The Odds Calculator can't be trusted!

[QiZhe]

Sounds like you're getting RNG-screwed.

The combat isn't as random as you're all making it. This game has RNG like any other game. The Random Numbers are generated with algorithms that are only so complex. If you've ever restarted a load and tried the same thing, you'd notice that the same thing happens (if you don't check the Regenerate Random Seed option). Combat is not truly random, so odds generators will not be perfect.

 

[ruff_hi]

Using a swordsman with raider 2, it states you get +55% when attacking a city.

Is that right? I thought it was +20% for raider 1 and that raider 2 moved that +20% to +25% (ie an additional 5%). Don't get me wrong - I am more than happy to be wrong in this situation as I was always opting for combat 1 instead of raider 2.

 

[BeefontheBone]

Swordsmen DO get +55% against cities - 10 for being swordsmen, 20 for raider I and 25 for raider II. Bonuses are summed and then the total bonus applied to the strength (apart from the ones which are applied as a penalty to the enemy's strength, Cover for instance) leading to a 55% bonus.

 

[DarkSchneider]

I finally got to see the other side of the questionably flawed odds calculator. I was doing a suicide catapult attack against a city, so I decided to see the odds for the attack for fun. It said I had a 1.8% chance of winning, so I attack, and to my amazement, the catapult wins! Something definately seems off about the odds calculator, especially with siege unit battles...

 

[JerichoHill]

Part of the problem is selective memory.

Hi, I'm a statistician.

Given positive or negative occurances, the average person will almost always overestimate the number of negative occurances and underestimate the number of positives that they have had over a time period. What is most powerful to us is that which we remember.

Hence, what I'm saying is that those of you who are saying you're getting more than 2% losses on 98% odds are only remembering the losses with more veracity than your wins.

 

[Rayjr]

I dunno there has always been something about the odds calculator that has bothered me. I have noticed that some units are just more likely to win than others when both have the very same odds. Since I have started using the romans and mass producing the preatorians I feel like they almost always win when the odds are 50% or better. when I was the americans and had similiar odds I felt like my units lost too often. Over all though I think its pretty decent.... Its civ ofcourse you are gonna catch it in the rear.

 

[Kilroy]

One thing that I've been wondering about, with the odds calc, is first strike entirely left out of the calculations. I don't see how it could (easily) be applied to the odds calculation. In case it isn't, that too should be taken into consideration before attacking.

It's not hard to incorporate it into the calculation. It involves maths of some type I assume.

 

[ruff_hi]

The other thing to remember with some units is that they have a 'withdrawal' probability. That is, they lose, but run away to sulk without being killed.

 

[DarkSchneider]

Part of the problem is selective memory.

Hi, I'm a statistician.

Given positive or negative occurances, the average person will almost always overestimate the number of negative occurances and underestimate the number of positives that they have had over a time period. What is most powerful to us is that which we remember.

Hence, what I'm saying is that those of you who are saying you're getting more than 2% losses on 98% odds are only remembering the losses with more veracity than your wins.

Well, maybe the real problem for the 2% odds is sample size, I've fought at most ten battles with those odds, all with collatoral damage units hoping to soften the defense. So the fact that I've already won one is unusual, but as I mentioned in the original post, the odds calculator gives odds of over 100% and is therefore flawed regardless.

 

[JohnnyRico]

I'm quite satisfied with the odds calculator. I haven't experienced more than perhaps two 98% losses in at least more than 1000 battles. If the percentage is around 60-70%, one has to understand that there is quite a risk of losing.
....

Typo? 100=1000?
Otherwise, what you're saying proves the original posters point. Let's see...

(1000-2)*100/1000 = 99.8%

which the calculator returns as 98%, hence broken... by about 1 order of magnitude! (2 in 1000 not the same as 2 in 100).

But anyway I keep noticing an odd thing... if your odds are around 66%, it's the kiss of death - you almost always LOSE. In these cases I alwasy see the following happeneing ( talking about units that have 3 members/soldiers in the animation, like axemen or archers).
Round 1.
your trio kills 2 of his unit soldiers and you lose 1 (perfectly in accordance with the ~66% odds)
Round 2.
His remaining guy walks up to your remaining two soldiers and wacks them one by one. (remmeber the roman Horatii/Curiatii fight? - same concept it seems, which would be cool if implemented on purpose)

Now i know that the units with 2-3 soldiers are not calculated as 2-3 soldiers and are supposed to be just 1 unit regardless of the graphics representation, but something odd is going on here nonetheless.

 

[JerichoHill]

You don't almost always lost at 66% you win 2 out of 3 times. However, units sustain damange, which makes them weaker for the next round.

Bingo.

Look, I could go into the mechanics and all about odds calculations, but I think 3 or 4 people here would get the math.

