|
|
Post by Harry Lillis "Bing" Crosby Jr on Mar 17, 2019 9:18:54 GMT
Yep survivability is what i go for too instead of cost efficiency, having one 4 stack mech infantry rather than 4 of each singularly would be wiser imo. I like encircling my enemy, so I leave the max stacked for my generals you mean for daily invasions? iirc they only allow 5 stack/non stack unit groups, and its rather hard to apply double morale without rumour generals with that, for me 3 quad stack infantry(2 mech and 1 motorized) is enough to apply basic rumour while yamashita dishes out damage safely |
|
|
|
Post by Risviltsov on Mar 19, 2019 4:00:24 GMT
Alright, I'm bored with NHL EA games and I'll approach this again.
|
|
|
|
Post by Risviltsov on Mar 19, 2019 6:16:32 GMT
Very intersting topic ^^!. I just have one opinion, you only count atk and HP of unit but don't count the armor, the retaliation and the skills (between different units). As same unit with different stacked formations, the armor and the number of retaliations you receive are also important. (Example: 2 single units have more HP and atk points than double stacked unit, but two time damage decrease by enemies armor - because attack twice, two time received retaliation with lower armor => loss more HP than double stacked unit). I'll simulate a combat scenario roughly 100 times in Java with a simple program of two objects, using the simple battle dynamics of WC4. Results will come tomorrow morning, suggesting I don't have a project in AP Computer Science starting in ~8 hours. |
|
|
|
Post by Risviltsov on Mar 19, 2019 6:25:34 GMT
Will range be taken into account, along with the ability of tanks to attack multiple times? I'd have to convert a multi-dimensional figure (of stats) to a point number. That'd require a lot of simulations to determine the value of range and skills. Unlike attribute multipliers, abilities like those are highly indefinite and I'd have to find the mathematical "e" (natural) in the scenario. Range would require a recreation of the game mechanics and awareness of the map. I'm not willing to spend hours on that yet, no matter how intelligent or high-motor I may show myself to be. I'd need to create a Java program that'd quickly simulate most battle combinations and come up with an average value, where precision to a good point would be highly desired. As is the case, that in itself would demand a lot of C.P.U. time --- something I can't really waste in my current situation, as I'm stuck with a below-average HP 3125 running Ubuntu Bionic 18.04. If I were to tackle any of these factors, I must find the relative values of intangible abilities like the aforementioned one. Hold out for a month or two, and I may be able to come around on this. |
|
|
|
Post by Risviltsov on Mar 19, 2019 6:31:39 GMT
If I can afford it, I tend to stack my units as I also take into consideration their potential strengthening through rank promotion due to exp gained. Either way I'd love to see the math on it. I really think that ranking up makes stacking worthwhile, in conquests at least. I'll do the mathematics tomorrow afternoon, before producing one of three possible dinners (Fettuccine y Alfredo, Risolian Pizza, o Macaroni y Queso) and after school tomorrow is dismissed. I have the spreadsheets, I'll do it soon. From experience, I do think that veteran corps (XXX) units will be more financially efficient than recruit-level divisions (XX). Makes me wonder how effective a ranking system not unlike Tanki Online's ranking system and scaling would influence the efficiency and therefore demand for veteran units. |
|
|
|
Post by Dorado St. Sebastian on Mar 19, 2019 13:30:44 GMT
Yes, and also maybe substitute requeson for ricotta en lasagna 😊.
|
|
|
|
Post by Risviltsov on Mar 20, 2019 15:18:17 GMT
Simulator is finally up and running. I'll be doing roughly 100,000 trials, ignoring abilities (to gauge an ability-less average). I'm doing this to create a matrix to solve for the importance of attributes.
Light Infantry vs Assault Infantry - W 0.987%, L 99.013%
Light Infantry vs Motorized Infantry - W 0.0%, L 100.0%
Light Infantry vs Mechanized Infantry - W 0.0%, L 100.0%
Light Infantry vs Armored Car - W 0.0%, L 100.0%
Light Infantry vs Light Tank - W 0.0%, L 100.0%
Light Infantry vs Medium Tank - W 0.0%, L 100.0%
Light Infantry vs Heavy Tank - W 0.0%, L 100.0%
Light Infantry vs Field Artillery - W 0.0%, L 100.0%
Light Infantry vs Howitzer - W 0.0%, L 100.0%
Light Infantry vs Rocket Artillery - W 0.0%, L 100.0%
Light Infantry vs Submarine - W 0.0%, L 100.0%
Light Infantry vs Destroyer - W 0.0%, L 100.0%
Light Infantry vs Cruiser - W 0.0%, L 100.0%
Light Infantry vs Carrier - W 0.0%, L 100.0%
Total: 987 wins, 1,339,013 losses
^ As said, there is absolutely NO regards for abilities. Hence, the bizarre and impossible.
