Combat Simulator

[Lanidrak]

Okay, so I've been lurking for a while and posting occasionally. Just want you all to know I'm still a part of this community, coming up with ideas to try and make the game better .

So, my suggestion is simple:

A browser based simulator for calculating the win percentage of a certain deck, against a certain AI category.

• You choose the cards in your deck
• You then choose the opponent category (eg. False Gods/AI3)
• You then choose the number of simulations, default 100
• You then hit 'Simulate'

The reason I'm suggesting this is because right now, the only way to test a deck properly is to head to the trainer and build it, then do countless battles against the AI. This not only takes a lot of time, if you make any changes to your deck - however slight - you have to start recording a new win/loss record.

This will be extremely difficult to code, considering that no amount of clever AI scripts and procedures can compare to the complexity of an actual human player. Especially when a Rainbow-Deck is involved.

Somethings to consider...

• Effective use of Sundials, play only when incoming damage is over 20, and only if you have Light Quanta for example...
• Stealing the appropriate cards, and never steal a weapon when Eternity is in play for yourself
• Waiting to play a creature until you can Quintessence it in the same turn
• Stalling with correct use of Phase Shields and Sundials
• Waiting to know when to use a Firestorm or two
• Using Fallen Druid's improved Mutate on your fodder (Skeletons/Fireflies)
• Avoiding deliberate creature control when your opponent has a Bone Wall
• Etc...

As you can see, the list is endless. But I think this will be an extremely valuable tool for evaluating the effectiveness of a deck, although it will never be perfect. I think, no matter the level of intuitive coding given to the 'Human Players Deck' will ever make it close to as though an actual human player was in control. It would even be safe to say that any win-rate given by such a simulator will be 10 to 20% off of the actual value.

Anyways, just an idea , your thoughts and discussion will be most welcome.

[bobcamel]

I also guess that many decks wouldn't be properly controlled by AI, especially those with Immolation for example. Still, having the AI do 100 fights with your deck against AI3 decks would be much less tedious than doing it yourself, I agree...

...and you come a step closer to realizing grinding is boring.

[akromat]

Are you suggesting that someone else does this or are you planning on doing it?

[918273645]

Well, I think it would be most helpful, but I'm not sure Elements is built for this kind of stuff. Zanzarino would know more than me though, me being such a noob.

[Lanidrak]

I wouldn't be able to do it myself... Don't have the technical 'know-how' for some of these sorts of things.

It would very much be up to Zanzarino and Chris... First, they'd need to seriously up-the-game of an AI level which can come 'as close as possible' to a human player. Then, just give us the simulator. It's all already implemented in the game - it would just be a bare-bones version of the game, ie. no graphics, no quests, no spinner etc... Just, your 30 to 60 cards, an opponent and however many times you want the battle to take place.

Then it shows you the win/loss ratio against a given opponent with the pre-selected deck you gave it.

[Warkthogus]

Lanidrak, it's interesting to see that someone else had this exact idea.  We were talking about it in chat a few days ago, and I'm going to get started with it.  Now remember, I'm not going to spend an obscene amount of time on it, mostly because I don't really enjoy programming AI, but I'll get the ball rolling on my end.  Anyone else want to start their own effort/help?

[teffy]

I think combat simulators don´t fit to Elements.
I wouldn´t trust the results. Computers can calculate much, but the question is, if the calculation is right
Combat Simulators can´t simulate your way of playing.

You can choose not to play a card. e.g to save a Deflag in hand for later and not to waste it on weaker permanents can decide games.
You have another priority of destroying/stealing enemy permanents than the simulator.
You make better choices, when you have to discard a card. That may be especially interesting in Rainbow Decks.

I think games against user - made decks could be possible, thats enough for simulating.

[jakubas]

Just do 100 matches in trainer

[Warkthogus]

Just do 100 matches in trainer

I don't think you understand the logic and reasoning at work here...

The current problem isn't that it's impossible to test a deck's effectiveness, but that it takes much too long to do so.