Table 6

OLS REGRESSION RESULTS OF EURODOLLAR CALL FUTURES

OPTIONS PRICING ERRORS ON SELECTED PARAMETERS

Panel A: Training Data Set GANN Results, N = 2000

Equation

a

b M

b t

b u

R2

D = a + b M + e

-2.7E-08

(0.000)

-7.6E-.08

(0.000)

    0.000
D = a + b t + e

-1.4E-07

(0.000)

 

3.0E-07

(0.000)

  0.000
D = a + b u + e

-2.7E-07

(0.000)

   

8.4E-.07

(0.000)

0.000

Panel B: Training Data Set CS OPM Results, N = 2000

D = a + b M + e

0.088

(33.183)

0.029

(13.883)

    0.088
D = a + b t + e

0.059

(11.217)

 

0.038

(3.980)

  0.008
D = a + b u + e

-0.141

(-45.059)

   

0.683

(77.258)

0.749

Panel C: HOLDOUT1 GANN Results, N = 6887

D = a + b M + e

1.5E-.04

(0.517)

-2.9E-04

(-1.322)

    0.000
D = a + b t + e

1.2E-03

(2.111)

 

-1.8E-03

(-1.859)

  0.001
D = a + b u + e

5.0E-05

(0.078)

   

7.0E-04

(0.379)

0.000

Panel D: HOLDOUT1 CS OPM Results, N = 6887

D = a + b M + e

0.084

(61.752)

0.028

(27.298)

    0.098
D = a + b t + e

0.054

(19.996)

 

0.038

(7.723)

  0.009
D = a + b u + e

-0.132

(-77.174)

   

0.642

(131.264)

0.714
Note: D = CGANN - C (or neural network call price - market call price) for the panels that relate to GANN and D = CCS - C for the panels that apply to the CS OPM. M = degree of moneyness ((100 - K) - F(t)), t = time to maturity, and u = volatility measure.

The terms a and b are the regression coefficients and e is the error term.

All t-statistics are below their respective coefficients in parentheses.

[Back to Document]