It's very much spot on. Not 100% accurate, but very close.

PS: I would be happy to talk about odds, but not here. Send me a PM and I can send you the math.

 

[Steve2000]

Part of the problem is selective memory.

Hi, I'm a statistician.

Given positive or negative occurances, the average person will almost always overestimate the number of negative occurances and underestimate the number of positives that they have had over a time period. What is most powerful to us is that which we remember.

Hence, what I'm saying is that those of you who are saying you're getting more than 2% losses on 98% odds are only remembering the losses with more veracity than your wins.

People need to pay more attention to the issue of selective memory.

Everyone who thinks they are getting screwed by the RNG, should start keeping a log of every single battle they fight (What the odds were, and what the results were). Better yet, fire up the worldbuilder and show yourself that with enough battles the results will come close to the odds presented.

EDIT: Also - losing a bunch of battles where you have favorable odds does not make the next battle more likely to win. In other words, say I have three 80% chance of victory battles that I lose. I then fight a fourth 80% battle. What are the odds of me losing that battle? They are still 20% - it doesn't matter how many previous battles I have won or lost, each result is determined independently.

 

[jar2574]

People need to pay more attention to the issue of selective memory.

Everyone who thinks they are getting screwed by the RNG, should start keeping a log of every single battle they fight (What the odds were, and what the results were). Better yet, fire up the worldbuilder and show yourself that with enough battles the results will come close to the odds presented.

Good points. I just shrug when I lose a 98% battle. It happens, but most people forget the 49 times they won before that ugly loss.

 

[Steve2000]

Good points. I just shrug when I lose a 98% battle. It happens, but most people forget the 49 times they won before that ugly loss.

Of course. I scream and yell every time I lose a battle like that too. Everyone has the right to be upset when they lose in unlikely situations, and come to this forum and vent about it. I just think it is important to get some perspective and not claim that the game is flawed.

 

[Mathemagician13]

Part of the problem is selective memory.

Hi, I'm a statistician.

I heard that 87% of all statistics are made up. Oh...wait..

Anyway. I've never seen odds over 100%, so I can't comment on that, but probability is just that...probable. The game would be boring if you won every fight you have more than 50% chance of winning, eh?


Edit: the / to end a quote goes this way. : )

 

[Xenocrates]

I tried an experiment with this; I won an 87% battle and then reloaded and fought a 20% battle instead and also won it. I lost a 20% battle and reloaded and also lost an 87% battle. I tried many times and this kind of result is common!

That lookup table of 'random numbers' makes more difference than I would like. Is there any way to reduce the 'random' influence on the battle? The unpredictability seems to take some of the strategy out of it for me.

 

[Littlewolf]

Guys, be realistic. Don't complain about the combat calculator. The occasional unlikely result is a normal part of probability. Someone wins a jackpot, although the chance is almost zero. So you have almost zero chance of losing that fight, but you still have to lose it from time to time.

Also, in real world, a cavalry can lose against a warrior (the horse stumbles over a rock, breaks his leg and the cavalry guy falls down and breaks his neck, the warrior wins!. It's unlikely, but it probably happened in the history. Or the swordsman gets a heartattack and dies against an archer, the archer wins. The game would be too predictable if you knew for sure that the next battle would be a win. It would practically have an effect of virtual reloading and maxmizing the wins.

 

[Gaizokubanou]

I won't dismiss the possibility of selective memory... however, I still got handful of 99.7% losses and once I even lost three 90%+ in a row. My modern armor on full strength fought against an archer and while I did win, I won with 4 life left.

In Civ4, anything lower than 80% seems like a huge gamble to me. I think the big problem is not that you can lose a 80 - 90% stuff often, but the fact that when you lose it you lose the unit completely. I think it could've been nice if Civ series just dealt away with chance happening and went with straight up consistent results.

 

[GenericKen]

I tried an experiment with this; I won an 87% battle and then reloaded and fought a 20% battle instead and also won it. I lost a 20% battle and reloaded and also lost an 87% battle. I tried many times and this kind of result is common!

That lookup table of 'random numbers' makes more difference than I would like. Is there any way to reduce the 'random' influence on the battle? The unpredictability seems to take some of the strategy out of it for me.

The most reliable way to beat the combat odds is to bring multiple units. The random distribution given the total combat odds for each combat is uniform, not normal, and if you want to achieve a normal distribution around your expected win %, you need to engage in many combats.


There are two issues with the combat calculator at work here, though. Occasionally the calculator will show odds of 100% and 0%, and these are pefectly valid. The display rounds off to the somethingth-decimal place, so these values are to be expected.

However, it is definately evident that there's something buggy with the combat calculator when it shows odds from 104% to 150%. It's quite clear to me, even just doing mental math, that the combat calculator is handling first strike all wrong.