Assuming you have done all your development, theoretically you should price your game to maximise your profit.
<profit> = (<units sold> * (<price per unit> - <cost to support sale>) )
Costs to support a sale would be an average per unit cost for things like tech support, community management, increased server load etc. Usually fairly negligible for simple games with many units sold, but for complex games with fewer units sold it might be a factor in choosing the price.
Of course, <units sold> is closely related to <price per unit>, because fewer consumers will buy your game if it's overpriced, and more will buy it if it's a bargain.
Theoretically there's a <price per unit> point which will maximize <profit>. The problem is that you don't know the details of the relationship. So how do you know where that point is? There are many factors at play:
- Many consumers are attached to the idea that a game's price should somehow relate to <average play time>. If a typical consumer spends 20 hours in total on your game, and it costs $20, then we can reason that they spent $1 per hour of entertainment. So that's why replay value comes in to the equation, it extends <average play time> sometimes by a lot. I've probably played hundreds of hours of Spelunky and it cost me only £11.99, so I guess it was a good purchase.
- Theoretically, consumers should also factor in <play quality>. That is <game value> = <average play time> * <play quality>. OK so you played a game for 20 hours but how much fun was each of those hours? If the game made you grind repetitively then not much fun. If each minute of the game was a unique enlightening experience then <play quality> is high.
- There are threshold effects, for example some consumers will simply buy straight away if they like the trailer and the game is less than say $2. Whereas once you go over $10 consumers will need more convincing.
- Sometimes a slightly higher price works in your favour if your product appears to be high-quality (e.g. snazzy trailer, awards, good reviews) -- the higher price will confirm this in the consumer's mind
- Certain types of game have a "maximum" price which has been established over time. For example The Witness attracted criticism for its $40 price tag because it's "only a puzzle game". Similar maximum prices may apply to games with the "indie" tag, 2D games etc.
So overall, I guess you should try to estimate the game value in terms of play time and play quality, then price it relative to games of similar game value but be aware of consumer expectations, costs to support sales and threshold effects.