Stinson,

Can you please take a quick look and fwd it to Sam
Smith (wsmith2@enron.com)

Vince


---------------------- Forwarded by Vince J Kaminski/HOU/ECT on 05/07/2001 09:59 AM ---------------------------


VKaminski@aol.com on 05/06/2001 07:59:33 PM
To:	vkamins@enron.com
cc:	 
Subject:	Newsletter


A reminder: this column serves two objectives. The firstis introduction of  
new members of the Research Group, and of the associates andsummer interns  
rotating through the group. The second objective is to reviewinteresting  
publications related to energy markets, quantitative finance andother topics  
of interest to the Enron audience. I have run out of the newmembers to  
introduce and it's time to some real work. 

Today's column is devoted to the issue of technicalanalysis. Financial  
economists tend to dismiss technical analysis (chartism) asvoodoo science.  
Over the last few years, however, a significant amount ofevidence that  
technical analysis works at least in some markets. A paperpublished recently  
by Carol Osler of Federal Reserve Bank of New York (CurrencyOrders and  
Exchange-Rate Dynamics: Explaining the Success of Technical Analysis)reviews  
two cases in which technical analysis proved quite successful inoffering  
useful predictions in the foreign currency markets. The firstprediction is  
that trends tend to be reversed at identifiable support orresistance levels.  
The second is that if a support or resistance level isbreached, trends tend  
to accelerate. 

The explanation of this success can be found, according toCarol Osler, in the  
microstructure of the F/X market: take-profit and stop-lossorders tend to  
cluster around certain typical levels (like exchange rate endingin a specific  
market at 00 or 50 level). For example, the author found thatthat roughly  
2.8% of all stop-loss buy orders are triggered when the rate hitsa round  
number, while 10.5% of take profit sales orders are triggered. Thisasymmetry  
explains the success of the first prediction of technical analysis. Asecond  
asymmetry is related to clustering of orders just above and just belowround  
numbers ending in 50 and 00. Stop-loss buy orders are infrequent at ratesjust  
below these levels, and cluster above these levels. Stop-loss sell  
orders,however, cluster just below 00 and 50 levels a relatively scarce at  
rates justabove these levels.  

The question remains, why conditional orders tend tocluster at certain  
levels. One possibility is cognitive advantage tradersderive from managing  
large volume of information using simplified rules.Another possibility is  
that they follow recommendations of technical analysisin managing their  
books. In this case, technical analysis becomes aself-fulfilling prophecy. 

Please, contact us for thecopy of the paper if this caught your attention.
 - wjk0506.doc