Corante

About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: derekb.lowe@gmail.com Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Emolecules
ChemSpider
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
PubChem
Not Voodoo
DailyMed
Druglib
Clinicaltrials.gov

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
Kilomentor
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
ChemBark
Realizations in Biostatistics
Chemjobber
Pharmalot
ChemSpider Blog
Pharmagossip
Med-Chemist
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
SimBioSys
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Business|Bytes|Genes|Molecules
Eye on FDA
Chemical Forums
Depth-First
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa


Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
FuturePundit
Aetiology
Gene Expression (I)
Gene Expression (II)
Sciencebase
Pharyngula
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net


Medical Blogs
DB's Medical Rants
Science-Based Medicine
GruntDoc
Respectful Insolence
Diabetes Mine


Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem


Politics / Current Events
Virginia Postrel
Instapundit
Belmont Club
Mickey Kaus


Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« Inspiration Is Where You Find It | Main | Does Glucophage Make Alzheimer's Worse? »

February 25, 2009

Single, Simple Numbers: Use At Your Own Risk

Email This Entry

Posted by Derek

I wanted to link to this excellent article by Felix Salmon over at Wired. He's talking about the mathematical formula that convinced many people on Wall Street that they'd figured out how to price out correlated risks in debt securities. As we all now know, they'd done no such thing, even though trillions of dollars ended up riding on the whole idea.

The article's well worth reading just on those terms. But it's also worth thinking about for what it says about other fields where the risks - and the correlations between different risks - can't be well measured. Such as drug discovery and development! Many examples in Salmon's article can be extended directly to our own industry: what are the risks of each compound in Company X's pipeline failing? If a compound with a similar mechanism wipes out over at Company Y, how have the odds now changed? What about patent risks - if a Supreme Court decision makes everyone rethink issues of infringement or obviousness, how correlated are the patent-busting exposure around the industry? And so on. . .

The difference is that we haven't (quite) convinced ourselves over here that we've got it all figured out, and we haven't issued billions of dollars in derivative securities on top of our individual drug development programs. Not yet, anyway. But if you come away from a study of the current situation with a mistrust of any formula that people try to use to quantify complex systems down to one easy-to-use number, well, you've come out ahead.

Comments (15) + TrackBacks (0) | Category: Business and Markets | Current Events | Drug Development


COMMENTS

1. processchemist on February 25, 2009 10:54 AM writes...

"we haven't issued billions of dollars in derivative securities on top of our individual drug development programs"

Surely not, as individuals. But there more than a consultant that sells methods to evaluate companies and pipelines from a financial point of view. Mathematically. An example?

Phase Success rate
Discovery 50%
In-vitro 67%
In-vivo 75%
Toxicology 80%
Total 40%

(taken from www.avance.ch)

Permalink to Comment

2. Biggest biotech on February 25, 2009 11:01 AM writes...

"we haven't issued billions of dollars in derivative securities on top of our individual drug development programs."

What do you mean? There may not billions of dollars of issued options on individual drug development programs, but there are cerainly tens, if not hundreds (in some cases), of millions of options trading dollars based on single drugs. Timely examples that come to mind are ARNA, DNDN, CYPB, ITMN, and SNTA, to name a few. These are all plays in individual drugs.

Good fun, really!

Permalink to Comment

3. RB Woodweird on February 25, 2009 12:13 PM writes...

Although not directly addressed by my good friend and drinking buddy Kurt Gödel, I think he set a reasonable example for not thinking that your precious little model contains all the answers.

Permalink to Comment

4. SteveM on February 25, 2009 1:04 PM writes...

I got into Operations Research (OR) after I got out of Chemistry. OR is applied mathematics and a lot of OR guys work(ed) as quant jocks on Wall Street.

I read the article attached to the link and what jumped out was obvious. Li, the Chinese quant guy imputed CDO risk from their trading prices. But the CDO's were initially mispriced, so the trading prices had to be mispriced! Why?

Because of simple GIGO. Most of the underlying subprime loans had little or no historic default rate data because there were exotic. In other words, there were too few data realizations for accurately generating a probability distribution for the default rate of people buying a $200,000 house with no money down and financed with a complex adjustable rate mortgage. How many would default? What would be time-phased default distribution?

Check your chicken entrails, because whoever priced those risks just yanked them out their backside. OBTW, any risk can be properly priced simply by acknowledging that the default distribution is a mile wide to the buyer. But who's going to by your CDO's with that kind of honesty?

About Pharma risk. Pharma is one of the few industries that have in-house Decision and Risk Analysis groups. (OR guys do that too.) Why the DA guys have failed and continue to fail to provide effective decision support is because they like to focus on decision factors they can quantitatively measure and explicitly monetize. Otherwise, they can't complete their decision trees. (Unfortunately, most OR guys have engineering mindsets.)

In Derek's post here, he mentions all kinds of qualitative or weakly quantitative risk factors that should enter into the decision formulation for drug development. And they are considered by Pharma management, but badly. There happen to be ways like the Analytic Hierarchy Process (AHP) for more accurately incorporating qualitative risk into a decision problem. But there are also organizational factors that preclude the acceptance of those techniques. (That's a whole other story.)

