El Farol Bar

El Farol Problem is a minority game problem, people have use it to describe financial markets...following is a section about it...

 

 

El Farol is a bar near the Santa Fe Institute that used to feature Irish music on Thursday nights. Brian Arthur, who is now the Citigroup Professor of Economics at the Institute, was born and raised in Belfast and enjoyed going to El Farol to listen to his favorite music. But there was one slight problem. Occasionally, El Farol was packed with rambunctious drunks whom Arthur wished to avoid. The chore of having to decide, week after week, whether to go to the bar led him to formulate

a mathematical theory he named the "El Farol Problem." "It has all the characteristics of a complex adaptive system," said Arthur.

Suppose there are one hundred people in Santa Fe who like going to El Farol to listen to music, but none of them wants to go to the bar if it's overcrowded. Now also suppose that, each week, El Farol publishes its Thursday night attendance. Over the past ten weeks, the figures were: fifteen, eighteen, eighty-three, sixty-six, forty-five, seventy-six, sixty-seven, fifty-six, eighty-eight, and thirty-seven.

Music lovers can use these past data to estimate how many people will show up at the bar next week.

Some may figure that approximately the same number of people will attend this week as came last week (thirty-seven patrons). Others might take an average of the prior ten weeks (fifty-five patrons) or a shorter four-week period (sixty-two patrons).

Now let's suppose that each person who wants to go to El Farol will go if he or she estimates that fewer than sixty other people will go that evening. All of the one hundred people will decide independently, using whatever predictor has proven to be the most accurate over the past several weeks. Because each person has a different set of predictors, some people will turn up at El Farol on any given Thursday night, and others will stay home because their model predicts that more than

sixty people will be at the bar. The following day, El Farol publishes the new attendance figure, and the one hundred lovers of Irish music update their models and get ready for next week's prediction.

The El Farol process, explains Arthur, can be termed an "ecology" of predictors. What he means is that, at any point, there is a subset of predictors that are deemed "alive," which means that at least one of the predictors is being used and the other predictors are "dead." Over time, some predictors will come to "life'' and others will "die."

Is the El Farol problem nothing more than a theoretical proposition used to help understand the difficulties of predicting complex adaptive systems, or does it actually exist in the market today?