ROI for patent application

The ROI of a patent would have to be determined by the likelihood of the commercial success of the product/system being patented. For example, the cell phone, or when Apple came out with their iPad - apparently, the commercialization of tablets had been tried before, but were not successful, and yet when Apple came out with their iPad, they apparently had the know-how of how to successfully market their iPads, so their ROI was obviously huge.

My response is from the standpoint of a technical expert, expert witness, and one who has evaluated patent portfolios for potential sale, licensing, or litigation.

This is the million dollar question. The answer is: nobody knows at the time of filing. To some extent, one would need to be able to predict future trends. Having reviewed many patent portfolios, I can say based on my own experience that less than 1% of the patents contained in the portfolios I've reviewed are suitable for licensing, sales, or litigation with any reasonable degree of success. To some extent, the law of large numbers applies.

It's not all random, however. You can markedly increase the chances of success by ensuring that your patent is not easy to work around (so that others don't circumvent your invention to avoid paying royalties), that there is a articulable value to your patent that makes it superior to what is currently available, that the market for your invention is large enough or could reasonably become large enough (the more companies who potentially use it, the more valuable it is), and that it is relatively easy to detect the invention in products (so you know who to sue if you suspect infringement). Also, spend a lot of time ensuring that you are not reinventing something that already exists, because that can cause your patent to be invalid (i.e. worthless), even if USPTO grants your patent. In general, an invention can be technically brilliant, while at the same time being a very poor patent material. Or it could be great patent material which happened to be poorly written (I've seen many of those). I hope this is helpful.