Terrible losses...what did you do (again) Gree ??

[manbeast]

the code tells you nothing about how casualties are determined. that is server side. unless you know an engineer who works at gree then you have no idea how they work besides what you see in the game.

code readers aren't hackers. anyone can do it.

don't need a couple thousand data points. do 5 attacks on someone with low stats and get zero casualties. 5 attacks on someone with high stats and 20 units die. if you don't see the correlation then you are blind.

casualties aren't just some simple equation you can plug consumption rates into. the way mark described it (and the only real info we have) is that each unit rolls a dice to see if it will live or die. if the first unit dies then the next unit rolling the dice has a lower chance of dying. if two units die then there is a lesser chance of the 3rd unit dying.

this is why consumption rates don't matter for overall casualty. attacking a strong player, you are always going to lose that first unit. even if all the units you have are extremely low casualty, the odds of the first one dying are still so great.

[stuartb]

Please allow me to be more clear. Independent third parties (read: hackers) who have dived into the code, have verified that statement. I saw it on these forums, but I remember it being credible.

This, to me, a mathematically inclined person, means that the burden of proof lies with the naysayers. Take a couple thousand data points, complete with relative A/D ratios and casualty numbers. Then we will believe it. Aggregate data from several people would be best.

Otherwise, I'm going with the guy with blue letters and the code scroungers who have no reason to lie.

I am pretty mathematically inclined myself, and I would have to see the posts to be able to comment on CCM's descriptions. I don't believe that the "consumption rate" itself differs by who you attack. If it is 0.015 for a unit, then it is 0.015 no matter who you attack. However, the way that the formula treats the consumption rate can alter the number of units that "die" in an attack. For example, say that you take 2,000 identical units, each of which has a consumption rate of 0.01, into battle with a rival that has 500 of the exact same units.

If the # of casualties is purely a function of how many that you take into battle, you would expect to lose 20. If the # of casualties is a function of how many that actually battle with the rival, though, then you would expect only 5 casualties because of the 2000 you take into battle, only 500 are engaged. The other 1,500 units are sitting on the sidelines watching the fight.

There are lots of ways to program the mathematics of the engagement, each of which will produce different #'s of casualties. I have not gone back and read each and every message, but I have read the ones where CCM says "I will not discuss the particulars of the calculation" and that speaks volumes.

I have been keeping some stats on A/D numbers and on casualties. The higher the rival's D, the more casualties. That doesn't mean that the "consumption rate" differs, but it does seem to show that the number and/or type of units that you attack does come into play when applying the consumption rate.

I don't have any reason to lie either. I just know a lot of math, and have been doing mathematical research for a living for a couple of decades now. Casualty rates may be constants, but the number of casualties is partially a function of the stats of the rival who you attack.

[Selfproclaimed]

I don't care what you've read or how smart you are at math. Reality is stats play a role, not just the units. If I can attack someone with really low stats compared to mine and not lose a single unit for hundreds of battles. But when I attack someone with stats near mine, ill lose some units....I see it how it is. I don't mind losing units. But don't try and insult people. Ive been on this forum since march. And don't know who the hell your talking about when you say by blues in this forum. Your starting to sound like the engineers that claimed the boosted units were working. The engineers themselves would tell ccm that they were working. When they were proven and tested that they weren't. So no, I don't give a **** what the code says. Stats play a major role. End of story.

[Shinazueli]

I am pretty mathematically inclined myself, and I would have to see the posts to be able to comment on CCM's descriptions. I don't believe that the "consumption rate" itself differs by who you attack. If it is 0.015 for a unit, then it is 0.015 no matter who you attack. However, the way that the formula treats the consumption rate can alter the number of units that "die" in an attack. For example, say that you take 2,000 identical units, each of which has a consumption rate of 0.01, into battle with a rival that has 500 of the exact same units.

If the # of casualties is purely a function of how many that you take into battle, you would expect to lose 20. If the # of casualties is a function of how many that actually battle with the rival, though, then you would expect only 5 casualties because of the 2000 you take into battle, only 500 are engaged. The other 1,500 units are sitting on the sidelines watching the fight.

There are lots of ways to program the mathematics of the engagement, each of which will produce different #'s of casualties. I have not gone back and read each and every message, but I have read the ones where CCM says "I will not discuss the particulars of the calculation" and that speaks volumes.

I have been keeping some stats on A/D numbers and on casualties. The higher the rival's D, the more casualties. That doesn't mean that the "consumption rate" differs, but it does seem to show that the number and/or type of units that you attack does come into play when applying the consumption rate.

I don't have any reason to lie either. I just know a lot of math, and have been doing mathematical research for a living for a couple of decades now. Casualty rates may be constants, but the number of casualties is partially a function of the stats of the rival who you attack.

I understand what you are saying. I respect that your data does not support my theory. I can't find the relevant thread.

Now, I'm very capable of doing statistical modeling, and I'm sure you could help me out with some of the aspects I'm missing.

I believe that with a few thousand data points, plotted with A/D ratio (attacker to defender) vs number of casualties (controlled, preferably, with the same type of unit, ideally, but that would be hard in practice).

