It may work for a group of coders who are employees dedicated to a company, and are thrown projects, but it won't do shit for contract programming if you need to have waterfall level breakdowns that in turn feed to estimated work points.

You need to have project types. X screens with Y fields with Z level of complexity for implementation. Q network nodes combined with S servers times C clients which in turn will drive up your complexity, timing issues and add to your pad factor. Add in L layers of management overhead review times and associated cost for redoing nonrequired cosmetic changes that allow the VPs to put their stamp on the visual interface.

I can go on and on and on. Anyone who truly thinks they can estimate a project without having years of experience in the particular field and implementation type is full of shit. You will be winging it for quite a while.

Obviously this teams needs a strong manager to whip them into shape to perform up to the CEO obviously perfectly realistic assumptions.

Take the job. Make sure it is a serious raise. Then once you are in the position, use it to educate the CEO 1-1 and also go on sales calls. Do ANYTHING you can to reduce the CEO's burden, but never cut him out of the decision loop. Hell, go on sales calls all by yourself if that is what it takes.

5 years later you get the company when he retires.