I was a bit puzzled at Jon's comment suggesting analysts numbers...

I was a bit puzzled at Jon's comment suggesting analysts numbers were not actually price targets but in fact risk adjusted valuations.

To me it defies logic that 7 different analysts could do their own independent risk adjusted valuations and, what with the considerable uncertainties involved, all end up within a few dollars of each other.

Are they all using the same model? How did they all end up with almost exactly the same values in their modelling? e.g. discount rate, drug pricing, time to peak sales, net margins, penetration rates, drug trial success likelihood, number of indications, etc.

If so, where do they get such accurate data from that everyone has almost exactly the same numbers?

For example, even Neuren states their patient population estimates have huge uncertainties. Phelan-McDermid prevalence estimated between 1/8,000 to 1/15,000, Angelman 1/12,000 to 1/24,000 and Prader-Willi 1/10,000 to 1/30,000. So the patient population number alone have uncertainties of over 100%.

To illustrate how sensitive a risk adjusted valuation is to even slight changes in assumptions I did some scenario analysis using the DCF model I've been using.

Adding a 10% increase or decrease in the model parameters, for example, varying the original 12% discount rate between 11% to 13%, the drug penetration rate between 18% and 22%, patient numbers (e.g. Pitt Hopkins) between 6,300 to 7,700, and varying drug pricing by +-10%, I get a +-36% change in the Net Present Value, meaning the difference between the lower (-10%) and upper (+10%) modelling numbers is a pretty large 72%.

So if your base case valuation was $30 a share, then just a +-10% variation in input values would produce a valuation output change between $19 to $42 a share.

Even a tiny +-5% change results in a +-18% variation in NPV i.e. $25 to $36.

So these analysts models must be within +-2% of each other in every value to end up with such a narrow range of similar valuations. How likely is that?

And again, the chances of this model spitting out a value that's within 30% of the current share price (the share price was $22-$24 a month ago when these analysts updates were issued) is pretty low. I'd suggest the only way this could happen is if you worked backwards from the current share price, and kept fiddling all the parameters in your risk adjusted model until you got the right number out the other end.

Hopefully Jon mis-spoke, as I'm hoping his comments the other day re comparing Neuren with Reata are closer to the mark on where he thinks Neuren should be valued.

All this said, I'd love to know how the professional analysts do calculate their published numbers (or 'risk adjusted valuations'). Does anyone know an analyst and can ask them?

I am sure they do indeed use some sort of valuation models. We've seen Hashan (and E&P) publish some quite detailed modeling. But it has me wondering if they simply use their models to work out if the company is under valued or over valued and then simply adjust the current share price up/down a bit to account for this. That would explain why the valuations always seem to be directly related to the current share price.

I note that some analysts have also talked about 'unrisked' valuations. But even these haven't made much sense to me. As if I do an unrisked valuation I end up with 10-20x the numbers they've published. Would love to know what risk factor I'm missing in my modeling that these guys put in to get such low numbers out the other end.