Biotech Values

[DewDiligence]

Re: Backgammon-style analysis:

>>Dew, regarding your backgammon style likelihood analysis, Now that the ISEL results are public, I'm curious as to what data you crunched and how you crunched it to come up with your likelihood analysis that you previously posted here. I'm not sure if this is proprietary or something that you want to keep private (in which case i'm sure you will think of an elegant and polite way of not answering), otherwise I'd be curious to know.<<

Hi Ant. I relied on the published data about the various Iressa and Tarceva clinical trials and statements made during conference calls by OSIP and other companies during the past few years.

What I think is as important as the data sources is the manner of thinking one uses to mentally process the data.

How many times have you read a stock recommendation like this:

“The drug works, it should be approved no later than ____, and this is a ___-sized market. With a ___ market share, sales will be____.” ?

Or a variation like this:

“There are ___ drug candidates competing for this market but the others have no chance to be successful because of ____, so that leaves the entire market for us, and sales will be ____.” ?

What’s missing in the above thinking is… probabilities. Few things in investing are black and white. Shades of gray rule!

As baby simple as this may seem, I have noticed that it is foreign to the way many investors (and non-investors) approach problems. When I ask someone to assign probabilities to event outcomes, I often get the response, “I can’t really put a number on it."

My contention is that you have to assign probabilities to future events to be a consistently successful investor, and the only question is whether you do it explicitly or implicitly. People who say they can’t or won’t assign probabilities to events are in reality doing it implicitly. I’ve found that the extra step of making this task explicit can impose a degree of discipline that’s quite helpful.

What does all this have to do with playing backgammon? Quiet a lot! The use of the doubling cube in backgammon (a relatively recent addition to this ancient game) means that you must assess your winning chances in a given position explicitly in order to decide whether the position is good enough to offer a double, or good enough to accept your opponent’s double. It is not sufficient merely to believe that your position is strong –you must make an effort to determine exactly how strong it is relative to the thresholds needed to offer or accept a double. This is done by looking at what can happen from the current position on the various sequences of future dice rolls, assigning a valuation to each of these sequences, and aggregating the individual-sequence valuations into an valuation of the parent (i.e. current) position.

(For the algorithmically inclined, this is a tree walk with a branching factor of 21, because there are 21 distinct dice combinations to be considered at each node in the tree. Experienced backgammon players conduct a shallow tree walk fairly quickly by recognizing nodes in the tree which are similar to a mental database of positions with a known valuation. Because of the combinatorial explosion of future sequences caused by the dice, backgammon players do not conduct the kind of deep tree walk that a chess master does.)

The above may seem esoteric, but I am absolutely convinced that having played backgammon for non-trivial amounts of money has reaped substantial dividends for me in the realm of investing.

Getting back to a concrete example, let’s take the iHub survey question about Celebrex. The survey asks whether Celebrex will be withdrawn from the market or not withdrawn. However, this is much too simpleminded. What you want to come up with is the probability that Celebrex will be withdrawn, which is not 100% or 0%, but something in between. Hence a more informative survey would have asked for the probability (rounded to, say, the nearest 10%) that Celebrex will be withdrawn. I didn’t construct the Celebrex survey this way because most people hate to give actual numbers and I didn’t want to do anything to discourage participation!

I think that's enough! I hope I answered your question and didn’t bore anybody. Regards, Dew