Journal on Policy and Complex Systems where S q
( N ) the same quantity as is defined in Lo ’ s ( 1991 ) test . Still , q should be defined .
In order to establish the truncation lag q for both tests , the present authors follow the approach proposed by Lima and Xiao ( 2004 ), instead of simply using the formula developed by Andrews ( 1991 ), which is the more traditional approach .
The formula developed by Andrews ( 1991 ) is given by : the sample size . The main reason for that is can be seen in Xiao ( 2002 ), where the author demonstrated that Andrews ’ ( 1991 ) formula not only captures the short-range dependence , but can also capture the long-range dependence . Hence , they propose the following formula ( Lima & Xiao , 2004 ):
where is an estimate of the first-order autoregressive coefficient .
Lima and Xiao ( 2004 ) proposed a different approach , by combining information available in the dataset , as in Andrews ( 1991 ) and a correction of bandwidth lag selection based only in where q * is the optimal choice .
Consequently , using these four tests and evaluating the Autocorrelation function , it is expected that it is possible to detect the presence of longrange dependency in a robust way .
A Multi-Agent Prediction Market based on Boolean Networks
The main idea of this first model
is to simulate the dynamics behind the interactions between individuals , an external source of information and a market clearer , which aggregates the individual beliefs , in order to compose public opinion ( or probability of outcome ) of events .
The basic structure of this model is composed of individual agents that update their respective beliefs according to a Boolean based rule , where they assume a binary belief state : 1 when they believe a specific event will happen ; 0 when they believe that a specific event will not happen . The factors that
18