Hi survivors !
I've always wanted to understand what happens in a State Of Survival battle. Some are just curious, but it's like the grail for others to be able to find the magic formula which is hidden behind ๐.
For six months now I have been trying to unravel Kings' little secrets. But the subject is very complex. It is with the new fighting system that I have made decisive discoveries that have allowed me to understand and reproduce now a complete battle.
You can read more on : battle mechanics
The battle mechanics of the game is realy complex. You have no idea how many formulas and parameters used๐ต. Kings must have had as much trouble making and calibrating a realistic engine as I did finding their formulas. At first I used an Excel file but that was quickly insufficient. So I developed an application to simulate and understand the result of a battle with an accuracy of > 99%.
โ ๏ธThe simulator only works for states with the new battle engine (currently below 409). This guide is writen for those who have avery good understanding of the game and who are looking for a better understanding of the battle mechanics in order to optimise them. It will not answer simple questions (what ratio should I use). If this is not your case, you don't have to do anything here โ ๏ธ
The application can simulate several types of battle (infected, colony attack, trap ...).
It is necessary to do in thid order :
Choose the type of battle and the maximum number of turns.
Define the attacker
Defining or choosing a defender
Attack ๐ก
It is very important at the beginning to "calibrate" the application, i.e. make sure that your configuration is correct. Small changes can give very different results.
When everything is ready, you can save your configuration and use the optimisation function which will calculate the best configuration for your fight ๐คฉ
Even if the application is designed to run on the phone, it is best to use it on a computer or tablet.
The choice of battle will determine the benefits that do or do not apply.
โ๏ธThis type of battle has been successfully tested
Use Max rounds with : 9999
Attacker / Defender:
Unckeck Infected and configure
โ๏ธThis type of battle has been successfully tested
Use Max rounds with : 9999
Attacker / Defender:
Unckeck Infected and configure
The heroes benefits against the settlements are automatic. But, your opponent has settlement defense, these values are not linked to heroes, you need to input these informations (based on previous battle) on Benefits part.
โ๏ธThis type of battle has been successfully tested
Use Max rounds with : 9999
Defender:
Use load buton and choose Infected level.
The type of troops, the number of infected and the statistics will be configured.
โ๏ธThis type of battle has been successfully tested
Use Max rounds with these values :
Level 1 : 2 rounds
Level 2-3: 3 rounds
Level 4-6: 4 rounds
Level 7-30: 5 rounds
Defender:
Use load buton and choose trap level.
The type of troops, the number of infected and the statistics will be configured.
โ๏ธ Results are not verified
After discussing it with MUFASA, we think these two battle use the old battle engine.
โ๏ธ Infected Horde is not implented yet
The attacker/defender part is similar.
"Infected" must be activated to differentiate them from a player. Bonuses apply in this case.
The load and save buttons allow you to load your configuration or all the game templates (trap, infected, boss).
Statistics should be copied from your chef profile.
โ ๏ธ This information is often a mess. You have to check it carefully!
These statistics already include :
Your searches
Your talents
Your chef's equipment & badges
But do not include the statistics of the heroes with their equipment.
You can also use a combat report by checking this button. In this case it includes the statistics of the heroes, but not their skills.
Use this part for your troops.
In the case of a legion, put the sum of the troops of all participants.
โ ๏ธ The application does not check the maximum of your troops, it is up to you to check. You should not put all your troops but only those who participate in the battle.
Plasma are also here :
Infantry T6 => T6 classic
Infantry T6 1* => T6 with plasma 1
Infantry T6 2* => T6 with plasma 2
Infantry T6 3* => T6 with plasma 3
Add all your heroes. This will make it easy to test and find the best configuration.
โ ๏ธ Only the selected heroes will participate in the fight.
You must fill in all this information accurately. This is not so simple.
For the equipment, please use the table below for the 9 levels:
Add the participants' skills in the case of a legion (maximum 4).
If your benefits are not already taken into account in your statistics, you can add them here.
These are the bonuses that you can buy in the colony of 6%,12% or other bonuses of the game like the governor's bonuses.
The combat ratio is identical of the game.
The score provides a simple indicator to know at which level the fight is really won or lost :
-100%: The attacker has lost everything and the defender has no loss.
