Cyclacel Pharmaceuticals Inc (NASDAQ:CYCC) soared throughout the US session on Thursday, gaining as much as 60% across the day. The company dipped into today’s session, and has given around 10% back since the markets opened, but there’s still plenty of of volume underpinning action, and it looks like we might see a return to the upside before the week draws to a close.

The question is, what’s driving the move?

A quick look at Cyclacel’s latest releases points to the announcement that the company is on track to put out top line data from its SEAMLESS trial before the end of the year. SEAMLESS is one of the largest ever studies conducted in elderly patients with AML, targeting those who are unfit for or refused intensive chemotherapy, and an efficacy readout for the drug in question – sapacitabine – would be a massive upside catalyst.

So that’s why markets are buying in, to load up ahead of the data hitting?

Well, not so fast. The answer to this question is probably yes – investors are loading ahead of the readout in anticipation of a a binary event. This anticipation, however, is misplaced. We think the company is due a sharp downside correction when the top line readout does hit, and that the recent gains are the result of investors getting in on a misunderstanding of what’s going on.

To explain, we must first look at the drug in question.

As mentioned, it’s called sapacitabine, and it’s targeting AML in what amounts to a last resort treatment indication. It’s a pretty substantial unmet need, and some analysts argue that there’s a $500 million peak sales market waiting for a company that meets it (and in turn, an introductory $2 billion valuation, or thereabouts, for Cycalcel, if the company can garner approval for sapacitabine. That’s why there’s excitement surrounding this one, but there’s a key point that many are overlooking, which goes back to the end of 2016.

Mid December, 2014, Cyclacel announced that an independent data monitoring committee (DMC) had informed the company that the trial had hit its predefined futility boundary, which meant it’s unlikely to significantly improve survival compared to intravenous decitabine (the control in the study) in AML for these patients.

However, the DMC recommended that Cyclacel keep monitoring patients until the pre-specified 424 deaths for the final analysis had occurred. There’s a key word here, that to some will form the basis of a bull thesis – the DMC recommended, not required, Cyclacel to continue. This suggests (perhaps) that the DMC saw something in that data that hinted at long tail efficacy. That’s far to speculative for us, however, so we’re not basing our bias on it.

So, the interim analysis basically told markets the drug didn’t work. Why are the same investors now buying in?

Well, the above long tail efficacy thesis may hold water, and in turn, may underpin a buy trigger for some, but here’s the issue: the interim analysis was made up pretty much entirely of deaths that occurred during the first six months of treatment, and accounted for 70% of the 247 deaths examined. So, a very large proportion of these patients died within 6 months. When you apply this group to the final data, it’s going to leave very little room for error for the remaining patients to demonstrate a stat sig benefit over control, even if there is a long tail benefit. In other words, 70% of let’s say one half of the whole population, so around 35% of the population, didn’t survive long enough for this long tail to kick in. To claw back anything substantial from the remaining 65% is going to require incredible results, and that’s very unlikely.

With this in mind, there are only really two outcomes comes topline readout, in our opinion.

The first, the trial fails, and the company crashes. This is probably the most likely of outcomes, even with some degree of long tail benefit shown.

The second, the company somehow manages to demonstrate that there is long tail benefit, but it’s not enough to (from a stats perspective) to warrant approval, and it must conduct a fresh trial. More money, more time, and more room for error.

Both outcomes could spell real trouble for Cyclacel and its shareholders.