|
|
||
|
The Technion Prediction Tournament Organized by: Ido Erev, Eyal Ert, and Alvin E. Roth |
||
|
|
||
2. The problem selection algorithm.The 60 problems in each set are determined according to the following algorithm:[1] The probability p is drawn (with equal probability) from one of the following sets (.01-.09), (.1-.9), (.91-.99) Two random draws are generated for the risky option (Xmax, Xmin): Xmin is drawn (with equal probability) from (-10, 0); Xmax is drawn from (0, +10). H'= Round(Xmax, .1)[2] L'= Round(Xmin, .1)
The Expected Value of the risky option is determined and an error term is added to create the value of the safe option: m = round(H'*p+L'*(1-p), .1); sd = min(abs(m-L')/2,abs(m-H')/2,2); e=rannor(0)*sd; m=m+e;
Finally the dataset is balanced to include equal proportion of problems that include nonpositive payoffs (loss domain), nonnegative payoffs (gain domain) and both positive and negative payoffs (mixed domain). if problem <21 then con=-max+min; if 20 < problem < 41 then con=0; if problem > 40 then con=+max-min; L = L'+con; M = round(m+con,.1); H = H'+con;
[1] The problem selection algorithm presented here is an improved version of the problem selection mechanism which was presented in the call for papers. [2] The function Round(X, .1) implies rounding X to the nearest decimal. The function asb(x) returns the absolute value of X. the function rannor(0) returns a randomly selected value from a normal distribution with a mean of 0 and standard deviation of 1.
|
||