No one is suggesting that program-survival bias renders investing in small biotech companies untenable. Rather, the idea is to try to make adjustments to one’s models that take into account program-survival bias and thereby arrive at interpretations of early- and mid-stage clinical data that are less bullishly biased than what you typically see on investment message boards.
"If a company ran few trials for one indication and kept going based on consistently good results in that narrow path, the bias is much, much smaller. "
Not necessarily. A post by Dew from a few years back found that "81% of phase III studies have lower response rates than preceding phase II studies" even though the P2 and P3 studies used identical chemotherapeutic regimens.
Generally in the real world there are almost none of the second category of companies/drugs - because by the time they have two successes they are done with trials (aka approved by the FDA). The closest you generally see is a company with a hit in ph ii on their pre-specified primary endpoint. Sounds good, right. But even most of these fail in the subsequent ph iii because the incidence of good drugs is sooo low and the false positive rate is so high.