Table 5

WILCOXON SIGNED-RANKS TEST RESULTS OF EURODOLLAR CALL

FUTURES OPTION APPROXIMATION METHODS

Panel A: Training Data Set

   

CGANN vs. Market Price

CGANN vs. CCS

Sample Group

N

Z-value

p-value

Z-value

p-value

Complete

Sample

2000 -0.591 0.555 -33.682 0.000

M* > .01

790 -0.151 0.880 -22.476 0.000

.01 < M < 1

517 -1.194 0.232 -18.494 0.000

M > 1

271 -1.467 0.142 -12.573 0.000

M < -.01

1202 -1.147 0.251 -24.828 0.000

-1 < M < -.01

577 -0.773 0.440 -19.790 0.000

M < -1

619 -0.947 0.343 -13.138 0.000

-.01 < M < .01

8 -0.280 0.779 -2.100 0.018

Panel B: Holdout Data Set, HOLDOUT1

Complete

Sample

6887 -3.305 0.001 -61.054 0.000

M* > .01

2698 -0.611 0.541 -41.830 0.000

.01 < M < 1

1782 -1.126 0.260 -34.768 0.000

M > 1

910 -2.517 0.012 -22.889 0.000

M < -.01

4159 -4.137 0.000 -43.272 0.000

-1 < M < -.01

1840 -2.157 0.031 -34.445 0.000

M < -1

2312 -3.666 0.000 -23.084 0.000

-.01 < M < .01

30 -0.051 0.959 -4.247 0.000
* M is a measure of the degree of moneyness, CGANN and CCS are the Genetic Adaptive Neural Network and Chen & Scott (1993) approximations, respectively. When comparing CGANN to the market price, the p-value is for a 2-tailed test. When comparing CGANN to CCS, the p-value is for a 1-tailed test.

[Back to Document]