Sentiment data in biotech has an obvious framing: find elevated sentiment before a catalyst, and position for the reversal. That framing is mostly right, but it hides almost everything useful about the signal. The interesting question isn’t whether sentiment is high. It’s whether sentiment peaked two weeks ago or is still building on the day. Those two situations look identical in a snapshot and trade differently. The trajectory of sentiment in the weeks before a binary event does something positioning data can’t: it tells you when the buying happened, not just that it happened. This sounds obvious. In practice, it’s easy to miss because most sentiment aggregators give you a level. You have to construct the temporal structure yourself. The more surprising thing: high sentiment before a catalyst is often a cleaner signal in the negative-outcome case than the positive one. If data disappoints and sentiment had been elevated, the unwind is fast and mechanical. Buyers came in on feeling rather than thesis, and they sell the same way. Positive data with elevated sentiment is messier. Sometimes it rallies. Sometimes it sells off on volume that looks wrong, and you spend the next three days trying to work out who was on the other side. One cut that doesn’t get enough attention: sentiment across different source types diverges significantly around catalysts in biotech. Social, sell-side, and specialist communities don’t move together. When they converge, all pointing the same direction in the same window, that convergence is the signal, and it behaves differently from any single source being elevated on its own. The setup matters more than the outcome. Whether something is sell-the-news can be more predictable from the pre-catalyst sentiment structure than from the data itself. That’s uncomfortable to write down, but I think it’s true.