Calculates average win percentage, then adds this in a matrix:
(l - minimum attack, t - maximum attack, d - defense, h - health = win percentage)
16l+22t+1d+80h=0.000705
|
|
|
|
Post by Risviltsov on Mar 20, 2019 18:31:58 GMT
I have little time, but here goes some more simulations:
Assault Infantry vs Motorized Infantry: W 22.865%, L 77.135%
Armored Car vs Light Tank: W 0.106%, L 99.894%
Howitzer vs Rocket Artillery: W 98.536%, L 1.464%
Motorized Infantry vs Mechanized Infantry: W 0.0%, L 0.0%
|
|
|
|
Post by Harry Lillis "Bing" Crosby Jr on Mar 21, 2019 4:35:32 GMT
I have little time, but here goes some more simulations: Assault Infantry vs Motorized Infantry: W 22.865%, L 77.135% Armored Car vs Light Tank: W 0.106%, L 99.894% Howitzer vs Rocket Artillery: W 98.536%, L 1.464% Motorized Infantry vs Mechanized Infantry: W 0.0%, L 0.0% very useful study here, now we know which units we should be using on a certain type of enemy  |
|
|
|
Post by Risviltsov on Mar 21, 2019 16:25:48 GMT
I have little time, but here goes some more simulations: Assault Infantry vs Motorized Infantry: W 22.865%, L 77.135% Armored Car vs Light Tank: W 0.106%, L 99.894% Howitzer vs Rocket Artillery: W 98.536%, L 1.464% Motorized Infantry vs Mechanized Infantry: W 0.0%, L 0.0% very useful study here, now we know which units we should be using on a certain type of enemy I haven't even got to step 3 yet. |
|
|
|
Post by Harry Lillis "Bing" Crosby Jr on Mar 21, 2019 16:26:25 GMT
yep but its all good progress man |
|
|
|
Post by Risviltsov on Mar 21, 2019 17:46:59 GMT
Simulations of units against themselves:
Format: (Unit: Win percentage of instigator/first to attack, loss percentage of instigator/first to attack)
Light Infantry: W 97.609%, L 2.391%
Assault Infantry: W 99.877%, L 0.123%
Motorized Infantry: W 94.043%, L 5.957%
Mechanized Infantry: W 90.301%, L 9.699%
Armored Car: W 80.371%, L 19.629%
Light Tank: W 79.417%, L 20.583%
Medium Tank: W 84.911%, L 15.089%
Heavy Tank: W 80.518%, L 19.482%
Field Artillery: W 99.964%, L 0.036%
Howitzer: W 99.869%, L 0.131%
Rocket Artillery: W 99.451%, L 0.549%
Submarine: W 82.468%, L 17.532%
Destroyer: W 78.838%, L 21.161%
Cruiser: W 75.689%, L 24.311%
Carrier: W 70.074%, L 29.926%
Average: W 87.56%, L 12.44%
Note: Does not account for skills nor retaliation.
Note 2: This simulation only considers: lowDamage, highDamage, defense, health, halfHealth (attack dropoff), unitType (unused), vsInfantry (unused), vsArmor (unused), vsNavy (unused), vsStronghold (unused), isAlive
|
|
|
|
Post by Risviltsov on Mar 21, 2019 19:09:48 GMT
|
|
|
|
Post by Risviltsov on Mar 21, 2019 19:12:15 GMT
Note: Does not account for skills, range, technology Note: Only factors - Minimum Attack, Maximum Attack, Health, Defense. Range would require a 2-dimensional cluster**** that uses too much memory and time. |
|
|
|
Post by Risviltsov on Mar 21, 2019 19:16:59 GMT
Now the question is: How the hell will I connect this data to cost?
|
|
