Card games are inherently complicated to understand and balance due to their intrinsic randomness. It can be hard as a designer to really grok how a variety of games will play out due to the pervasiveness of that randomness. This article simulates a very basic card game so as to help you understand a standard spine for card games and provide a foundation for you to use when building one of your own.
We're going to take a Hearthstonelike basis for this article, where cards have a mana cost and the player only draws a single card per turn. We are going to define balance as the existence of at least two decks that have a win split of no worse than 4060. By having it such that each deck wins at least 40% against the other, it should be hard for players to accurately determine which one is favored and so they will feel free to play whichever they find more fun.
We're only going to have vanilla cards here and give them a value based on their mana cost and a base value for being a card. We should be able to generate an aggressive deck and a control deck that together qualify as balanced from this.
The early part of a card game like this tends to be mana constrained as players have lots of cards, but limited mana. The late game becomes card constrained as the players have spent most of their cards, but have lots of mana. So, an aggro deck is one that can play a lot of cards in the early game and a control deck is one that has spent more total mana by the late game.
From this, it's easy to see that increasing the base value of the cards will benefit aggro decks and increasing the value mana gives to a card benefits control decks.
Base Card Value 0Mana Value 1 
Let's start by taking a pair of decks to represent aggro and control respectively. These decks both have a slightly more midrange curve than what you would expect in most actual card games as it's a twodeck metagame, but they should be good enough for a demonstration.
For this simulation, we're going to set the player health to be 100 and change the ratio of mana value to health value. We're going to use the 3 mana 5/5 as our static card and change the values for the other cards based on the chosen ratio. Given these two decks, find a ratio for the base card value and mana value that result in a win split no worse than 4060.

Observations:
By changing the starting health, we very naturally expect to see a very different values for balance. For this simulation, I'm going to shift the starting health to 50 and leave the decks the same. Once again, you should try to find a ratio that results in a split of no worse than 6040.

Observations:
Another axis that we can play with is card draw. By having the players draw more cards, we greatly change the texture of the game and so need to change our values accordingly. For this simulation, both players are going to draw 2 cards every turn and we'll keep the starting health at 100. Once again, try to get a win split in the 4060 range.

Observations:
Here's all the variables that I've used above in one simulation for you to play with.
Card Value 5Mana Value 5Starting Health 50Cards Per Turn 1 
This is the smallest version of the article that I could make. Possibly future iterations could include:
If any of these seem interesting to you, please tell me and I'll be sure to put them in the next iteration of this article.
This is an early draft of this article, and I'd like to hear your thoughts on it. In particular:
< The Fear and Allure of Pure Narrative Gaps In Defense of LifeLike Games >