists combat odds to a single decimal place of a percent...

No, according to PieceOfMind, if you have an advantage of 1000 times on each role, then it is guaranteed. But each battle consists of several roles. So, if you had a 1 in a thousand chance of winning each round, needing two rounds for victory, and the enemy needed to win just one, then you would have 1 in 1 000 000 odds of winning the battle.

Keep in mind that with both units full strength and no other bonuses even Modern Armour vs Scout is just 40:1, so getting 1000:1 would be rare

QFT.

It is very rare to have the 1000 times strength advantage. Combat odds of almost 100% (even 99.99999% if it could be displayed) do not necessarily imply that 1000 times factor has been reached.

In other words, I am talking about rounding error on a number that is not actually visible to the player (though it can be calculated very easily as roundup(1000*A/(A+D)) where A and D are the modified strengths). The 99.9% odds that are shown are a different thing altogether.

I might add, losing a unit with the enemy unscratched when you had 99.9% odds is very rare.

Also, paradoxically, in general an attacker with 0.1% odds of success is more likely to get away unscathed if he is injured than one that is is not injured. If that's confusing, think about why two attackers - one healthy and one injured - would both have 0.1% success odds. If the cavalry in the above example was already very injured, the result is less surprising than if it were a full health cavalry.

 

[PieceOfMind]

I once lost defending a full health infantry on a hill against a cavalry. The combat log says i have 99.99% chance of winning and the ai cavalry was unscratch.

Further to my previous post...

If your infantry had 99.9% chance of winning then that already tells you the cavalry was injured. When attacking units are already so injured that one hit would kill them, it's fallacious to think if they kill the defender it was exceptional if they did it unscratched. Maybe this is obvious but if it would only take one hit to kill the attacker then the odds of the unit winning the battle are the same as winning unscratched.

Now, as for your particular example, another problem is that cavalry always have a minimum 30% withdraw chance. The combat log can report numbers like 0.0% or 100.0% (unlike the hover text for odds) so let's assume the combat log said 0.0% in your example.

I was able to reproduce a similar battle:

In my example here, to kill the infantry without getting hurt the cavalry needs to land 8 hits each having 25.5% chance of hitting, for it to kill the infantry. 0.255^8 = ~ 1/56000. A 1 in 56000 chance is pretty unlucky but I suspect the combat log did not really read 0.0% but instead 0.1%...

Now let's assume you meant to say 99.9% originally and so we can assume the combat log reported 0.1% odds. Here's how it might have looked.

This battle will show up as 0.1% in the combat log as shown.

Now we need 8 hits at 29.6% per round. 0.296^8 = ~1/17000. Honestly I think that's getting pretty close to what you observed. You were pretty unlucky to witness a 1 in 17000 event, but the odds of actually losing that infantry could have been around about 0.14%.