GBG's Algorithm

[PianoFil]

Does anyone know what the whole algorithm is for GBG for type of goods and amounts of those goods for building the GBG buildings is?

I think part of it is number of players per age in a guild determines what goods and amounts are from tile to tile.. But I am curious what the rest is for determining goods and amounts. Unless I'm wrong about number of players per age in a guild.

 

[Mor-Rioghain]

Do you mean the price of constructing a province building? You're mixing up the two major components, I think. Those components are that A) the era of goods is based on your members' eras at opening of the season, not by how many members are in each era and B) the price for a province building is 'set,' regardless of what goods are needed to build it that round., e.g., a Fortified Field Barracks will always cost "X" goods. It's the goods that are the variable not the # required. (The distribution of the goods for a building may also vary. Example: If the building costs 7,500, the RNG might generate something like 112, 573, and 6815 but the total will always be 7,500. Later that same day the same building could be split into 3 very different numbers of each good but the overall total will still be 7,500).

The two components can sometimes collide for some distastrous results in a guild's treasury, especially if the primary HQ building uses up a good that starts cropping up throughout the season. Suddenly that nice pad doesn't sound so good anymore! lol (Believe it or not - and boy do I wish I'd have taken a screenshot of this! - about 6 months ago I build our Main building to the tune of 75K -- purified water! The Goods Gods were not on our side that day!)

 

[PianoFil]

So if I'm understanding you correctly, the era of goods that are mainly used in a GBG season are based on the eras that a guild has players in regardless of how many in each era at the start of the run. And the RNG of said goods for a given building always total up to 7500 regardless of individual numbers of each goods.

Did I understand that right?

 

[xivarmy]

So if I'm understanding you correctly, the era of goods that are mainly used in a GBG season are based on the eras that a guild has players in regardless of how many in each era at the start of the run. And the RNG of said goods for a given building always total up to 7500 regardless of individual numbers of each goods.

Did I understand that right?

I believe the frequency a good is chosen is weighted by how many people are in the era in that guild. But certainly no era where you have no people will be chosen.

But ultimately when the goods are chosen yes it's pure RNG as to how much of each that adds up to the total.

So you might only have 1 iron age player in your 80-person guild. But there is *some* chance (i believe ~3/80 because 3 goods will be chosen) that an Iron-Age good is chosen as one of the 3 goods for the 100k cost big building in the HQ. And as much chance as either of the other two chosen that you then need *mostly* that iron age good.

A new component is that lower leagues have cost-divisors now. So that it's more likely they can build more stuff. i.e. the same building in platinum league will cost half of diamond league. To as low as 10% of diamond-cost in copper. https://support.innogames.com/kb/ForgeOfEmpires/en_DK/4290 has a table with the cost-multiplier depending on the guild's league.

 

[WillyTwoShoes]

I believe the frequency a good is chosen is weighted by how many people are in the era in that guild.

I don't think that is correct. At least not officially (who knows what really happens).

My understanding is that the frequency or amount of a needed good for attrition builds has nothing to do with the number of players in a given age. Any age occupied by at least 1 player within your guild at GBG start is the only factor used to determine what age goods may be required on any given GBG.
-

 

[xivarmy]

I don't think that is correct. At least not officially (who knows what really happens).

My understanding is that the frequency or amount of a needed good for attrition builds has nothing to do with the number of players in a given age. Any age occupied by at least 1 player within your guild at GBG start is the only factor used to determine what age goods may be required on any given GBG.
-

Nothing official. I can start tracking it - it should not be hard to show if an era of goods is significantly more common than another - particularly if you have at least 1 of those "1 player" eras.

My understanding is just based on general feeling not evidence as such though. That my SASH treasury goods get asked for an awful lot less than my inactive players' FE goods.

 

[Mor-Rioghain]

So if I'm understanding you correctly, the era of goods that are mainly used in a GBG season are based on the eras that a guild has players in regardless of how many in each era at the start of the run. And the RNG of said goods for a given building always total up to 7500 regardless of individual numbers of each goods.

Did I understand that right?

You are correct.

@xivarmy : Nope. This has been gone over and over multiple times during different iterations/upgrades, etc. and the # of people in any given era is irrelevant to what or how much of a good is chosen. It's simply "random."