+100%: The attacker has no losses and the defender has lost everything.
+50%: You have lost half your troops and the defender has lost everything
...
The closer you get to 0%, the more the result may switch from one side to the other.
The statistics give information about what has been calculated. It is important to check that this part is identical to what is found in the game.
The graphs represent the evolution of the remaining troops in each round.
It is possible to analyse a battle with the battle report. The battle report already contains all the advantages (heroes ...). To avoid that the simulator adds these benefits again, you have to use the option below.
This function is very interesting to see what was problematic in a battle and to see the best configuration to use.
It is also possible to do it from an espionage report, but this one does not have all the statistics and requires a knowledge of its opponent.
You have to take the statistics part, the troops and select the heroes with their level (which is not displayed in the report, but here 80 corresponds to 8.1).
You have to manually set the benefit part for non-standard benefits to have the same as the report.
Manual version which seemed to me to be optimised
Version after optimisation of the ratios, less infantry and more cavalry and fighters
Just click on the "optimisation" button and the application will calculate the best (one of the best) solution for you.
The program starts by testing all the heroes with a troop ratio of 33% 33% 33% to find the three best heroes.
The programme then searches the best ratio. Impossible to test all possibilities (crazy computer time) : I use the Nelder-Mead algorithm in order to find a good configuration faster (between 75 and 100 fights are usually tested to get a good result).
After the calculation is finished, the battle is launched to display the result.
On the graph above you can see a big difference in score by changing the ratios (-90% to +90%) the programme will progressively look for the peak/maximum that gives the best result.
It is very complicated, even with a good knowledge of State of Survival, to have the right settings in the application. It is very important to have the good configuration, as the fight is by rounds, mistakes can make a big difference at the end.
Here is the technique to check if everything works correctly :
Choose a fight against an infected with 9999 turns maximum.
Configure your army with only generation 1, 2 or 3 heroes (Generation 4 is not yet very reliable)
Choose a level of infection at your level
Start the fight
Make sure that the deviation of the combat result is <1% of the total of your troops, for example here:
for the attacker 84422-82419 = 3 So 3/117,265 is 0.003%.
for the defendant 78812 - 78812 = 0 So 0.000%.
Combat statistics must be 100% accurate. If they are not, you have a problem with the stats, heroes or equipment.
And here stats are in the same order in the game ๐.
Currently the results are enough reliable to be used in the game.
GEN 1,2,3 : <1% โ๏ธ
GEN 4 : <5% ๐
Infected : <1% โ๏ธ
Infected Rally : <1% โ๏ธ
Settlement : <1% โ๏ธ
Settlement Rally : <1% โ๏ธ
Gathering : <1% โ๏ธ
Influencer Trap : <1% โ๏ธ
Capital / Fortified Bunker : <1% โ๏ธ
Reservoir Raid : Not tested โ
Showdown : <10% (I don't understand this gap) โ
Mix with big gap on tier (tier 1 with tier 10) : <3% ๐
Many rounds > 50 : <5% ๐
Errors accumulate if you have several GEN 4s, troop mixes and many turns (legions). Tests have shown for these cases deviations of up to 25% - 30% and certainly more. But the average in these complicated cases is 8% deviation.
Why these errors whereas the simulator uses the same formulas as the game?
There are rounding errors which are minimal.
Several Gen 4 skills (like Tyler's skill 2, Julie's skill 1, Ernie's skill 3) are not correctly implemented, because I didn't understand the calculation, they are approximations. These are the biggest deviation.
For the mix of troops of different tiers the game calculates a result for each tiers. I didn't find out how this calculation is done, the simulator uses a weighted power which adds a small deviation.
The simulator is a web application that works on mobile and computer. Working on a computer is easier.
It is translated into three languages, the language will be automatically chosen according to your configuration :
English
French
ะ ััั
The link to access to the simulator : https://battle.ultimate-guide.ovh/battle/
Use your discord account to login. This allows you to have your environment with your configurations.
On a mobile phone, you can add it to the desktop to have it as an application.
I stopped playing and updated the simulator at the end of 2020, it is no longer up to date is accessible now. The project is in opensource: https://github.com/request-laurent/sos.battle, players are looking to update it
Credit : request