Table 8

PAIRED-COMPARISON T-TEST RESULTS OF EURODOLLAR PUT

FUTURES OPTION APPROXIMATION METHODS

Panel A: Training Data Set

   

PGANN vs. Market Price

PGANN vs. PCS

Sample Group

N

t-statistic

p-value

t-statistic

p-value

Complete

Sample

2000 0.000 1.000 28.854 0.000

M* > .01

1202 -0.948 0.343 18.801 0.000

.01 < M < 1

577 -3.473 0.001 18.795 0.000

M > 1

619 2.530 0.012 9.677 0.000

M < -.01

790 0.828 0.408 22.880 0.000

-1 < M < -.01

517 1.122 0.262 17.828 0.000

M < -1

271 -0.258 0.796 15.277 0.000

-.01 < M < .01

8 1.215 0.264 2.307 0.027

Panel B: Holdout Data Set, HOLDOUT1

Complete

Sample

6887 -0.796 0.426 52.931 0.000

M* > .01

4159 -1.704 0.088 34.451 0.000

.01 < M < 1

1840 -5.953 0.000 33.707 0.000

M > 1

2312 4.535 0.000 18.699 0.000

M < -.01

2698 0.579 0.563 42.215 0.000

-1 < M < -.01

1782 0.973 0.331 32.977 0.000

M < -1

910 -0.593 0.553 28.613 0.000

-.01 < M < .01

30 0.490 0.628 4.231 0.000
* M is a measure of the degree of moneyness, PGANN and PCS are the Genetic Adaptive Neural Network and Chen & Scott (1993) approximations, respectively. When comparing PGANN to the market price, the p-value is for a 2-tailed test. When comparing PGANN to PCS, the p-value is for a 1-tailed test.

[Back to Document]