bad great generals

[ohcrapitsnico]

That's interesting combat isn't affected by difficulty level. My only quarrel with the system is that I'm pretty unlucky but if you say that's the case I hope I get lucky with my combat soon.

 

[ohcrapitsnico]

This is mainly the reason why I avoid using Great Generals as warlords. Uber units can still be killed easily. Also, promotions take a long time to get after around 5 or 6 of them, so your units growth will come to a screeching halt at one point or the other.

Ahhh yes now that's a great alternative to the very unpredictable combat(maybe just predictable and bad combat). Now I just need to improve the great instructor and military academy as they're pretty weak considering the difficulty in acquiring warlords and how easily you can surpass the great instructor and the military academy by civics, barracks, stables, west point, heroic epic, leader traits, and so on.

 

[DsevenO]

Yes, you do need hundreds of fights if you want to be accurate. That's just the way statistics work.
Note that you have actually lost 3 fights out of 10. That's not 15 in 50, it could theorically be 15 in 50.
By doing 10 fights only the results will vary too widely to be significant. it would be like throwing a dice 10 times, and calcullating the odds solely based on those 10 throws.

I'm well aware of the finer points of statistical mathematics, but like I said, the layman doesn't need a scientifically accurate statistical analysis of the combat odds to determine whether or not said odds are even in the ballpark of what they purport. I may have only specifically counted 10 encounters, but I can say I paid close attention to the other 40 or 50 similar encounters and the odds just don't seem statistically probable; possible certainly, but highly improbable.

 

[AriochIV]

I don't know whether or not the odds display is actually statistically correct, but I do know that human perception of probability can be extremely wrong. I do notice many times however when my attacking unit has a very small chance of success and manages to win. A 10% chance can happen a lot over the course of a game. Of course it's still irritating when your Warlord has a 99.4% chance of winning and dies.

Since Warlords can be so easily lost, I've started using them more as a focus of the army than as a workhorse combat unit, giving the unit promotions like Medic and Sentry.

 

[Duraska]

Correct me if I'm wrong, but I don't think first strike factors into the combat odds. This means that if you attack a longbowman your odds of winning are actually lower than they appear (because he'll likely damage your troops a little with the first strike).

 

[Mr. Civtastic]

I lose Great Generals all the time, while winning battles with lesser odds. Its real annoying. Now I dont attack with a GG unless the odds are in the 90%. Illl save him in a stack until I absolutely KNOW he will win.

And yes, I feel like I lose more fights when I have the advantage then I did in vanilla civ 4.

 

[DsevenO]

I don't know whether or not the odds display is actually statistically correct, but I do know that human perception of probability can be extremely wrong. I do notice many times however when my attacking unit has a very small chance of success and manages to win. A 10% chance can happen a lot over the course of a game. Of course it's still irritating when your Warlord has a 99.4% chance of winning and dies.

Since Warlords can be so easily lost, I've started using them more as a focus of the army than as a workhorse combat unit, giving the unit promotions like Medic and Sentry.

You make a good point - highly elevated odds probably do skew our perception of how often those odds come out in our favour.

 

[DrewBledsoe]

It isn't fixed in anyone's favour, it IS a true RNG (well as true as an algo to produce a "random" number ever can be), but sometimes it just feels like it's against you.

Quick Example: I was using WB to experiment with different grt general increment settings, and had 30 axes on my side against 30 axes on AI side..all at 50% odds I did a stack attack, and lost 25 out of 30 combats.

If that had happened in a real game, the perfectly natural human response would be to think "there's something wrong, that wasn't random, I'm being picked on".

But it was just a purely random occurance, with a bunch of results going "the wrong way", and if I'd won 25 out of 30 50% combats, I'm sure I'd rationalise it with something like "My brave, tough axemen etc."

No cheating, just random


Edit: Oh yeah, someone mentioned barb combat, well that does change with lvl, on easy lvls you get a decent bonus against barbs, which diminishes with each lvl until you get none.

 

[scipian]

William the Conquerer did quite well though. The chance to withdraw is really nice. Does anyone know if it stacks on units that already have a chance to withdraw? Would a chariot warlord or a trebuchet warlord have a really high chance of withdrawing?

Yes. Once I got a warlord immortal to have a 90% withdrawal. Loved it.