On a general note, I do remember that when the camps were first adjusted with this pricing scheme way back -- 3? 4? years ago -- that the announcement stated something to the effect that the 3 goods chosen would be in "low, medium, and high" ranges but despite the general belief that this referred to the era specifically, it actually has to do with the way the "price" is calculated, i.e., for a 5K bldg, 500 (sm) + 1500 (med) + 3000 (large) = 5K goods ... BUT ... I believe that they also said that the goods were from a low, medium, and high era. That is probably where the initial confusion began because an example of this in play for a 5K bldg would be 500 Purified Water + 1500 Bromine + 3000, so the Purified Water is the "lowest good" in terms of the 3 eras + it's the smallest quantity of the 3. Make sense?

 

[xivarmy]

@xivarmy : Nope. This has been gone over and over multiple times during different iterations/upgrades, etc. and the # of people in any given era is irrelevant to what or how much of a good is chosen. It's simply "random."

To how much of a good is chosen is simply random.

To how *often* a good is chosen certainly appears not to be. Logging just a HQ and one sector is already showing a relatively consistent distribution to *how often* an era of goods is chosen:

Era​	Players​	Total Frequency​	HQ​	E4B​
CE​	2​	12​	3​	9​
FE​	3​	17​	6​	11​
AF​	1​	4​	2​	2​
OF​	2​	14​	6​	8​
SAM​	1​	5​	1​	4​
SASH​	1​	1​	0​	1​
​	​	​	​	​
​	10​	​	18​	35​

Obviously not conclusive with only 53 selections so far. I'll add more later

 

[Mor-Rioghain]

The AI's choice of a specific era or even a specific good may well play into the algorithm for that season but overall the system was designed to address the growing concerns over wildly variant availability of goods on the market and the server, i.e., Global Trading monopolies, hoarding, etc. It stands to reason that the algorithm would be a general one with one or another "switches" set to address such deficits and surpluses. Can you imagine how complex it would be if it had to consider each era of each player each time a sector unlocked or even daily (at reset) rather than every other Thursday at GB opening? Talk about a nasty lag!!

 

[xivarmy]

The AI's choice of a specific era or even a specific good may well play into the algorithm for that season but overall the system was designed to address the growing concerns over wildly variant availability of goods on the market and the server, i.e., Global Trading monopolies, hoarding, etc. It stands to reason that the algorithm would be a general one with one or another "switches" set to address such deficits and surpluses. Can you imagine how complex it would be if it had to consider each era of each player each time a sector unlocked or even daily (at reset) rather than every other Thursday at GB opening? Talk about a nasty lag!!

Well good news, it doesn't consider which goods every time it unlocks It only considers it at the start, or possibly after a guild has kicked their last members of an era out (haven't tested it in years, it used to). Usually it keeps the same goods the rest of the season as was set at the start. This is the most common reason guilds run out of a good is that a high traffic sector winds up being rebuilt over and over all season with the same building draining the same good.

 

[xivarmy]

With a bit more data and confidence intervals:

Era​	Players​	Total Frequency​	Theory​	Wilson Min (95%)​	Avg​	Wilson Max (95%)​
CE​	2​	44​	20.0%​	17.6%​	23.3%​	30.0%​
FE​	3​	51​	28.1%​	20.9%​	27.0%​	34.0%​
AF​	1​	14​	10.6%​	4.3%​	7.4%​	12.2%​
OF​	2​	41​	20.0%​	16.2%​	21.7%​	28.3%​
SAM​	1​	20​	10.6%​	6.7%​	10.6%​	16.0%​
SASH​	1​	19​	10.6%​	6.3%​	10.1%​	15.4%​
​	​	​	​	​	​	​
​	10​	189​	​	​	​	​

It's extremely probable that the 2 and 3 player eras are more commonly asked for than the 1 player eras. (The Wilson Min probability for all 2+ player eras is higher than the Wilson Max for all 1 player eras)

It's consistent (though far from proven) that it's simply a direct weight of the number of players as to which good is chosen - the theory ranges are all inside the wilson interval. And most quite close to the average observed.

It's not even remotely proven that the times when only 2 goods show up that a good was chosen twice. This takes more away from the more probable goods in expected occurrence than the improbable goods. But may have no impact on the expected number of goods needed if the two occurrences are generated separately and then just combined to show the users. But this would be the simplest explanation in my eyes for why sometimes 2 goods or even 1 good are shown instead of 3. The other simple and logical alternative would be that 0s are being rolled, but 2-goods-required buildings seem too common for that to me.

I'm not likely to look to record more data unless someone presents a contrary finding - i'm satisfied that the frequency with which a good is chosen in GBG correlates to the number of players in the guild.