🤑 Indie game store🙌 Free games😂 Fun games😨 Horror games
👷 Game development🎨 Assets📚 Comics
🎉 Sales🎁 Bundles

So how are we supposed to strategize now?

A topic by MGBlitz81 created 192 days ago Views: 652 Replies: 17
Viewing posts 1 to 10

I am slowing coming to terms with Alpha 5. It's a little better than the first iteration, but there are things that I can't understand here still and would like to hear others opinions on the topic.

Most specifically, I'm talking about other classes...

So great, we have other classes. How are we supposed to perform against other AI classes? What if I have a great AI using the assault class that was in alpha 4 and I find that it doesn't work against machine gun class but it works quite well against shotguns both using the same AI? What is it that I can do that will successfully defend against both these types of bots if I don't know what the enemy is using? It seems pointless to create an AI that is specific towards one type of enemy AI and class now. But how do we know which to use if we can't see what the enemy is going to do before hand? I mean, it's obvious that some classes perform better than others. So are we supposed to rely on luck of the draw to win a match? Then you have to take into factor the shotgun rush that is more easily countered, but still exists.

I don't know, it doesn't feel very fair anymore. More like a random match vs random bots. I can use my best AIs but it doesn't matter if you are using machine gun bots that fire from long range only. Then again if I use an AI that is stronger against machine gun bots, shotgun bots come and take all the resources. If I try to meld the two strategies, I weaken my overall attack and any decent AI using focused fire and same bots can beat me. It's rock, paper, scissors all over again.

Perhaps it would be nice to only be able to use the bots that the game creator is using? It makes it much more fair in terms of AI creation. You join a game and you can see that the creator is using 3 shotgun bots and 2 machine gun bots. The game then forces you to use the same set up. That would be a real challenge. You, of course, would get the same opportunity to choose the bot set up when you get to create games as well. Seems like a much more fair game to me.

(Edited 1 time)

first right now assault isnt balanced dont use them. Second don't focus on ONLY using one type of bot some compliment each other nicely e.g. machine gun + sniper. This will massively narrow your chances of being hard countered by bot types. Also check my 5.2 math to figure out how your bot should react to different classes. There is a filter that let's you select by bot type and it's INCREDIBLY useful. No bot beats another at EVERY range. Calculate what range you need to fight and at what health/shield point you need to win the battle then kite the opponent until your bots have the advantage.


Edit: This will help would be nice if it was sticky lol

Eh, mgblitz in alpha 4.x had 1600+ points, so he knows. But your input is true as well.

hdeffo saied right

Use machine gun like a wall and sniper togheter i rewrote my AI on it. I said more use the same AI in 3 games map. Only on specific map: circle, split and mind use specific AI becouse they are not generic map: example in circle no distance from enemy.

In your AI u can test the bot in enemy AI and understand what u must do. If your enemy use only shot gun u can execute your AI tree for this enemy units.... I use on split team u must know what your enemy will do and move before

I have to say: I'm amazed at the level in the top 30 league 4. You guys really made it difficult to reach the top 3.

Also: split team is my nightmare.

split team is currently a rock paper scissors game. Shotguns beat Snipers, Machine guns beat shotguns, and snipers beat machine guns. I am currently using a probability analysis algorithm I wrote to determine which of these units to send out each match so that more often than not my bots are the counter to my opponent's

I'm really interested by those sentences of yours "probability analysis algorithm". I mean, I know what you mean in theory but i would really like to see what you use

(+1)

Google how to predict rock paper scissors games. There's There's few articles on how to write a rock paper scissors engine. I simply adapted one I wrote for that into an engine to predict taking sides. I said probability analysis algorithm just to sound fancy 😂 it's still technically probability calculations

Developer

Did you notice the new class conditions and filters?

(Edited 1 time)

What mean % weakshieldratio and other % ratio?

Developer

https://itch.io/t/47059/question-normal-vs-ratio-c...

Yes, but it still doesn't matter much if you don't have the proper bots in play vs the enemies bots.

I guess what I'm trying to say is that every combination of bots should have a fair chance against any combination of enemy bots assuming you both are using successful strategies in your AI.

Keep in mind I only just unlocked sniper. Still plan on playing with it a bit but it seems the only way to go is sniper/machine gun combo.

(Edited 2 times)

To beat a machine/sniper combo take out the sniper first since this is nearly all of the opponents long range damage output. Now position you bots just outside of medium range near each other. Some machine guns will continue to fire at you from long range, if this is the case then fight at long range 2v1 you win. If it tries to close have one bot go to long range and the other fire at medium then keep alternating as shields deplete. No matter of the machine gun focuses on current or closest target it will lose.



P.s. you're welcome by the way 😂 i'm the one who really started this sniper/machine combo and only a few players up until now have figured out how to counter me. Now i need new team

the trick... :-)

U need change your strategies fast see your enemy understand what he do good and what you do wrong: in a game a was beat from machine gun & sniper... i study the math... then rewrite ALL my AI with shotgun.

In future with new AI i will do the same



NOW strategies (for me) sniper & (1-2) machine gun (the wall)

Kill snipers first: test the enemy unit...

Find trick for firing on single unit: like use of two simetric AI... Two snipers firing on the same unit is better an have more chances of kill enemy unit.

Your machine gun must stop any rush attac to yours snipers at any price (life...) never got back your snipers.

I don't say all the trick but listen this: u must test your change of your AI vs the original AI. Becouse i see an AI with e simple change of a test (like ANY... I didn't remember well... from CLOSESEST to FURTEST) to lose vs original.AI....

Bug or features i don't know but change the game winner...

(+1)

I think I just need to sit down and rewrite AIs like ntk said. I haven't really done any adjusting since alpha 4 beside throwing a few nodes together for the sniper class.

I rewrote everything at every update:

  • First, assault was the only bot
  • Second, shotgun was overpowered and I only used it
  • Third, it was the snipers for one or two days I think (5.1)
  • Finally, bots are very well balanced except assault (5.2)

So now: when you start close to the enemy, use machine guns, and when he's far use a combination of machine guns and sniper. But this last one is difficult: too many snipers lose against shotguns. This morning I took advantage of the sniper trend, I went full shotgun for 4 or 5 games and won them all, but it was very risky.

I did for the assault. I was not convinced by the new bot, now I started again but, yes, I see assault not so useful at the moment. It should be coded to have its sweet spot at medium range I think.

who has password for admin?