Also, my luck seems to have gotten much worse. I think I'm losing many more 90+ battles. Of course, now it a actually is what I should lose (1 in 10 90% battles I should lose0.

 

[Mango]

Yes, the computer cheats. No matter that it's more complicated to implement a cheating Random Number Generator, no matter that Firaxis is purposedly making customers unhappy by cheating on RNG, they've decided to simply screw everybody and make them loose all the fights they expect to win.



No, your interpretation is right. 5 out of 100 does not mean that you WILL loose 5 fights for every 100 you fight. It means that, ON AVERAGE, given a sufficiently large number of fights, you can expect to loose 5 out of a 100.
10 is way too small a sample to validate/infirm a probability. If you want to make a study that will actually mean something, your sample size needs to be at least 800. Meaning, fight 800 fights with 99% combats odds, and you should have a fairly good (but NEVER perfect) idea of whether or not the odds are right.

You don't need that large of a sample size to confirm that something is screwy. You simply need to know how to calculate the odds of an event happening.

If someone really did lose 3 out of 10 fights where their odds were >99%, then we can calculate the odds of this happening quite easily. It's simply (0.99^7) * (0.01^3) or .00000093. That's a .000093% chance of this event occuring, or about one-in-a-million (literally). And that's only assuming a 99% win percentage. If it were higher that number would jump significantly.

 

[ohcrapitsnico]

When I do look back on my combat experiences its seemed about the same amount won at low odds that did at high odds.


Regardless Firaxis should do some balancing with their combat system, for instance stronger/technically advanced units should receive bonuses against weaker/outdated units. In my opinion I wouldn't have the combat system based merely on a game of chance.

 

[D'Artagnan59]

Today as the Vikings, I attacked Karakorum with 2 berserkers. Both had a 90% chance of victory. AND BOTH LOST! And I've lost Alexander on a 99.6%.

 

[Mango]

I'm wondering if first strike chances are calculated in combat odds. After all they're "chances" so it would make sense for the odds to not be correct if those 2 first strike chances turn into 2 first strikes. I don't know anything about coding though.

 

[Eternalsteelfan]

Always invest in Tactics.

 

[sydhe]

If someone really did lose 3 out of 10 fights where their odds were >99%, then we can calculate the odds of this happening quite easily. It's simply (0.99^7) * (0.01^3) or .00000093. That's a .000093% chance of this event occuring, or about one-in-a-million (literally). And that's only assuming a 99% win percentage. If it were higher that number would jump significantly.

Actually, that's the percentage chance of winning the first seven battles in a row, then losing the last three in a row. The chance of losing some three battles out of ten, rather than three particular, is nowhere near that. After all, the chance you will lose the first three battles in a row is .0001%

 

[DaviddesJ]

But it was just a purely random occurance, with a bunch of results going "the wrong way", and if I'd won 25 out of 30 50% combats, I'm sure I'd rationalise it with something like "My brave, tough axemen etc."

If that really happened, I don't think it was purely random. The odds are about 1 in 10,000. More likely that there's a bug than that you just happened to get the 1 in 10,000 outcome.

 

[DaviddesJ]

If someone really did lose 3 out of 10 fights where their odds were >99%, then we can calculate the odds of this happening quite easily. It's simply (0.99^7) * (0.01^3) or .00000093.

Nope. It's 120*0.00000093 = 0.00011, or about 1 in 9000. Still too small to be likely.

 

[sydhe]

The chances you'll lose 3 out of 10 is (.99^7)*(.01^3)*120=.000112. The 120 is the number of ways of losing 3 out of 10. (This doesn't include the possibility you'll lose more than 3.)

 

[sydhe]

As you said.

 

[DrewBledsoe]

Nope. It's 120*0.00000093 = 0.00011, or about 1 in 9000. Still too small to be likely.

Isn't that a bit like say since the odds of winning various countries' national lotteries are often around 14,000,000 to 1, that it is impossible for anyone to ever win?

And strangely I can't ever remember losing a single combat with even 98% odds in my favour, and I must have fought literally thousands upon thousands of combats (and never stack attack , only for testing purposes), which is also statistically highly unlikely.

Oh well, I'll probably lose half a dozen in my next game, but thats even more unlikely