Quest For The Holy Grail

[DarkSider]

Value of troops sent shouldn't be calculated for the simple reason the other side doesn't know exactly what you sent and from their perspective they might face your entire lethal army.
So from defenders perspective they shouldn't be deprived of any benefits since you didn't send as much as they prepared for and also you shouldn't gain more for sending less since again they don't know what they'll get to fight and might even not defend a flak mob since your hack shows you have so much lethal potential.

To give a more convincing example think of it this way: you are attacking with a couple friends an alliance which pulls a massive last tick defence. You know it will be paintfull but decide to take them all in one go. BR shows entire defence was fake and since they had so little troop score defending and you had so much attacking that you where ready to take on them all, your gains will be dramatically reduced. Take this to extreme vs a full SO/tl alliance and you'll always take massive risks for minimum gains.

 

[BlackWolf]

Can someone explain why such system is needed in first place? I have not yet figured out why this game needs some fairness system when you can easily hard code things like land cap, which already should be enought to prevent attacking unfairly.

Another good example has been changing battles to be fought vs all, unless from same alliance or NAPs, or something along those lines.

I dont understand why some complicated system is needed to make things "fair" when fact is that things wont ever be fair.

 

[DarkSider]

It's not a system to make things fair it's a judge that says the land cap should be 10% or 13% or 15%. You said there is already a system in place that changes the land cap but the cap is affected by this Judge to say how big the cap should be given the circumstances. Same this judge should probably be in place for bounty to avoid farming for big benefits at the bottom of your range.

 

[CFalcon]

EDIT: I would make this easier to read, but I can't remember/don't know if there are any spolier/hidden text tags for the new forums

OK, I've had a bash at doing this, however it's not short, simple or pretty. To be honest I don't think a method exists that takes into account all the variables you want it to, and that is also simple. We're probably going to have to settle for complex and fair, or nothing.

Warning: if you don't like maths, look away.

My effort is based around the values of the attacking and defending mobs, the total troop values of the attacking and defending alliances, the value of other mobs attacking the defending alliance at the same time, and the number of defenders online. I also realise my attempt is very much centred on alliance play, but I can't think why it shouldn't work for solos.

I've taken the repeat attack modifier to be 1 when neutral, and increasing from 1 with more repeat attacks (assumed it can't reach double figures though).

I've basically come up with 5 'factors' which each take care of certain aspects of fairness. Each factor is worked down to a fraction which in the neutral/fair case (equal attacking/defending) should equal 1. I've then averaged these factors to come up with a final fraction which again is 1 when all fair, greater than 1 when you're attacking unfairly, and less than 1 when attacking fairly.
The important thing to note is that the averaging can be weighted to give a certain aspect of fairness more significance, so in the current version I've weighted the difference between the attacking and defending mobs' scores to be twice as significant as the factor which takes into account other incoming on the defending alliance and how many they have online.
This means it should be quite easy to tweak if people don't like the results given in certain situations. It also means other factors could easily be included if people think of things that should be included that I haven't covered.

So, in essence:

FC = 200 - 100m(10h + 6i + 5k + j)/22

please excuse the odd variables, I started with others, but how it came out was these.

Before you ask "wtf is that?", and if you have the stomach for it, please read what each variable means:

Spoiler

m takes care of what eta your mob was sent at, and is 5/(eta sent)

h is the difference between the attacking and defending mob values (including the targets troops).
h is where the repeat modifier is included
so if there are 3 attackers, A, B and C, with repeat attack mods a, b and c respectively, and defenders X, Y and Z then
h = (A*a + B*b + C*c)/(X + Y + Z)

i takes into account the difference between the alliance you are in and the alliance you are attacking, so if you're attacking someone about the same size, but you're in a large alliance and he's in a small one, then you don't really deserve as much eff/BH, whatever (at least I think so). To accommodate solo into this, I've just taken the 'alliance score' of a solo to be 20x the solos score, so if the solo is attacking an alliance that is on average larger than him, he will benefit. When two people from different alliances are attacking the same target, the mean alliance score is used, but a mean weighted towards whichever attacker is sending more.
so with attacking mob score A with alliance score a, solo attacking mob score B with 'alliance' score b, and the target is part of alliance c,
i = ((A/(A+B))*a + (B/(A+B))*b) / c


k takes into account how much incoming the target alliance is already having to deal with, and how many people they have online to deal with this incoming. In my book, if you send an overwhelming attack when the target alliance is already under stress, you should lose even more eff/BH etc. Conversely, if you are attacking an alliance that doesn't have any incoming, you shouldn't be penalised for sending more than is strictly neccesary to just attack the target. Everyone knows that if you're attacking a half decent alliance that isn't under stress, you don't just send enough to tackle the target, you send enough to handle possible defence as well.
If an alliance has a total of G incoming value, the alliance's troop value is F, and the number of players online is N, then
k = (20G)/(FN)

j is the 'psychological factor', which I was a bit iffy about, but which I've put in after seeing the argument put forward by DS, that if you think you're taking on a much larger force then you are, you still deserve something for having the balls to do it, or if you're taking on a much larger force than you realise, you don't really deserve the credit if it somehow works out. As you can see in the final equation I've only given it a small weighting.
j is based on highest score that can be made from players troops, given the size
of the mob. The calculator tries to make up the mob size by adding most expensive units,
up to number of that unit player actually has, then goes to the 2nd most expensive etc. This is more programming than maths, that I can't do without seeing exactly how.

j would then = total possible incoming score / total possible defending score

 

[Max]

Fantasic ideas CF, I'll have to write a few mathematical musings of my own when I get a chance

 

[Max]

On the topic of an eta calculation, I would suggest not to punish those who send at eta 4 or 5, but only eta 3 or less.

