How they do it

David Stone, Cowen & Co.

"We put together our models based on a timeline that assumes success," said Stone. "So, for example, unless there's reason to assume that a product will take more than one year at the FDA, I use one year. To be conservative, companies often use 18 months. But as an analyst you can be as much in danger of being too late as of being too early."

The amount of market research Stone does on a candidate compound varies. "Sometimes we rely on what the company tells us buttressed by what we know, by consultants and by hallway conversations with doctors. Sometimes we do primary research. But a lot of times, because of other uncertainties, the range of possibilities depends on how the data looks and overwhelms things like price sensitivity. Market research can give you a good answer about an existing product, but it's not as good for new markets. So we don't do as much of that as one might think."