Daily Speculations The Web Site of Victor Niederhoffer and Laurel Kenner

 

The Chairman
Victor Niederhoffer

Home

About Victor Niederhoffer

 
Write to us at:(address is not clickable)

7/9/2005
Update: Decision-Making Under Certainty

There is no field in the world where decision-making under uncertainty is so retrograde as in medicine. Many treatment options are only available for use after hundreds of millions of dollars of research lasting more than 15 years. Countless innovators stay out of the field because economic incentives are lacking. Numerous treatments -- e.g., cancer-preventing molecules such as statins -- can only be used sub rosa because it would cost billions of dollars and dozens of years to do a proper double-blind prospective study of side effects that would have any reasonable power, say 95% that differences of a few percentage points might occur.

Strangely, in our own field, there is something even worse. That's the strain that says because that prices do not conform to a normal distribution, catastrophes are much more likely to occur than a typical observer might believe, and that normal statistical methods, normal decision-making under uncertainty, normal speculating in providing volatility insurance, are all doomed to failure.

Two quacks I met 43 years ago at Harvard are prominent in this modality, and it has not been entirely unprofitable for me that the world of investment ideas is in their thrall and that they are revered with the same awe as the prophets of pessimism, the Sage, the Palindrome, the weekly financial columnist who has never been bullish, and members of their pessimistic, zombie-loving cult.

Because I am always trying to stay up on the latest in decision-making under uncertainty rather than advising all others to avoid same by going to a horror movie, buying Treasury bills, or retiring to a monastery, I make it a point to go to the best college bookstores in whatever city I visit. In Britain, my usual rounds include Foyles, Blackwell's and the Oxford and Cambridge bookstores. On a recent trip I bought a highly recommended book on the statistical synthesis of evidence, "Meta Analysis by the Confidence Profile Method,"  by David Eddy, Vic Hasselblad and Ross Shacter.

The book deals with problems where there are:

  1. multiple pieces of evidence( macro and micro predictors)
  2. different experimental designs (e.g., a one-year study and a one-month study with different outcomes (loss or gain)
  3. different measures of effect (a Sharpe or a percentage of variation caught, biases to internal validity (survival bias)
  4. biases to experimental validity (e.g., the period before 9/11 and after)
  5. indirect evidence (the Sage has a war chest of $40 billion through buying deadbeats and clever manipulation of their deferred losses
  6. mixed comparisons (one study looks at P/Es relative to the mid-market and another at stocks in the Dow
  7. gaps in experimental evidence (no study ever uses completely prospective data for seasonality). The authors describe a general method to solve that problem, and give examples of how to apply it in nine areas of medical decision-making. The method involves maximizing a likelihood function with all the parameters and experiments given proper weight.

Good self teaching techniques are included.

While there would not be any direct gain to immediate profitability for stock market investors by learning these techniques or studying this technique, the reader will be able to develop a frame of mind that is much better attuned to proper decision-making in this all too uncertain world.

More writings by Victor Niederhoffer