How about an eta fairness modifier for each attacker, such that those rushing at eta 1, 2 or 3 are deemed as more unfair.

To keep it simple, I would suggest that if one sends at an eta greater than or equal to 4, there is no modification (in other words, the 'fairness modifier' is kept equal to 1) and any etas below this (1, 2 and 3) have a modifier of 0.5. I believe that rushing at eta 1 with Jeeps is just as damaging as rushing at eta 3, in terms of damage potential, so that there should be a clear cut-off which is easy for the playerbase to navigate

More to come later I think...

 

[Max]

I also believe that a comparison calculation should be based on the value of the troops attacking, compared to the value of the troops belonging to the defender.

This way, when a player sends a single hippy to attack, the FC remains unchanged as it does not depend on the total value of the attacker.

A consequence of this would be that it is considered 'fair' that a player can send a small proportion of his army to attack a player at 40% of them, which I don't see as being a problem.

Perhaps a formula based on:

Y= total value of the target's troops + value of any defending troops
X= total value of the attacking troops (lethal or not)

So that the principle definition of fairness is is the ratio Y / X, like you mentioned earlier Azzer.

The higher this ratio is, the greater the fairness, as the more defending troops there are compared to the attacking ones. A ratio of one implies that the attacking and defending troop scores are equally matched.

The question is: should the other factors be added to the main one (as in CF's model) or used as a product. A product system would look like the following:

FC = a*b*c*d*...

where each of the unknowns are factors from 0-1 and can be scaled by a constant as necessary for normalisation. However, I think that CF's previous model would prove more successful, so long as the parameters are tweaked to perfection!

 

[Iamsmart]

I haven't read the entire thread, but I just read Max's post and I completely agree. I have been thinking that exact thought for over a month (whenever Azzer was chatting about it in #Bushtarion)

 

[Ridiculous_Lee]

Instead of going to the insanity complex level, why not just calculate based on value of troops at the battle each tick, and importance of that tick based on value of staff distracted/disabled/killed/bribed. Lets look at it a little closer.

Calculate the value of troops on each side before battle. Then look at how much actual fighting went on (the importance factor). This value will then determine for both sides how much their bounty hunting, and effectiveness are modified by that tick. Say the attacker is 3 times the defender, and the defender gets zeroed.

You end up with a value of (3/1)*25 (this value is 25/100 because 1/4 of the total troops died) for the defender =(75), and 1/3*25 for the attacker =(8.33).

These numbers are arbitrary right now, but could be converted into percent changes on landcap/effectiveness/bounty. Say anything below 50 results in losses from the normal, and anything above 50 results in gains from the normal.

Another important thing to look at is the battle over multiple ticks. If you follow this example, and combine the ratios for each tick, even though the attacker would be infinity times stronger than the defender at the close tick, the importance of the battle would be zero because nothing at all happened. This would allow the land cap code to work on the actual fighting that happened on attacking for 3. I'm going to go test out some values, and I'll get back to this suggestion.


I fail to see what aspect of fairness this doesn't cover. Blackwolf stated that you should not be penalized/rewarded based on what actually happens but what you think will happen, but I think that is hogwash. If you're expecting to fight a huge battle and lose tons of forces, and then you don't, you should still be land capped because you have an overwhelming force there. You sent way more than was necessary. Your reward is getting FREE land, even if it is reduced from 15% to 10%. That's 10% you didn't have to bleed for instead of 15% that you got slaughtered for. You can't have your cake and land for 15% every time too.

As for etas, that's the power of low eta units. I think a better solution would be to increase their etas only for attacking or whatnot. Rushing shouldn't be punished out of the virtue of it being rushing. It's a viable tactic, and I don't want to see it disappear.

Thanks for reading this whole thing, I was trying to be concise =/

 

[[Ninja-7]]

What about, for each attacker: (100 x (total defending score/total attacking staff score) x (your sent staff score/your total staff score) x (ETA/5))-((k) x (no. of repeat attacks))

And for each defender: (100 x (total possible attacking staff score/total defending score) x (your sent staff score/your total staff score) x (ETA/5))-((k) x (no. of repeat attacks))

Where:
k is a constant we’ll need to find through testing. Say an arbitrary 5
and ETA is the sent ETA

Attacking:
Total attacking score and total attacking staff score are those of any mobs that are launched, including yours, at the time you hit attack. Defending scores are current. BUT the FC on the 1st tick of battle is used for the final land calculations.

Defending:
Total defending score and staff score are current, those that exist on those ticks. The total possible attacking staff score could be worked out using: ((Score-(land, cash, intelligence, seeds and plants score)/(total no. of staff)) x (no. of staff sent)

And the no. of repeat attacks doesn’t include the 1st attack: e.g on 1st attack, would be 0, in 2nd, would be 1 etc.

E.g. a 1st time attack on someone equal to your size with all your troops nets you 100%

On double your size with all your troops nets 200%

Or half your size with half your troops nets 25%

A piggy or boost will net far less, because the original attacker’s score is included, but the original attacker stays unaffected.

Sending just 1 person on with a low init would not gain you massively, because of the total sent staff score/your total staff score clause.

You'll still get good land for a good battle, even if you manage to wipe out an opponent fairly and he has nothing left on the last tick.

This will encourage fighting around your rank, sending all your troops, and will stop new players being owned by an expertly crafted squad, sent by a much higher ranked player with better troops to choose from, with the same value as them.

Although if we disclude the sent staff/total staff clause, then we can disable this if lots of people don't like....
How does this sound?
What else should I have included?

sorry for all the brackets...