Modeling S&P 500 Futures Mispricing Using a Neural Network.
The Review of Futures Markets, Vol. 12, No. 2, 1993
Posted: 2 May 2001
In this article we present a model of S&P 500 futures mispricing that is able to capture all major stylized facts observed in actual S&P 500 mispricings behavior. The model itself is inspired by theoretical considerations as well as empirical observations. The model's parameters are estimated by combining stochastic simulation with a neural network. Only a small number of sample statistics is required as input. From the simulated mispricings and futures price series it is clear that the simulated data fit actual S&P 500 mispricings data very well.The approach may easily be extended to other asset classes where the available information on the variables to be modeled is for some reason limited and/or where the model to be estimated is too complex to do so directly.
JEL Classification: G13
Suggested Citation: Suggested Citation