I propose a new thread, with preformatted entries.

Ie: (your (calculated) unboosted attack) (their displayed defense) (#of unit lost) (name of unit)

I believe a good null hypothesis would be that A/D stats play no role in overall units lost, with the alternate hypothesis as A/D stats do play a role in casualty rates.

Either way, even if you don't support my theory, the model would then be able to predict, within a reasonable degree of certainty, at what cutoff ratio you begin to lose units. This would be helpful to many people.

I can start a new thread, or we could use this one. Please, help me be constructive. If that doesn't motivate you, you could also realize that if the model says the alternative hypothesis is correct, you'd have indisputable evidence that Gree lied to our faces.

Side note, please post no fewer than ten battle results with complete details, in a nice format, for data entry purposes.

Last note, I'd simply do this myself but I'm camping hard. Sorry.

[Mcdoc]

It's cool that you guys are looking at this from a statistical point of view - but if you check out my thread about the casualty glitch

( http://www.funzio.com/forum/showthre...ty-rate-glitch )

I have found a method of attack that let's me go weeks at a time without experiencing a single lost unit. The first few pages are on topic - unfortunately the last page has turned into an inquisition - but some really good info about losses

In fact - a few people have already documented their series of attacks that you can examine to support your theory or challenge mine

[Desas]

Mcdoc - when you don't loose units for hundreds of battles with your strategy (certain BO number) I was just wondering whether your army has some high casualty units?

Mine does have - and with recent changes I started loosing units even when getting only 30-50 bp (attacking less than 10k def players with my 53k att.

As for the mathematical discussion there is a good saying regarding theory and practise "theoretically I have a horse, but practically he is dead" - I totally disagree with Gree saying that whom you attack means nothing for your casualty rate - it's simple and clear on my side of statistics:
constant: my att=53k
defender1 (less than 10k) - losing 0-2 units (usually 0-1)
defender2 (less than 20k) - losing 0-2 units (usually 1-2)
defender3 (less than 30k) - losing 1-3 units (usually 2-3)
defender4 (less than 40k) - losing 2-4 units (usually 3-4)

[Agent Orange]

I had been using a similar formula Desas in terms of attacking. My attack score is just under 90K and I tested a few theories attacking players less than 10K defence. Up until the other day, when this thread started I could completely overwhelm those players with no losses.

But something has changed in that my loss rates have increased to the point that it is not worth doing PvP. As above that attack scenario would generate single or multiple losses of Super Hornets, Elite Ops or Transport Raiders! I always intel my rival before attacking to see what kind of mix they have of air defence units and even with very weak or low number of units my losses are still high.

So where does that leave me? I don't know, now the comment was something about high defence stats giving you the 'glitch' which might be possible in that my defence stats are very high but in terms of attacking in the past I took casualties so I would say I did not have such a glitch.

From looking at those who attacked me in the past who had the glitch, what I could see from those players is that they probably did not have high defence skill points because I could return the attack without loosing. Just take high levels of casualties so I would have suspected that those with the 'casualty' glitch had very high attack skill points and not high defence skill points.

If Gree targeted those with high defence skill points they probably went the wrong direction I think.

It obviously is some sort of calculation in game. Perhaps what is going wrong is all these units that are supposed to produce certain results other than not working caused the devs to make some changes to the way in which the game calculates losses and that is what has mess the whole thing up.

I've been playing for over a year and it seems to me things started to go bad with the Stealth Frigate change but then continued to get worse as these 'low casualty' loot units started turning up, and come to think of it the boost buildings as well.

[Poopenshire]

Currently lowered my allies and my boosted attack (assault bear not counted) is 96K.

I can attack and recieve casualties as follows:

Def raw stats

<15,000 no casualties
15,000 to 19,000 about 1 casualt in every 5 to 10 attacks
19,000 to 25,000 1 casualty per attack and 2 every 5 to 10 attacks
>25,000 forget about it!!!!!!!!!!!!!!!!!!

[Agent Orange]

One other observation, when these event units first rolled out the thing that I noticed was that the units didn't appear to do anything Hmm as reported but what I did notice that those of us who did not get the unit had our results change relative to the unit coming out. I realize that didn't make a lot of sense... basically I couldn't help but wonder if the unit didn't actually add anything to you if you got it basically kept you at status quo but for those of us who did not get the unit our stats changed in a negative way that would reflect the difference between having the unit and not.

Perhaps also why people getting certain units were always reporting that they did not appear to do anything.

[LUEZANA]

I am at over 30K A/D. I attacked someone as low as 700 A/D and still lost units every hit even though I won. I am attacking the same people I used to attack that had less than 10K A/D and i would lose nothing. now i lose 1-3 units every time. Something has definitely changed.

[Huckleberry658]

I am only level 64' ...... I have a goal of win 7 battles with alliance of 200 or more. I was able to work my attack defense to 5500 and 6500. I've spent 25 dollars since I started. All of a sudden I wasn't getting points for the last week. In a few hundreds track wins I've prob only gotten 4 or 5 rewards , I've have gone as far as 35 wins in a row without my bonus which was supposed to be every seven wins, and I'm making sure to attack above 200. What gives?