So Pharma is where it is. And it probably belongs there.

Permalink to Comment

5. tyrosine on February 25, 2009 3:05 PM writes...

"we haven't issued billions of dollars in derivative securities on top of our individual drug development programs"

I like this quote. I attended a recent meeting with a VC that invests in the biotech/pharma space. He mentioned that currently, with no IPO or acquisition possibility, venture funds have no way to cash out their investments. He also mentioned that the average life of a venture fund is shorter than the average time it takes to develop a drug, and thus there is a major problem for VCs. One proposed solution (albeit, not his favorite) was to issue such securities on products in the pipeline so that VCs could get out some value if they needed to hand back money to their investors.

Permalink to Comment

6. petros on February 25, 2009 4:06 PM writes...

Long Term Capital Managment's model was based on Nobel prize winning theories and still failed oiwng billions (chicken feed now of course)

Permalink to Comment

7. milkshake on February 25, 2009 5:52 PM writes...

From what I was reading LTCM - after being taken over and wound down by the government - actually ended up with a mild profit.

Permalink to Comment

8. Anonymous on February 25, 2009 5:59 PM writes...

Thanks for linking this, it was a very well presented article and made an obscure subject more than approachable.

Appreciated.

I'd be lax to 'correlate' the lessons to the pharma industry however (little joke). However, using such a mathematical model to find hidden correlations (pathway modeling?) could be interesting (way over my head though). I've often felt that not enough mathematics is focused on biological sciences.

Zz

Permalink to Comment

9. SteveM on February 25, 2009 8:59 PM writes...

Re: Post #1 processchemist

No offense. But I checked out the advance.ch web site. More junk mathematics. They claim "Real Options" as a area of expertise. RO was a consulting fraud out of the box when it was introduced over 10 year ago.

Real life product development violates every axiom of the financial options theory that underlies RO. RO died a miserable but deserving death after the dot bomb collapse.

I'm only bringing this up because if an outfit like advance is selling discredited RO, what does that say about the rest of their product line? And there are tons of advances out there selling voodoo mathematics to corporations apart from Wall Street.

The biggest mistake management teams often make is becoming enamored with eye candy charts generated with GIGO analysis. And they have the blood stained balance sheets to prove it.

Permalink to Comment

10. processchemist on February 26, 2009 2:19 AM writes...

@SteveM

no offense taken. I reported the numbers, but I'm more than dubious about any statistical or math modelling about assessing values of a preclinical pipeline.

Permalink to Comment

11. biggest biotech on February 26, 2009 9:18 AM writes...

Ya, the Avance stuff is complete BS.

In one of their downloads they give a fancy formula for WACC, then add some BS about "intuitive discount rate". Sad, that fancy equation with some greek symbols fools people. If you read the equation you realize, from a science point of view, its complete BS.

Permalink to Comment

12. Philip on February 26, 2009 12:10 PM writes...

Interesting! Today's quote of the day on my Google home page. I guess it's an old problem.

Today's scientists have substituted mathematics for experiments, and they wander off through equation after equation, and eventually build a structure which has no relation to reality.
- Nikola Tesla

Permalink to Comment

13. Bruce Hamilton on February 26, 2009 1:48 PM writes...

For those curious about earlier warnings about these shifting sand foundations of the Wall Street Palace, there's an interesting, readable, earlier article on Gaussian Copula and Credit Derivatives ( September 2005 ) by Professor Steve Hsu available at...

http://infoproc.blogspot.com/2005/09/gaussian-copula-and-credit-derivatives.html

Permalink to Comment

14. Hap on February 26, 2009 2:40 PM writes...

1) Single numbers figure well for those who don't have the time, the intelligence, or the willingness to figure out what's going on. They are likely to be misused and their caveats forgotten because the people likely to use them aren't likely to remember them or care.

2) The Wired article summarized it best as saying that the people involved in making investments based on the accuracy of the Gaussian copula "were making too much money to stop". If people don't want to listen, then having better or more incomplete indicators won't help. You can blame the people who were willing to tell the investment people what they wanted to hear, but if the investment managers weren't willing to take advice from people who didn't tell them what they wanted to hear, then the model of blame seems rather incomplete.

Permalink to Comment

15. Darkside on February 26, 2009 9:02 PM writes...

The "stimulus" bill is a giant derivative, based on dirtier mathematics than Wall Street ever used.

Does anyone get the impression that we've lost the ability to relate cause and effect?

Permalink to Comment

POST A COMMENT




Remember Me?



EMAIL THIS ENTRY TO A FRIEND

Email this entry to:

Your email address:

Message (optional):




RELATED ENTRIES
Weirdly, Tramadol Is Not a Natural Product After All
Thiola, Retrophin, Martin Shkrell, Reddit, and More
The Most Unconscionable Drug Price Hike I Have Yet Seen
Clinical Trial Fraud
Grinding Up Your Reactions
Peer Review, Up Close and Personal
Google's Calico Moves Into Reality
Reactive Groups: Still Not So Reactive