I've used a number of methods in the past to do that, but all of them, no matter how elaborate, are highly inaccurate. The reason for this is that, at any given point, you have to compare the game you have in hand with something that's out in the market right now.
Here's where you'll err:
- Do you REALLY know how much they are making? (Are you sure their numbers are accurate? Were they trying to prove a point?) - I tend to aim 10-20% lower
- They are out RIGHT NOW, and you are NOT. Not only did they pick up on the wave before you did (which means your game will be deja vu to a lot of potential players) but they also came out in a different time period, which inherently means the market was in a different position. If there's a big tower defense frenzy, you can bet there will be more in the couple of following weeks/months, which means, if you plan to release then, you might get swarmed and come up at the botom, unnoticed. Consider the delay between the data you have on a given product and your estimated time to market as you do this. The faster you can react, the better you can earn.
- Possibly, your reference game is not "like" your game. This introduces a lot of risk: people might "get" it, or not. If they do, they might "like" it or not. All the more reasons to increase or decrease their sales
- Can you spend as much money on marketing as the base game? Can you get better/worse exposure? Can you hit the best platforms to reach your audiences? Are these different from your reference game? Is the art style more or less appealing to certain crowds?
- Are you a good judge of your own product's level of quality?
In brief, nothing you wouldn't already know if you have created a persona.