Searching for Future Prices, 12/14/05



A market exists to bring buyers and sellers together. Together buyers and sellers agree on a price to exchange goods. Anthropomorphizing the above we have a market searching for an equilibrium price.

Is it possible to predict the algorithm a market is using to search for equilibrium?

If it is possible to predict, it should be possible to front run the algorithm.

An overview of search algorithms is here:
http://www.dcc.uchile.cl/~rbaeza/handbook/search_a.html
My guess is that a market uses two algorithms to search for equilibrium: one when the market is under accumulation and another when the market is under distribution. Perhaps a binary search for accumulation and a sequential search for distribution.

The code looks like this:
if (accumulation) then
if (c > yesterday's close) then
sell with target yesterdays close
if (c < yesterday's close) then
buy with target yesterdays close
if (distribution) then
if (c > yesterdays close) then
buy with target (c[t] + c[t-1])/dx
if (c < yesterdays close) then
sell with target (c[t-1] - c[t])/dx
Where dx is a time-based fraction representing the amount of time left in the day.

The definitions of accumulation and distribution are the most difficult part of the model. For day traders, the first hour high/low makes a good starting point.


With thanks to Dr. McDonnell for the comment that got me thinking about this.

Henry Carstens
Vertical Solutions
carstens@verticalsolutions.com