Daily Speculations The Web Site of Victor Niederhoffer & Laurel Kenner Dedicated to the scientific method, free markets, deflating ballyhoo, creating value, and laughter;  a forum for us to use our meager abilities to make the world of specinvestments a better place.

Home

Write to us at: (address is not clickable)

Method in the Markets

Almost every field of human activity has benefited from the adoption of the scientific method. Chemistry, physics, biotechnology and a myriad of other activities have made spectacular advances by the basic tools of observation, recording, classification, formulating a falsifiable hypothesis, testing that hypothesis, publishing the results for review by peers, and going back to the drawing board. Popular investment literature is the exception to this great trend; it is rife with guru-worship, superstition and mythology. One of our main purposes in publishing this free Web site is to practice scientific inquiry and to promote its ideals. This goal drives and infuses every thought written here. We offer the following posts as directly applicable to the subject of science. 

05/25/2004
Law of Small Numbers

In studying the law of small numbers (the clustering illusion), the inordinate tendency to think that what's happening in small samples can be generalized to large samples, I came across the studies of the hot hand in basketball, the iniquitous effects of Medicare on small hospitals, the tendency of primes and Fibonacci series to violate all rules developed from the first 10 numbers, the defects and lack of power in Kahnemann's original paper on this subject, and one new word, strobogrammatic, a number that's the same when turned upside down (like 1691, 1881, 1961 and 6009).

In addition to the new word, the review of all the papers made me highly wary of inferences based on samples of 10 or 12 numbers, and led me to hope we will see very few posts on how the stock markets predict this or that about the elections this year, especially when such studies suffer from the confirmation bias, the tendency to present evidence consistent with what you already believe. -- Victor Niederhoffer

05/1/2004
Father Random and the Market Mistress

In our zeal to ascertain the tea leaves in such things as relative strength, buy stops during outside days, election seasonals, let us always remember Father Random. He creates cycles out of moving averages and x-day changes such as relative strength, he likes to gives a preponderance of winners to trades that wait until the first profit at an open versus a stop out at a big loss averaging 4% over the period covered, and he likes to find cutoff points relative to years, conventions, and parties, that yield correct predictions either way in 5 of the last 6 elections, especially if we ignore one or two.
-- Victor Niederhoffer

05/14/2004
The Alluring Studies

As the Specs return from studying deception in Vegas , a study crosses their desk by a very learned and able professor, with great facility in the world of practical forecasting, and corrective psychology. The study shows that there is a strong tendency to go up in the last hour using Dow Jones levels over a 20-year period. We immediately undertook a nine-year study to see if it would help to break the bank with the following results (with able assistance from Mr. Downing, the versatile former assistant to the great Sam Eisenstadt). 

Breaking up the day into various intervals, cumulating the move, and then dividing these further classified by retrospectively selected starting and stopping points is always going to show differences. The query remains as to whether these differences are consistent with those engendered by Father Random, the Market Mistress, the muse of deception, or Mr. Grind. 

A study from 1996 to the present of every interval of the day, using prices that one could actually trade, i.e., S&P futures during this period, reveals that the move from 3:20 to 4:00 averages 0 during this period. More useful and inconsistent with Father Random is the move from 11 to 12 which averages -0.1. 
-- Victor Niederhoffer

5/10/2004
Response to a Well-Meaning Spec

A well-meaning spec sends us data that the number of new lows is substantially greater than the number of new highs, and such has been the case at 3 or 4 market major declines. The post is a la Sornette, a report on one side of the distribution of subsequent moves, the rises, without taking into account of the subsequent declines, or expectations or uncertainty. Regrettably, such is all too typical of our field and it prompts the following response. ----

The post on indicators the new high to new low ratio, leads one to suggest a template for study of indicators. First its definition so that we may replicate it, ( presumably this is a continuous 12 month extreme indicator, rather than the old extreme year to date indicator), its qualitative raison d'etre, amount of implicit searching to find it ( i.e. other indicators discarded), descriptive statistics of the distribution of subsequent returns on both sides with its complete record of successes and failures, the uncertainty attached to the predicted distribution, the risk associated with its predictions and appropriate money management based thereupon, the execution costs, ease of acting upon same in real time (including whether its hypothetical vis-a-vis the slow moving Fisher effected index or actually tested on futures date (((see memo to consulting psychologist earlier today))), its out of sample success, its sensitivity to changing definitions and slight differences in when it implemented,( i.e. close or open), its harmony of result vis-a-vis the magnitude of the level of the indicator((((a standard test for velikofkian mumbo), and its incremental predictivity over and above its correspondence with extreme moves say in the last 30 days, and then generalizations of the effect with augmented sharper predictions ,methods of adjustment vis-a-vis ever changing cycles, and specific ways of validating it and refuting it for the future .

Such a template will enable us to differentiate ourselves from mumboists and promoters of all stripes and we thank Mr. Moody for this platform
-- Victor Niederhoffer

3/18/2004
Tom Ryan on Regression Biases
(a good reason to read Stephen Stigler's
"Regression Toward the Mean and the Study of Change"

02/25/2004
Terrible Problem

A terrible problem is that there is so much bad out there along with the good. Most people have no way of differentiating the two, so they throw out the baby with the bath water. That's why it's so hard to find good books. So many of them are so plausible, but they just guide you to making your more than your fair contribution to the market's costs. This is true of our worst-selling PracSpec book, still in the main unavailable at B&N, for example. Even my local B&N has not carried it, even though the other, EdSpec, sold many thousands of copies at that B&N and others. The reason is that B&N's buyers can't differentiate it from the hundreds of other books they have on candlesticks, Eliot waves, chart patterns, and the ways to trading success. But it could not be otherwise; if things were different, then people wouldn't make such big contributions to the market infrastructure, the energy cost of the system.

So much of what happens in the market is designed to prevent the indirect approach from teaching people the right way to trade. So much is involved with the symptoms of malaise -- for example, trading with many short-term randomnesses overlaid upon the core of value.

I am always amused when people say of the PracSpec book things like it is so obvious, everyone already knew that. It could have been boiled down to 40 pages. Yes, that's true. But then how would people know how to differentiate it from the other obiter dicta, and how would they be able to change. How would they know about the greatness of Artie?
-- Vic 

2/20/2004
Vic Speaks at Yale

2/3/2004
Beware of Retrofits

01/27/2004
Ignorance:

Frank Knight said, "It ain't ignorance that does the most damage in economics, it's knowin' so darned much that ain't so." This holds true in investments also.

Modern Fundamental Analysis:
As we showed in our book, Practical Speculation, both qualitatively and quantitatively, the Benjamin Graham method of fundamental analysis never worked and still doesn't. But that doesn't mean nothing can be learned from financial statements. We reviewed the latest academic work, interviewed the most au courant researchers and topped it all off with our own counting, to come up with better ways to pick stocks.

The Difficulties of Science:
Momentous advice given by Enrico Fermi to the young Freeman Dyson in 1953, recounted in an extraordinary essay in the Jan. 22, 2004, issue of Nature, holds a crucial lesson for all who would apply counting to the field of speculation.